
From fenix@google.com  Mon Feb  3 09:21:08 2014
Return-Path: <fenix@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id B63061A017E for <hybi@ietfa.amsl.com>; Mon,  3 Feb 2014 09:21:08 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.313
X-Spam-Level: 
X-Spam-Status: No, score=-1.313 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, J_CHICKENPOX_65=0.6, RP_MATCHES_RCVD=-0.535, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id PO5uGFlKfUFz for <hybi@ietfa.amsl.com>; Mon,  3 Feb 2014 09:21:06 -0800 (PST)
Received: from mail-qa0-x236.google.com (mail-qa0-x236.google.com [IPv6:2607:f8b0:400d:c00::236]) by ietfa.amsl.com (Postfix) with ESMTP id E512A1A019A for <hybi@ietf.org>; Mon,  3 Feb 2014 09:21:05 -0800 (PST)
Received: by mail-qa0-f54.google.com with SMTP id i13so10327020qae.41 for <hybi@ietf.org>; Mon, 03 Feb 2014 09:21:05 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=WHCD5n+JgyRiUDmLkLooKH/EENpANJPz0JokoRcAuNk=; b=nEHXIcuaxiTHQ5KJoqBXad/317XS96GLp5nnUokvYA7Dan9mLsxWmA9wYZyQ9GL3RG hYld30xbdCA4FHW0OKD/HjiiHo/ylMhgeDAxpqD4ZUnoqkYnJafcgddr9CDJL3aJYGKH bWmqoDmPsBxyB8/LnszOo5Dboq+6y0GiTayWIgZGvQoSCG95lyzL8rbSteMUDbojcHOE zdFf1/9zuMYgO2+pnW/3XcT5XKIBM1mINSiXtvx6JFjk5O8dUVdctI3/MnjU8Pa9HM2p c5dq/9qMnZEkWgRKtx9Ogzn3zJy3bD3mWTGdlqT0UnUI1Mv/UgghiblTeE7YjXEch4zS 2wDw==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=WHCD5n+JgyRiUDmLkLooKH/EENpANJPz0JokoRcAuNk=; b=jR2iQX4b2HTJ+gk5EsgaAxNWf6uXiD1ime6Op42NS+AT/nKMxeiKIZpMjGb74egsvv WRjnOZMhuysANvE3NCf7oW4Y7f/1fQzbC27VJv4bhWYF7jg6fpDAS/eG+xg8lJLstlnu bnNF64K38c8Hm7IPhJhZUMxPdfNCm9E4mZUqTk7zL0g8wzuNEZCV7ibgNV0vPwMsEhIK nu7IeT+zDobvtXKeH21cXhRmekkx6Ryx0OoycLfLzw8SVbsuK7lMpEkqDmYygVnK0rrO XcOuOrJKOYtUb4rOBrMZTItaaWcdyfIKJQeeBVR9isQBQv1bjWwW00UQVxjjYecyN1Cf HjqA==
X-Gm-Message-State: ALoCoQk2IN9S5TD0VKyUqja6gLOTVkRTEiPiwnpBsLMwNqplah/Xo045UCg1aL/vIJh5HYHYOvbhrOJ8bD9a6JpWIkXrYJieS5BaFnRxxWNOQT///qNf6P1vRaH1GVX50rdMyRp8/qvuus56lWAwpES5r+jPVqRz7RlVU1ms9u+GCqLRhzii6ArmtION9yC7juD4qApE/cBq
MIME-Version: 1.0
X-Received: by 10.140.21.179 with SMTP id 48mr54355560qgl.78.1391447733518; Mon, 03 Feb 2014 09:15:33 -0800 (PST)
Received: by 10.229.13.7 with HTTP; Mon, 3 Feb 2014 09:15:33 -0800 (PST)
In-Reply-To: <634914A010D0B943A035D226786325D4446C1E9735@EXVMBX020-12.exch020.serverdata.net>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <CABihn6GSrU8oJa4nbMJiDDo2Z+-c4fT69roVFxcHYM4XYFToRQ@mail.gmail.com> <F1F115D4-29EB-457B-B02C-71E46D3E1C81@zaphoyd.com> <CABihn6GXxUwHwLd+926VPEkZxHRZUUi-3dw0PoNaO6aK8X4sEQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446C1E8D2C@EXVMBX020-12.exch020.serverdata.net> <CAGzyod7pt_PzjexmnZj8dSMg3BooUmuMb6xB5P0sG1t6S5FXZw@mail.gmail.com> <634914A010D0B943A035D226786325D4446C1E925F@EXVMBX020-12.exch020.serverdata.net> <CABihn6GrDwu=74WEACTrd8RMwDe78G-Q3X+e_OUrYbPcFm8gZQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446C1E9735@EXVMBX020-12.exch020.serverdata.net>
Date: Mon, 3 Feb 2014 09:15:33 -0800
Message-ID: <CAGzyod6_9H=a+uejDTEd_6HfbD2e2ynkA2frCnhk0TboricgfQ@mail.gmail.com>
From: Roberto Peon <fenix@google.com>
To: Tobias Oberstein <tobias.oberstein@tavendo.de>
Content-Type: multipart/alternative; boundary=001a11c11feee160b904f183abba
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 03 Feb 2014 17:21:08 -0000

--001a11c11feee160b904f183abba
Content-Type: text/plain; charset=UTF-8

I'm interested in supporting the W3C WS API.
I'm interested in correctly muxing, demuxing, and loadbalancing of
streams-of-messages without needing to understand application-layer
semantics.
I'm interested in enabling a single transport-layer session for both HTTP,
WS, and potentially other content.

I'm not at all interested in needing to implement multiple separate
protocols in order to implement the WS API.
I'm uninterested in small-number-of-byte savings.
I am uninterested in any extensions or extension frameworks that require
any extension-specific knowledge in intermediaries.


On Fri, Jan 31, 2014 at 4:13 AM, Tobias Oberstein <
tobias.oberstein@tavendo.de> wrote:

> Hello Yutaka,
>
>
>
> my priorities:
>
>
>
> 1. allows to provide W3C WebSocket API
>
> 2. no "80%-WebSocket-protocol-compatible" design (either compatible with
> full RFC6455 incl. any extensions or outright independent of WS)
>
> 3. does not put servers that provide both native WS and SPDY (but not any
> SPDY/WS) on same origin at a disadvantage
>
> 4. minimizes code coupling between WS and SPDY within implementations
>
>
>
> /Tobias
>
>
>
> *Von:* yhirano@google.com [mailto:yhirano@google.com] *Im Auftrag von *Yutaka
> Hirano
> *Gesendet:* Freitag, 31. Januar 2014 11:41
> *An:* Tobias Oberstein
> *Cc:* Roberto Peon; Peter Thorson; hybi@ietf.org
>
> *Betreff:* Re: [hybi] Discontinuation of mux standardizaton in favor of
> WS/HTTP/2.0
>
>
>
> Seriously: _what_ is the goal of this effort?
>
> My motivation is, specifying a multiplexed bidirectional message protocol
> conforming to the W3C WebSocket API on HTTP/2.0.
>
>
>
>  - The W3C WebSocket API
>
>  - Multiplexing
>
> is mandatory for me.
>
> Since It is built on HTTP/2.0, I prefer strongly the following:
>
>  - A mechanism that enables intermediaries to transfer WS over HTTP/2.0
> without fully understanding WS over HTTP/2.0.
>
>
>
> The remaining, though they are preferable, are not mandatory for me.
>
>  - RFC6455 conformance
>
>  - small amount of bandwidth efficiency
>
>  - the spec cleanness
>
>  - less-invasiveness
>
>
>
> Note: It does not mean I can't accept WS-TUNNEL.
>
>
>
> Tobias, Roberto, Peter, can you share your priority?
>
>
>
>
>
> On Thu, Jan 30, 2014 at 8:54 PM, Tobias Oberstein <
> tobias.oberstein@tavendo.de> wrote:
>
> >I don't find the frame-size arguments very compelling since I haven't
> seen or met anyone with a substantial deployment who cares significantly
> about the overhead difference given packet, TLS, etc. overheads.
> >There is no deployment that we know about that uses frame-based
> extensibility- we can (and should) remove it from the standard since it
> hasn't proven to be interesting.
>
> There is a draft for a WebSocket extension ("permessage-priority") being
> currently developed that makes use of this and we will start an
> experimental implementation in Autobahn. It _is_ interesting.
>
> In any case, the ability for framing, and the ability for WS extensions to
> interleave WS messages is part of RFC6455.
>
> Why remove something from RFC6455 just to fit the limited capabilities
> that SPDY offers?
>
>
> >I'm not sure I agree with your statement that libraries supporting HTTP/2
> will include necessarily also have WS framing. It seems unlikely to me-- I
> know of at least one large deployment where this isn't at all true.
>
> I'd put it like this: any networking library that wants to support
> "messaging" on the Web will need to support native WebSocket, plus
> optionally any SPDY-based messaging.
>
> Since, WebSocket is deployed _today_ and HTTP2 won't be everywhere anytime
> soon.
>
> E.g. HTTP/2 support on Windows seems to be coming only on Windows 8+
> (since they have HTTP in-kernel). So native WebSocket is the only messaging
> that works on any IE version on Windows <8.
>
>
> >The HTTP/2 standard has (or shortly will have, as soon as the pull
> request goes in) all support necessary to handle any stream-of-message
> protocol given support of END_MESSAGE.
> >I'd strongly prefer that we map WS to those primitives so it, and any
> other stream-of-message protocols can be handled using loadbalancer/routing
> infrastructure which is and can remain protocol agnostic.
>
> That's fine, let's just not call it WebSocket-over-SPDY then;)
>
> Seriously: _what_ is the goal of this effort?
>
> /Tobias
>
>
>
> -=R
>
> On Wed, Jan 29, 2014 at 5:56 AM, Tobias Oberstein <
> tobias.oberstein@tavendo.de> wrote:
> >The proposal is mapping one WS Message to one HEADERS + multiple DATAs.
> >For large messages, it can achieve less overhead than WS-TUNNEL at the
> cost of some RFC6455 compatibility.
> How can it achieve less overhead?
>
> A SPDY HEADER with information from first WS frame seems to be always
> bigger than a SPDY DATA header (8 octets) + WS frame header.
>
> The only situation I can think of: a large, _fragmented_ WS message is
> reframed into a SPDY HEADER+DATAs sequence.
>
> Effectively, you do away with the individual WS frame information. And
> this exactly breaks full RFC6455 compatibility.
>
> /Tobias
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi
>
>
>

--001a11c11feee160b904f183abba
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><span style=3D"font-family:arial,sans-serif;font-size:13px=
">I&#39;m interested in supporting the W3C WS API.</span><div style=3D"font=
-family:arial,sans-serif;font-size:13px">I&#39;m interested in correctly mu=
xing, demuxing, and loadbalancing of streams-of-messages without needing to=
 understand application-layer semantics.</div>
<div style=3D"font-family:arial,sans-serif;font-size:13px">I&#39;m interest=
ed in enabling a single transport-layer session for both HTTP, WS, and pote=
ntially other content.</div><div style=3D"font-family:arial,sans-serif;font=
-size:13px">
<br></div><div style=3D"font-family:arial,sans-serif;font-size:13px">I&#39;=
m not at all interested in needing to implement multiple separate protocols=
 in order to implement the WS API.</div><div style=3D"font-family:arial,san=
s-serif;font-size:13px">
I&#39;m uninterested in small-number-of-byte savings.</div><div style=3D"fo=
nt-family:arial,sans-serif;font-size:13px">I am uninterested in any extensi=
ons or extension frameworks that require any extension-specific knowledge i=
n intermediaries.=C2=A0</div>
</div><div class=3D"gmail_extra"><br><br><div class=3D"gmail_quote">On Fri,=
 Jan 31, 2014 at 4:13 AM, Tobias Oberstein <span dir=3D"ltr">&lt;<a href=3D=
"mailto:tobias.oberstein@tavendo.de" target=3D"_blank">tobias.oberstein@tav=
endo.de</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div lang=3D"DE" link=3D"blue" vlink=3D"purp=
le"><div><p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family=
:&quot;Calibri&quot;,&quot;sans-serif&quot;;color:#1f497d">Hello Yutaka,<u>=
</u><u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></spa=
n></p><p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&q=
uot;Calibri&quot;,&quot;sans-serif&quot;;color:#1f497d">my priorities:<u></=
u><u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></spa=
n></p><p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&q=
uot;Calibri&quot;,&quot;sans-serif&quot;;color:#1f497d">1. allows to provid=
e W3C WebSocket API<u></u><u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">2. no &quot;80%-WebSocket=
-protocol-compatible&quot; design (either compatible with full RFC6455 incl=
. any extensions or outright independent of WS)<u></u><u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">3. does not put servers t=
hat provide both native WS and SPDY (but not any SPDY/WS) on same origin at=
 a disadvantage<u></u><u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">4. minimizes code couplin=
g between WS and SPDY within implementations<u></u><u></u></span></p><p cla=
ss=3D"MsoNormal">
<span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-=
serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></span></p><p class=3D"MsoNo=
rmal"><span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot=
;sans-serif&quot;;color:#1f497d">/Tobias<u></u><u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></spa=
n></p><div style=3D"border:none;border-left:solid blue 1.5pt;padding:0cm 0c=
m 0cm 4.0pt">
<div><div style=3D"border:none;border-top:solid #e1e1e1 1.0pt;padding:3.0pt=
 0cm 0cm 0cm"><p class=3D"MsoNormal"><b><span style=3D"font-size:11.0pt;fon=
t-family:&quot;Calibri&quot;,&quot;sans-serif&quot;">Von:</span></b><span s=
tyle=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif&q=
uot;"> <a href=3D"mailto:yhirano@google.com" target=3D"_blank">yhirano@goog=
le.com</a> [mailto:<a href=3D"mailto:yhirano@google.com" target=3D"_blank">=
yhirano@google.com</a>] <b>Im Auftrag von </b>Yutaka Hirano<br>
<b>Gesendet:</b> Freitag, 31. Januar 2014 11:41<br><b>An:</b> Tobias Oberst=
ein<br><b>Cc:</b> Roberto Peon; Peter Thorson; <a href=3D"mailto:hybi@ietf.=
org" target=3D"_blank">hybi@ietf.org</a></span></p><div class=3D"im"><br><b=
>Betreff:</b> Re: [hybi] Discontinuation of mux standardizaton in favor of =
WS/HTTP/2.0<u></u><u></u></div>
<p></p></div></div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p><div><blo=
ckquote style=3D"border:none;border-left:solid #cccccc 1.0pt;padding:0cm 0c=
m 0cm 6.0pt;margin-left:4.8pt;margin-right:0cm"><p class=3D"MsoNormal"><spa=
n style=3D"font-size:10.0pt;font-family:&quot;Arial&quot;,&quot;sans-serif&=
quot;">Seriously: _what_ is the goal of this effort?</span><u></u><u></u></=
p>
</blockquote><div><div class=3D"h5"><div><p class=3D"MsoNormal">My motivati=
on is, specifying a multiplexed bidirectional message protocol conforming t=
o the W3C WebSocket API on HTTP/2.0.<u></u><u></u></p></div><div><p class=
=3D"MsoNormal">
<u></u>=C2=A0<u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- The W3C W=
ebSocket API<u></u><u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- Mul=
tiplexing<u></u><u></u></p></div><div><p class=3D"MsoNormal">is mandatory f=
or me.<u></u><u></u></p>
</div><div><p class=3D"MsoNormal">Since It is built on HTTP/2.0, I prefer s=
trongly the following:<u></u><u></u></p></div><div><p class=3D"MsoNormal">=
=C2=A0- A mechanism that enables intermediaries to transfer WS over HTTP/2.=
0 without fully understanding WS over HTTP/2.0.<u></u><u></u></p>
</div><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div><div><p cla=
ss=3D"MsoNormal">The remaining, though they are preferable, are not mandato=
ry for me.<u></u><u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- RFC64=
55 conformance<u></u><u></u></p>
</div><div><p class=3D"MsoNormal">=C2=A0- small amount of bandwidth efficie=
ncy<u></u><u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- the spec cle=
anness<u></u><u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- less-inva=
siveness<u></u><u></u></p>
</div><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div><div><p cla=
ss=3D"MsoNormal">Note: It does not mean I can&#39;t accept WS-TUNNEL.<u></u=
><u></u></p></div><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div=
><div><p class=3D"MsoNormal">
Tobias, Roberto, Peter, can you share your priority?<u></u><u></u></p></div=
><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div></div></div></di=
v><div><div class=3D"h5"><div><p class=3D"MsoNormal" style=3D"margin-bottom=
:12.0pt">
<u></u>=C2=A0<u></u></p><div><p class=3D"MsoNormal">On Thu, Jan 30, 2014 at=
 8:54 PM, Tobias Oberstein &lt;<a href=3D"mailto:tobias.oberstein@tavendo.d=
e" target=3D"_blank">tobias.oberstein@tavendo.de</a>&gt; wrote:<u></u><u></=
u></p><blockquote style=3D"border:none;border-left:solid #cccccc 1.0pt;padd=
ing:0cm 0cm 0cm 6.0pt;margin-left:4.8pt;margin-right:0cm">
<div><p class=3D"MsoNormal" style=3D"margin-bottom:12.0pt">&gt;I don&#39;t =
find the frame-size arguments very compelling since I haven&#39;t seen or m=
et anyone with a substantial deployment who cares significantly about the o=
verhead difference given packet, TLS, etc. overheads.<br>
&gt;There is no deployment that we know about that uses frame-based extensi=
bility- we can (and should) remove it from the standard since it hasn&#39;t=
 proven to be interesting.<u></u><u></u></p></div><p class=3D"MsoNormal">
There is a draft for a WebSocket extension (&quot;permessage-priority&quot;=
) being currently developed that makes use of this and we will start an exp=
erimental implementation in Autobahn. It _is_ interesting.<br><br>In any ca=
se, the ability for framing, and the ability for WS extensions to interleav=
e WS messages is part of RFC6455.<br>
<br>Why remove something from RFC6455 just to fit the limited capabilities =
that SPDY offers?<u></u><u></u></p><div><p class=3D"MsoNormal" style=3D"mar=
gin-bottom:12.0pt"><br>&gt;I&#39;m not sure I agree with your statement tha=
t libraries supporting HTTP/2 will include necessarily also have WS framing=
. It seems unlikely to me-- I know of at least one large deployment where t=
his isn&#39;t at all true.<u></u><u></u></p>
</div><p class=3D"MsoNormal">I&#39;d put it like this: any networking libra=
ry that wants to support &quot;messaging&quot; on the Web will need to supp=
ort native WebSocket, plus optionally any SPDY-based messaging.<br><br>Sinc=
e, WebSocket is deployed _today_ and HTTP2 won&#39;t be everywhere anytime =
soon.<br>
<br>E.g. HTTP/2 support on Windows seems to be coming only on Windows 8+ (s=
ince they have HTTP in-kernel). So native WebSocket is the only messaging t=
hat works on any IE version on Windows &lt;8.<u></u><u></u></p><div><p clas=
s=3D"MsoNormal" style=3D"margin-bottom:12.0pt">
<br>&gt;The HTTP/2 standard has (or shortly will have, as soon as the pull =
request goes in) all support necessary to handle any stream-of-message prot=
ocol given support of END_MESSAGE.<br>&gt;I&#39;d strongly prefer that we m=
ap WS to those primitives so it, and any other stream-of-message protocols =
can be handled using loadbalancer/routing infrastructure which is and can r=
emain protocol agnostic.<u></u><u></u></p>
</div><p class=3D"MsoNormal">That&#39;s fine, let&#39;s just not call it We=
bSocket-over-SPDY then;)<br><br>Seriously: _what_ is the goal of this effor=
t?<br><span style=3D"color:#888888"><br><span>/Tobias</span></span><u></u><=
u></u></p>
<div><div><p class=3D"MsoNormal" style=3D"margin-bottom:12.0pt"><br><br>-=
=3DR<br><br>On Wed, Jan 29, 2014 at 5:56 AM, Tobias Oberstein &lt;<a href=
=3D"mailto:tobias.oberstein@tavendo.de" target=3D"_blank">tobias.oberstein@=
tavendo.de</a>&gt; wrote:<br>
&gt;The proposal is mapping one WS Message to one HEADERS + multiple DATAs.=
<br>&gt;For large messages, it can achieve less overhead than WS-TUNNEL at =
the cost of some RFC6455 compatibility.<br>How can it achieve less overhead=
?<br>
<br>A SPDY HEADER with information from first WS frame seems to be always b=
igger than a SPDY DATA header (8 octets) + WS frame header.<br><br>The only=
 situation I can think of: a large, _fragmented_ WS message is reframed int=
o a SPDY HEADER+DATAs sequence.<br>
<br>Effectively, you do away with the individual WS frame information. And =
this exactly breaks full RFC6455 compatibility.<br><br>/Tobias<br>_________=
______________________________________<br>hybi mailing list<br><a href=3D"m=
ailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><u></u><u></u></p></div></div><=
/blockquote></div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div></di=
v></div>
</div></div></div></blockquote></div><br></div>

--001a11c11feee160b904f183abba--

From webmaster@zaphoyd.com  Mon Feb  3 12:36:52 2014
Return-Path: <webmaster@zaphoyd.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 226191A01D6 for <hybi@ietfa.amsl.com>; Mon,  3 Feb 2014 12:36:52 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 1.401
X-Spam-Level: *
X-Spam-Status: No, score=1.401 tagged_above=-999 required=5 tests=[BAYES_50=0.8, HTML_MESSAGE=0.001, J_CHICKENPOX_65=0.6] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id gH3Mp5DVx_y1 for <hybi@ietfa.amsl.com>; Mon,  3 Feb 2014 12:36:48 -0800 (PST)
Received: from leaf102.mfilter.dimenoc.com (tree.mfilter.dimenoc.com [72.29.89.3]) by ietfa.amsl.com (Postfix) with ESMTP id 84B7F1A020C for <hybi@ietf.org>; Mon,  3 Feb 2014 12:36:48 -0800 (PST)
Received: from localhost (localhost [127.0.0.1]) by leaf102.mfilter.dimenoc.com (Postfix) with ESMTP id A5D3D81C569 for <hybi@ietf.org>; Mon,  3 Feb 2014 15:36:40 -0500 (EST)
X-Virus-Scanned: amavisd-new at leaf.mfilter.dimenoc.com
X-Amavis-Alert: BAD HEADER SECTION, Header line longer than 998 characters: References: <CA[...]
Received: from leaf102.mfilter.dimenoc.com ([IPv6:::ffff:127.0.0.1]) by localhost (leaf102.mfilter.dimenoc.com [IPv6:::ffff:127.0.0.1]) (amavisd-new, port 10024) with ESMTP id k3ppwdqC7R6y for <hybi@ietf.org>; Mon,  3 Feb 2014 15:36:37 -0500 (EST)
Received: from sh78.surpasshosting.com (sh78.surpasshosting.com [72.29.64.142]) by leaf102.mfilter.dimenoc.com (Postfix) with ESMTP for <hybi@ietf.org>; Mon,  3 Feb 2014 15:36:37 -0500 (EST)
Received: from ranna.uchicago.edu ([128.135.45.206]:60130) by sh78.surpasshosting.com with esmtpsa (TLSv1:AES128-SHA:128) (Exim 4.80.1) (envelope-from <webmaster@zaphoyd.com>) id 1WAQFt-0002o3-5I; Mon, 03 Feb 2014 15:36:37 -0500
Content-Type: multipart/alternative; boundary="Apple-Mail=_0A27658C-8BE7-4662-9994-2DDB24D455AB"
Mime-Version: 1.0 (Mac OS X Mail 7.1 \(1827\))
From: Peter Thorson <webmaster@zaphoyd.com>
In-Reply-To: <CAGzyod6_9H=a+uejDTEd_6HfbD2e2ynkA2frCnhk0TboricgfQ@mail.gmail.com>
Date: Mon, 3 Feb 2014 14:36:25 -0600
Message-Id: <3CF0431C-3AF2-4F6D-82C6-3680976C0FA7@zaphoyd.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <CABihn6GSr...
To: Yutaka Hirano <yhirano@chromium.org>
X-Mailer: Apple Mail (2.1827)
X-AntiAbuse: This header was added to track abuse, please include it with any abuse report
X-AntiAbuse: Primary Hostname - sh78.surpasshosting.com
X-AntiAbuse: Original Domain - ietf.org
X-AntiAbuse: Originator/Caller UID/GID - [47 12] / [47 12]
X-AntiAbuse: Sender Address Domain - zaphoyd.com
X-Get-Message-Sender-Via: sh78.surpasshosting.com: authenticated_id: webmaster@zaphoyd.com
X-Source: 
X-Source-Args: 
X-Source-Dir: 
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 03 Feb 2014 20:36:52 -0000

--Apple-Mail=_0A27658C-8BE7-4662-9994-2DDB24D455AB
Content-Transfer-Encoding: quoted-printable
Content-Type: text/plain;
	charset=windows-1252

My priorities:

Important:
1. W3C WebSocket API and muxing go without saying.
2. Dedicated RFC6455 connections remain a viable option in modern =
browsers.
3. Efficient and clearly defined method for detecting HTTP/2.0 messaging =
support, choosing HTTP/2.0 vs raw RFC6455, when to fall back, etc.
4. Payload compression with appropriate configurability (either via =
permessage-deflate support or a native mechanism)

Nice to haves:
5. Avoid significant regressions from RFC6455 in wire overhead, latency, =
and per message fixed costs for very small (<100byte) messages.
6. Clean spec/avoiding redundant and unnecessary features of RFC6455.

Only if it doesn=92t conflict with 1-6:
7. Options for future compatibility with simple RFC6455 extensions

As long as #2 and #3 are satisfied I am less concerned by performance =
details or RFC6455 compatibility.

Question for Roberto:
> I'm not at all interested in needing to implement multiple separate =
protocols in order to implement the WS API.


Can you clarify under what context you are referring here? Is your =
meaning that Chrome.app should not have to implement both HTTP/2.0 and =
RFC6455? Or that one should be able to create an HTTP/2.0-only product =
without needing to pull in an RFC6455 library? Something else?

Question for Yutaka:
- A mechanism that enables intermediaries to transfer WS over HTTP/2.0 =
without fully understanding WS over HTTP/2.0.

Does the introduction of a "message end=94 flag that HTTP/2.0 =
intermediaries would have to understand as =93do not coalesce across=94 =
violate this constraint?

On Feb 3, 2014, at 11:15 , Roberto Peon <fenix@google.com> wrote:

> I'm interested in supporting the W3C WS API.
> I'm interested in correctly muxing, demuxing, and loadbalancing of =
streams-of-messages without needing to understand application-layer =
semantics.
> I'm interested in enabling a single transport-layer session for both =
HTTP, WS, and potentially other content.
>=20
> I'm not at all interested in needing to implement multiple separate =
protocols in order to implement the WS API.
> I'm uninterested in small-number-of-byte savings.
> I am uninterested in any extensions or extension frameworks that =
require any extension-specific knowledge in intermediaries.=20
>=20
>=20
> On Fri, Jan 31, 2014 at 4:13 AM, Tobias Oberstein =
<tobias.oberstein@tavendo.de> wrote:
> Hello Yutaka,
>=20
> =20
>=20
> my priorities:
>=20
> =20
>=20
> 1. allows to provide W3C WebSocket API
>=20
> 2. no "80%-WebSocket-protocol-compatible" design (either compatible =
with full RFC6455 incl. any extensions or outright independent of WS)
>=20
> 3. does not put servers that provide both native WS and SPDY (but not =
any SPDY/WS) on same origin at a disadvantage
>=20
> 4. minimizes code coupling between WS and SPDY within implementations
>=20
> =20
>=20
> /Tobias
>=20
> =20
>=20
> Von: yhirano@google.com [mailto:yhirano@google.com] Im Auftrag von =
Yutaka Hirano
> Gesendet: Freitag, 31. Januar 2014 11:41
> An: Tobias Oberstein
> Cc: Roberto Peon; Peter Thorson; hybi@ietf.org
>=20
>=20
> Betreff: Re: [hybi] Discontinuation of mux standardizaton in favor of =
WS/HTTP/2.0
>=20
> =20
>=20
> Seriously: _what_ is the goal of this effort?
>=20
> My motivation is, specifying a multiplexed bidirectional message =
protocol conforming to the W3C WebSocket API on HTTP/2.0.
>=20
> =20
>=20
>  - The W3C WebSocket API
>=20
>  - Multiplexing
>=20
> is mandatory for me.
>=20
> Since It is built on HTTP/2.0, I prefer strongly the following:
>=20
>  - A mechanism that enables intermediaries to transfer WS over =
HTTP/2.0 without fully understanding WS over HTTP/2.0.
>=20
> =20
>=20
> The remaining, though they are preferable, are not mandatory for me.
>=20
>  - RFC6455 conformance
>=20
>  - small amount of bandwidth efficiency
>=20
>  - the spec cleanness
>=20
>  - less-invasiveness
>=20
> =20
>=20
> Note: It does not mean I can't accept WS-TUNNEL.
>=20
> =20
>=20
> Tobias, Roberto, Peter, can you share your priority?
>=20
> =20
>=20
> =20
>=20
> On Thu, Jan 30, 2014 at 8:54 PM, Tobias Oberstein =
<tobias.oberstein@tavendo.de> wrote:
>=20
> >I don't find the frame-size arguments very compelling since I haven't =
seen or met anyone with a substantial deployment who cares significantly =
about the overhead difference given packet, TLS, etc. overheads.
> >There is no deployment that we know about that uses frame-based =
extensibility- we can (and should) remove it from the standard since it =
hasn't proven to be interesting.
>=20
> There is a draft for a WebSocket extension ("permessage-priority") =
being currently developed that makes use of this and we will start an =
experimental implementation in Autobahn. It _is_ interesting.
>=20
> In any case, the ability for framing, and the ability for WS =
extensions to interleave WS messages is part of RFC6455.
>=20
> Why remove something from RFC6455 just to fit the limited capabilities =
that SPDY offers?
>=20
>=20
> >I'm not sure I agree with your statement that libraries supporting =
HTTP/2 will include necessarily also have WS framing. It seems unlikely =
to me-- I know of at least one large deployment where this isn't at all =
true.
>=20
> I'd put it like this: any networking library that wants to support =
"messaging" on the Web will need to support native WebSocket, plus =
optionally any SPDY-based messaging.
>=20
> Since, WebSocket is deployed _today_ and HTTP2 won't be everywhere =
anytime soon.
>=20
> E.g. HTTP/2 support on Windows seems to be coming only on Windows 8+ =
(since they have HTTP in-kernel). So native WebSocket is the only =
messaging that works on any IE version on Windows <8.
>=20
>=20
> >The HTTP/2 standard has (or shortly will have, as soon as the pull =
request goes in) all support necessary to handle any stream-of-message =
protocol given support of END_MESSAGE.
> >I'd strongly prefer that we map WS to those primitives so it, and any =
other stream-of-message protocols can be handled using =
loadbalancer/routing infrastructure which is and can remain protocol =
agnostic.
>=20
> That's fine, let's just not call it WebSocket-over-SPDY then;)
>=20
> Seriously: _what_ is the goal of this effort?
>=20
> /Tobias
>=20
>=20
>=20
> -=3DR
>=20
> On Wed, Jan 29, 2014 at 5:56 AM, Tobias Oberstein =
<tobias.oberstein@tavendo.de> wrote:
> >The proposal is mapping one WS Message to one HEADERS + multiple =
DATAs.
> >For large messages, it can achieve less overhead than WS-TUNNEL at =
the cost of some RFC6455 compatibility.
> How can it achieve less overhead?
>=20
> A SPDY HEADER with information from first WS frame seems to be always =
bigger than a SPDY DATA header (8 octets) + WS frame header.
>=20
> The only situation I can think of: a large, _fragmented_ WS message is =
reframed into a SPDY HEADER+DATAs sequence.
>=20
> Effectively, you do away with the individual WS frame information. And =
this exactly breaks full RFC6455 compatibility.
>=20
> /Tobias
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi
>=20
> =20
>=20
>=20
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi


--Apple-Mail=_0A27658C-8BE7-4662-9994-2DDB24D455AB
Content-Transfer-Encoding: quoted-printable
Content-Type: text/html;
	charset=windows-1252

<html><head><meta http-equiv=3D"Content-Type" content=3D"text/html =
charset=3Dwindows-1252"></head><body style=3D"word-wrap: break-word; =
-webkit-nbsp-mode: space; -webkit-line-break: =
after-white-space;"><div>My =
priorities:</div><div><br></div><div>Important:</div><div>1. W3C =
WebSocket API and muxing go without saying.</div><div>2. Dedicated =
RFC6455 connections remain a viable option in modern =
browsers.</div><div>3. Efficient and clearly defined method for =
detecting HTTP/2.0 messaging support, choosing HTTP/2.0 vs raw RFC6455, =
when to fall back, etc.</div><div>4. Payload compression with =
appropriate configurability (either via permessage-deflate support or a =
native mechanism)</div><div><br></div><div><div>Nice to =
haves:</div><div>5. Avoid significant regressions from RFC6455 in wire =
overhead, latency, and per message fixed costs for very small =
(&lt;100byte) messages.</div></div><div>6. Clean spec/avoiding redundant =
and unnecessary features of RFC6455.</div><div><br></div><div>Only if it =
doesn=92t conflict with 1-6:</div><div>7. Options for future =
compatibility with simple RFC6455 extensions</div><div><br></div><div>As =
long as #2 and #3 are satisfied I am less concerned by performance =
details or RFC6455 compatibility.</div><div><br></div><div>Question for =
Roberto:</div><div><blockquote type=3D"cite"><div dir=3D"ltr"><div =
style=3D"font-family: arial, sans-serif; font-size: 13px;">I'm not at =
all interested in needing to implement multiple separate protocols in =
order to implement the WS =
API.</div></div></blockquote></div><div><br></div><div>Can you clarify =
under what context you are referring here? Is your meaning that =
Chrome.app should not have to implement both HTTP/2.0 and RFC6455? Or =
that one should be able to create an HTTP/2.0-only product without =
needing to pull in an RFC6455 library? Something =
else?</div><div><br></div><div>Question for Yutaka:</div><div><div =
class=3D"gmail_extra"><div class=3D"gmail_quote"><blockquote =
class=3D"gmail_quote" style=3D"margin: 0px 0px 0px 0.8ex; =
border-left-width: 1px; border-left-color: rgb(204, 204, 204); =
border-left-style: solid; padding-left: 1ex;"><div lang=3D"DE" =
link=3D"blue" vlink=3D"purple"><div style=3D"border-style: none none =
none solid; border-left-color: blue; border-left-width: 1.5pt; padding: =
0cm 0cm 0cm 4pt; position: static; z-index: auto;"><div class=3D"h5"><p =
class=3D"MsoNormal">- A mechanism that enables intermediaries to =
transfer WS over HTTP/2.0 without fully understanding WS over =
HTTP/2.0.</p></div></div></div></blockquote></div></div></div><div>Does =
the introduction of a "message end=94 flag that HTTP/2.0 intermediaries =
would have to understand as =93do not coalesce across=94 violate this =
constraint?</div><br><div><div>On Feb 3, 2014, at 11:15 , Roberto Peon =
&lt;<a href=3D"mailto:fenix@google.com">fenix@google.com</a>&gt; =
wrote:</div><br class=3D"Apple-interchange-newline"><blockquote =
type=3D"cite"><div dir=3D"ltr"><span =
style=3D"font-family:arial,sans-serif;font-size:13px">I'm interested in =
supporting the W3C WS API.</span><div =
style=3D"font-family:arial,sans-serif;font-size:13px">I'm interested in =
correctly muxing, demuxing, and loadbalancing of streams-of-messages =
without needing to understand application-layer semantics.</div>
<div style=3D"font-family:arial,sans-serif;font-size:13px">I'm =
interested in enabling a single transport-layer session for both HTTP, =
WS, and potentially other content.</div><div =
style=3D"font-family:arial,sans-serif;font-size:13px">
<br></div><div style=3D"font-family:arial,sans-serif;font-size:13px">I'm =
not at all interested in needing to implement multiple separate =
protocols in order to implement the WS API.</div><div =
style=3D"font-family:arial,sans-serif;font-size:13px">
I'm uninterested in small-number-of-byte savings.</div><div =
style=3D"font-family:arial,sans-serif;font-size:13px">I am uninterested =
in any extensions or extension frameworks that require any =
extension-specific knowledge in intermediaries.&nbsp;</div>
</div><div class=3D"gmail_extra"><br><br><div class=3D"gmail_quote">On =
Fri, Jan 31, 2014 at 4:13 AM, Tobias Oberstein <span dir=3D"ltr">&lt;<a =
href=3D"mailto:tobias.oberstein@tavendo.de" =
target=3D"_blank">tobias.oberstein@tavendo.de</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 =
.8ex;border-left:1px #ccc solid;padding-left:1ex"><div lang=3D"DE" =
link=3D"blue" vlink=3D"purple"><p class=3D"MsoNormal"><span =
style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif=
&quot;;color:#1f497d">Hello Yutaka,<u></u><u></u></span></p><p =
class=3D"MsoNormal"><span =
style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif=
&quot;;color:#1f497d"><u></u>&nbsp;<u></u></span></p><p =
class=3D"MsoNormal"><span =
style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif=
&quot;;color:#1f497d">my priorities:<u></u><u></u></span></p><p =
class=3D"MsoNormal"><span =
style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif=
&quot;;color:#1f497d"><u></u>&nbsp;<u></u></span></p><p =
class=3D"MsoNormal"><span =
style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif=
&quot;;color:#1f497d">1. allows to provide W3C WebSocket =
API<u></u><u></u></span></p><p class=3D"MsoNormal"><span =
style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif=
&quot;;color:#1f497d">2. no "80%-WebSocket-protocol-compatible" design =
(either compatible with full RFC6455 incl. any extensions or outright =
independent of WS)<u></u><u></u></span></p><p class=3D"MsoNormal"><span =
style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif=
&quot;;color:#1f497d">3. does not put servers that provide both native =
WS and SPDY (but not any SPDY/WS) on same origin at a =
disadvantage<u></u><u></u></span></p><p class=3D"MsoNormal"><span =
style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif=
&quot;;color:#1f497d">4. minimizes code coupling between WS and SPDY =
within implementations<u></u><u></u></span></p><p class=3D"MsoNormal">
<span =
style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif=
&quot;;color:#1f497d"><u></u>&nbsp;<u></u></span></p><p =
class=3D"MsoNormal"><span =
style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif=
&quot;;color:#1f497d">/Tobias<u></u><u></u></span></p><p =
class=3D"MsoNormal"><span =
style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif=
&quot;;color:#1f497d"><u></u>&nbsp;<u></u></span></p><div =
style=3D"border-style: none none none solid; border-left-color: blue; =
border-left-width: 1.5pt; padding: 0cm 0cm 0cm 4pt; position: static; =
z-index: auto;">
<div><div style=3D"border:none;border-top:solid #e1e1e1 =
1.0pt;padding:3.0pt 0cm 0cm 0cm"><p class=3D"MsoNormal"><b><span =
style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif=
&quot;">Von:</span></b><span =
style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif=
&quot;"> <a href=3D"mailto:yhirano@google.com" =
target=3D"_blank">yhirano@google.com</a> [mailto:<a =
href=3D"mailto:yhirano@google.com" =
target=3D"_blank">yhirano@google.com</a>] <b>Im Auftrag von </b>Yutaka =
Hirano<br>
<b>Gesendet:</b> Freitag, 31. Januar 2014 11:41<br><b>An:</b> Tobias =
Oberstein<br><b>Cc:</b> Roberto Peon; Peter Thorson; <a =
href=3D"mailto:hybi@ietf.org" =
target=3D"_blank">hybi@ietf.org</a></span></p><div =
class=3D"im"><br><b>Betreff:</b> Re: [hybi] Discontinuation of mux =
standardizaton in favor of WS/HTTP/2.0<u></u><u></u></div><div><br =
class=3D"webkit-block-placeholder"></div></div></div><p =
class=3D"MsoNormal"><u></u>&nbsp;<u></u></p><div><blockquote =
style=3D"border:none;border-left:solid #cccccc 1.0pt;padding:0cm 0cm 0cm =
6.0pt;margin-left:4.8pt;margin-right:0cm"><p class=3D"MsoNormal"><span =
style=3D"font-size:10.0pt;font-family:&quot;Arial&quot;,&quot;sans-serif&q=
uot;">Seriously: _what_ is the goal of this =
effort?</span><u></u><u></u></p>
</blockquote><div><div class=3D"h5"><div><p class=3D"MsoNormal">My =
motivation is, specifying a multiplexed bidirectional message protocol =
conforming to the W3C WebSocket API on =
HTTP/2.0.<u></u><u></u></p></div><div><p class=3D"MsoNormal">
<u></u>&nbsp;<u></u></p></div><div><p class=3D"MsoNormal">&nbsp;- The =
W3C WebSocket API<u></u><u></u></p></div><div><p =
class=3D"MsoNormal">&nbsp;- Multiplexing<u></u><u></u></p></div><div><p =
class=3D"MsoNormal">is mandatory for me.<u></u><u></u></p>
</div><div><p class=3D"MsoNormal">Since It is built on HTTP/2.0, I =
prefer strongly the following:<u></u><u></u></p></div><div><p =
class=3D"MsoNormal">&nbsp;- A mechanism that enables intermediaries to =
transfer WS over HTTP/2.0 without fully understanding WS over =
HTTP/2.0.<u></u><u></u></p>
</div><div><p class=3D"MsoNormal"><u></u>&nbsp;<u></u></p></div><div><p =
class=3D"MsoNormal">The remaining, though they are preferable, are not =
mandatory for me.<u></u><u></u></p></div><div><p =
class=3D"MsoNormal">&nbsp;- RFC6455 conformance<u></u><u></u></p>
</div><div><p class=3D"MsoNormal">&nbsp;- small amount of bandwidth =
efficiency<u></u><u></u></p></div><div><p class=3D"MsoNormal">&nbsp;- =
the spec cleanness<u></u><u></u></p></div><div><p =
class=3D"MsoNormal">&nbsp;- less-invasiveness<u></u><u></u></p>
</div><div><p class=3D"MsoNormal"><u></u>&nbsp;<u></u></p></div><div><p =
class=3D"MsoNormal">Note: It does not mean I can't accept =
WS-TUNNEL.<u></u><u></u></p></div><div><p =
class=3D"MsoNormal"><u></u>&nbsp;<u></u></p></div><div><p =
class=3D"MsoNormal">
Tobias, Roberto, Peter, can you share your =
priority?<u></u><u></u></p></div><div><p =
class=3D"MsoNormal"><u></u>&nbsp;<u></u></p></div></div></div></div><div><=
div class=3D"h5"><p class=3D"MsoNormal" style=3D"margin-bottom:12.0pt">
<u></u>&nbsp;<u></u></p><div><p class=3D"MsoNormal">On Thu, Jan 30, 2014 =
at 8:54 PM, Tobias Oberstein &lt;<a =
href=3D"mailto:tobias.oberstein@tavendo.de" =
target=3D"_blank">tobias.oberstein@tavendo.de</a>&gt; =
wrote:<u></u><u></u></p><blockquote style=3D"border:none;border-left:solid=
 #cccccc 1.0pt;padding:0cm 0cm 0cm =
6.0pt;margin-left:4.8pt;margin-right:0cm">
<div><p class=3D"MsoNormal" style=3D"margin-bottom:12.0pt">&gt;I don't =
find the frame-size arguments very compelling since I haven't seen or =
met anyone with a substantial deployment who cares significantly about =
the overhead difference given packet, TLS, etc. overheads.<br>
&gt;There is no deployment that we know about that uses frame-based =
extensibility- we can (and should) remove it from the standard since it =
hasn't proven to be interesting.<u></u><u></u></p></div><p =
class=3D"MsoNormal">
There is a draft for a WebSocket extension ("permessage-priority") being =
currently developed that makes use of this and we will start an =
experimental implementation in Autobahn. It _is_ interesting.<br><br>In =
any case, the ability for framing, and the ability for WS extensions to =
interleave WS messages is part of RFC6455.<br>
<br>Why remove something from RFC6455 just to fit the limited =
capabilities that SPDY offers?<u></u><u></u></p><div><p =
class=3D"MsoNormal" style=3D"margin-bottom:12.0pt"><br>&gt;I'm not sure =
I agree with your statement that libraries supporting HTTP/2 will =
include necessarily also have WS framing. It seems unlikely to me-- I =
know of at least one large deployment where this isn't at all =
true.<u></u><u></u></p>
</div><p class=3D"MsoNormal">I'd put it like this: any networking =
library that wants to support "messaging" on the Web will need to =
support native WebSocket, plus optionally any SPDY-based =
messaging.<br><br>Since, WebSocket is deployed _today_ and HTTP2 won't =
be everywhere anytime soon.<br>
<br>E.g. HTTP/2 support on Windows seems to be coming only on Windows 8+ =
(since they have HTTP in-kernel). So native WebSocket is the only =
messaging that works on any IE version on Windows =
&lt;8.<u></u><u></u></p><div><p class=3D"MsoNormal" =
style=3D"margin-bottom:12.0pt">
<br>&gt;The HTTP/2 standard has (or shortly will have, as soon as the =
pull request goes in) all support necessary to handle any =
stream-of-message protocol given support of END_MESSAGE.<br>&gt;I'd =
strongly prefer that we map WS to those primitives so it, and any other =
stream-of-message protocols can be handled using loadbalancer/routing =
infrastructure which is and can remain protocol =
agnostic.<u></u><u></u></p>
</div><p class=3D"MsoNormal">That's fine, let's just not call it =
WebSocket-over-SPDY then;)<br><br>Seriously: _what_ is the goal of this =
effort?<br><span =
style=3D"color:#888888"><br><span>/Tobias</span></span><u></u><u></u></p>
<div><p class=3D"MsoNormal" =
style=3D"margin-bottom:12.0pt"><br><br>-=3DR<br><br>On Wed, Jan 29, 2014 =
at 5:56 AM, Tobias Oberstein &lt;<a =
href=3D"mailto:tobias.oberstein@tavendo.de" =
target=3D"_blank">tobias.oberstein@tavendo.de</a>&gt; wrote:<br>
&gt;The proposal is mapping one WS Message to one HEADERS + multiple =
DATAs.<br>&gt;For large messages, it can achieve less overhead than =
WS-TUNNEL at the cost of some RFC6455 compatibility.<br>How can it =
achieve less overhead?<br>
<br>A SPDY HEADER with information from first WS frame seems to be =
always bigger than a SPDY DATA header (8 octets) + WS frame =
header.<br><br>The only situation I can think of: a large, _fragmented_ =
WS message is reframed into a SPDY HEADER+DATAs sequence.<br>
<br>Effectively, you do away with the individual WS frame information. =
And this exactly breaks full RFC6455 =
compatibility.<br><br>/Tobias<br>_________________________________________=
______<br>hybi mailing list<br><a href=3D"mailto:hybi@ietf.org" =
target=3D"_blank">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" =
target=3D"_blank">https://www.ietf.org/mailman/listinfo/hybi</a><u></u><u>=
</u></p></div></blockquote></div><p =
class=3D"MsoNormal"><u></u>&nbsp;<u></u></p></div></div>
</div></div></blockquote></div><br></div>
_______________________________________________<br>hybi mailing =
list<br><a =
href=3D"mailto:hybi@ietf.org">hybi@ietf.org</a><br>https://www.ietf.org/ma=
ilman/listinfo/hybi<br></blockquote></div><br></body></html>=

--Apple-Mail=_0A27658C-8BE7-4662-9994-2DDB24D455AB--

From fenix@google.com  Mon Feb  3 12:50:22 2014
Return-Path: <fenix@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 3C42E1A022F for <hybi@ietfa.amsl.com>; Mon,  3 Feb 2014 12:50:22 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.313
X-Spam-Level: 
X-Spam-Status: No, score=-1.313 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, J_CHICKENPOX_65=0.6, RP_MATCHES_RCVD=-0.535, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id h4TmnnGJR_sa for <hybi@ietfa.amsl.com>; Mon,  3 Feb 2014 12:50:19 -0800 (PST)
Received: from mail-qc0-x232.google.com (mail-qc0-x232.google.com [IPv6:2607:f8b0:400d:c01::232]) by ietfa.amsl.com (Postfix) with ESMTP id CBBE71A020C for <hybi@ietf.org>; Mon,  3 Feb 2014 12:50:18 -0800 (PST)
Received: by mail-qc0-f178.google.com with SMTP id m20so12328184qcx.9 for <hybi@ietf.org>; Mon, 03 Feb 2014 12:50:18 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=QnXiiOc9Ku3LipEFDlHQ8JjIJG0E/XsSpyOFF2Tthug=; b=mJDfDFfapNT/6yLrig54g8BpSTvXnvFuXipgW0H0f3vYTAG60Y2NsYRa2x97A0b0Ou yggIdQnZvXe8CYZIH5T55kq6zF+pYY7pT8tYBVqVpWriHHW+Q8Lks9tapmKKeJJ3V6l7 i364Be3uWRc5N9NlCSjMVg+AjEJ4V6n9nABAdjQsyaTBIKbK5oQOSE0OXZpdy6sX7ncI F6qrSVfH/psYRvLaWA3Y2dbsDu1ZGmet28kOlOtgfflbTxKkZsKNtFDaljioNsSYf/+c lBA9wl9xKUcyiS4eqkJ3YJWo2fT3IHt+VYyv0uHvsWKQqQdqJbQyXPQL1qLEoxmSamhT BBSg==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=QnXiiOc9Ku3LipEFDlHQ8JjIJG0E/XsSpyOFF2Tthug=; b=jpfItf5bdNDu6gaMPkdlzyZexQh6zk9Xq0wEGHdrWhMlUesROSZCQJb7OJQp+Yz49K jvmudUW8Bkvv9u7ofI/ZTfvDSmufqfYd8wgI7DGIc/AqVXjOv+6p/SgilzNd/kXYyHXx 0bq2XsW2Zi3QVTDc2FrTtsyatIHgzMS6zRRQwZ82tFrcVYJYfrKGrD9qOxtplzrxgrkE 01gWSFoLCMg+wu0rnTJo63OO6Q2qH1dktxLWgHds6VxY5AhvO491yjBWiMV/u1pJWkL0 r8Je/MEwrlPVvYEcczxdOl2/qKNLg7NZhQdWhqkeBhbwwq0+VKMX7ZXfIOwJOEcUD4DF bM0A==
X-Gm-Message-State: ALoCoQmIN92RJuhMxhL9QZagIA++gK9gCkcTeyK0/vNG0hxwkk1occhq8473GZN6AC2JRCGR4KNG+AG4vB4qMmZotTQcI5U782kxpXuHlZMZei98FZMxhWgxFXEIjRMQQC4Hd1bI97sh4XW29rz3MyDbxRYjUqG6lwZ275NTnwEAVtU6l1E6v2EHZT6QG3t7IbsvGZ7fVf7r
MIME-Version: 1.0
X-Received: by 10.140.100.240 with SMTP id s103mr55797588qge.38.1391460618120;  Mon, 03 Feb 2014 12:50:18 -0800 (PST)
Received: by 10.229.13.7 with HTTP; Mon, 3 Feb 2014 12:50:17 -0800 (PST)
In-Reply-To: <3CF0431C-3AF2-4F6D-82C6-3680976C0FA7@zaphoyd.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <CAGzyod6_9H=a+uejDTEd_6HfbD2e2ynkA2frCnhk0TboricgfQ@mail.gmail.com> <3CF0431C-3AF2-4F6D-82C6-3680976C0FA7@zaphoyd.com>
Date: Mon, 3 Feb 2014 12:50:17 -0800
Message-ID: <CAGzyod56_vxUGg1cqPE9_SxSQnxTOxseqh+qu+MC3TkUTGmcvg@mail.gmail.com>
From: Roberto Peon <fenix@google.com>
To: Peter Thorson <webmaster@zaphoyd.com>
Content-Type: multipart/alternative; boundary=001a11c16cccdcc50a04f186aba1
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 03 Feb 2014 20:50:23 -0000

--001a11c16cccdcc50a04f186aba1
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

On Mon, Feb 3, 2014 at 12:36 PM, Peter Thorson <webmaster@zaphoyd.com>wrote=
:

> My priorities:
>
> Important:
> 1. W3C WebSocket API and muxing go without saying.
> 2. Dedicated RFC6455 connections remain a viable option in modern browser=
s.
> 3. Efficient and clearly defined method for detecting HTTP/2.0 messaging
> support, choosing HTTP/2.0 vs raw RFC6455, when to fall back, etc.
> 4. Payload compression with appropriate configurability (either via
> permessage-deflate support or a native mechanism)
>
> Nice to haves:
> 5. Avoid significant regressions from RFC6455 in wire overhead, latency,
> and per message fixed costs for very small (<100byte) messages.
> 6. Clean spec/avoiding redundant and unnecessary features of RFC6455.
>
> Only if it doesn=E2=80=99t conflict with 1-6:
> 7. Options for future compatibility with simple RFC6455 extensions
>
> As long as #2 and #3 are satisfied I am less concerned by performance
> details or RFC6455 compatibility.
>
> Question for Roberto:
>
> I'm not at all interested in needing to implement multiple separate
> protocols in order to implement the WS API.
>
>
> Can you clarify under what context you are referring here? Is your meanin=
g
> that Chrome.app should not have to implement both HTTP/2.0 and RFC6455? O=
r
> that one should be able to create an HTTP/2.0-only product without needin=
g
> to pull in an RFC6455 library? Something else?
>

Yup. One shouldn't have to write parsers for both 6455 and HTTP/2 if one is
only interested in supporting WS via HTTP/2.

Separately, I wouldn't at all mind having an ALPN identifier for RFC6455 WS
on port 443. I view that as a side-benefit of the HTTP/2 work, but not
HTTP/2 itself.

-=3DR


>
> Question for Yutaka:
>
>> - A mechanism that enables intermediaries to transfer WS over HTTP/2.0
>> without fully understanding WS over HTTP/2.0.
>>
> Does the introduction of a "message end=E2=80=9D flag that HTTP/2.0 inter=
mediaries
> would have to understand as =E2=80=9Cdo not coalesce across=E2=80=9D viol=
ate this
> constraint?
>
> On Feb 3, 2014, at 11:15 , Roberto Peon <fenix@google.com> wrote:
>
> I'm interested in supporting the W3C WS API.
> I'm interested in correctly muxing, demuxing, and loadbalancing of
> streams-of-messages without needing to understand application-layer
> semantics.
> I'm interested in enabling a single transport-layer session for both HTTP=
,
> WS, and potentially other content.
>
> I'm not at all interested in needing to implement multiple separate
> protocols in order to implement the WS API.
> I'm uninterested in small-number-of-byte savings.
> I am uninterested in any extensions or extension frameworks that require
> any extension-specific knowledge in intermediaries.
>
>
> On Fri, Jan 31, 2014 at 4:13 AM, Tobias Oberstein <
> tobias.oberstein@tavendo.de> wrote:
>
>> Hello Yutaka,
>>
>>
>>
>> my priorities:
>>
>>
>>
>> 1. allows to provide W3C WebSocket API
>>
>> 2. no "80%-WebSocket-protocol-compatible" design (either compatible with
>> full RFC6455 incl. any extensions or outright independent of WS)
>>
>> 3. does not put servers that provide both native WS and SPDY (but not an=
y
>> SPDY/WS) on same origin at a disadvantage
>>
>> 4. minimizes code coupling between WS and SPDY within implementations
>>
>>
>>
>> /Tobias
>>
>>
>>
>> *Von:* yhirano@google.com [mailto:yhirano@google.com] *Im Auftrag von *Y=
utaka
>> Hirano
>> *Gesendet:* Freitag, 31. Januar 2014 11:41
>> *An:* Tobias Oberstein
>> *Cc:* Roberto Peon; Peter Thorson; hybi@ietf.org
>>
>> *Betreff:* Re: [hybi] Discontinuation of mux standardizaton in favor of
>> WS/HTTP/2.0
>>
>>
>>
>> Seriously: _what_ is the goal of this effort?
>>
>> My motivation is, specifying a multiplexed bidirectional message protoco=
l
>> conforming to the W3C WebSocket API on HTTP/2.0.
>>
>>
>>
>>  - The W3C WebSocket API
>>
>>  - Multiplexing
>>
>> is mandatory for me.
>>
>> Since It is built on HTTP/2.0, I prefer strongly the following:
>>
>>  - A mechanism that enables intermediaries to transfer WS over HTTP/2.0
>> without fully understanding WS over HTTP/2.0.
>>
>>
>>
>> The remaining, though they are preferable, are not mandatory for me.
>>
>>  - RFC6455 conformance
>>
>>  - small amount of bandwidth efficiency
>>
>>  - the spec cleanness
>>
>>  - less-invasiveness
>>
>>
>>
>> Note: It does not mean I can't accept WS-TUNNEL.
>>
>>
>>
>> Tobias, Roberto, Peter, can you share your priority?
>>
>>
>>
>>
>>
>> On Thu, Jan 30, 2014 at 8:54 PM, Tobias Oberstein <
>> tobias.oberstein@tavendo.de> wrote:
>>
>> >I don't find the frame-size arguments very compelling since I haven't
>> seen or met anyone with a substantial deployment who cares significantly
>> about the overhead difference given packet, TLS, etc. overheads.
>> >There is no deployment that we know about that uses frame-based
>> extensibility- we can (and should) remove it from the standard since it
>> hasn't proven to be interesting.
>>
>> There is a draft for a WebSocket extension ("permessage-priority") being
>> currently developed that makes use of this and we will start an
>> experimental implementation in Autobahn. It _is_ interesting.
>>
>> In any case, the ability for framing, and the ability for WS extensions
>> to interleave WS messages is part of RFC6455.
>>
>> Why remove something from RFC6455 just to fit the limited capabilities
>> that SPDY offers?
>>
>>
>> >I'm not sure I agree with your statement that libraries supporting
>> HTTP/2 will include necessarily also have WS framing. It seems unlikely =
to
>> me-- I know of at least one large deployment where this isn't at all tru=
e.
>>
>> I'd put it like this: any networking library that wants to support
>> "messaging" on the Web will need to support native WebSocket, plus
>> optionally any SPDY-based messaging.
>>
>> Since, WebSocket is deployed _today_ and HTTP2 won't be everywhere
>> anytime soon.
>>
>> E.g. HTTP/2 support on Windows seems to be coming only on Windows 8+
>> (since they have HTTP in-kernel). So native WebSocket is the only messag=
ing
>> that works on any IE version on Windows <8.
>>
>>
>> >The HTTP/2 standard has (or shortly will have, as soon as the pull
>> request goes in) all support necessary to handle any stream-of-message
>> protocol given support of END_MESSAGE.
>> >I'd strongly prefer that we map WS to those primitives so it, and any
>> other stream-of-message protocols can be handled using loadbalancer/rout=
ing
>> infrastructure which is and can remain protocol agnostic.
>>
>> That's fine, let's just not call it WebSocket-over-SPDY then;)
>>
>> Seriously: _what_ is the goal of this effort?
>>
>> /Tobias
>>
>>
>>
>> -=3DR
>>
>> On Wed, Jan 29, 2014 at 5:56 AM, Tobias Oberstein <
>> tobias.oberstein@tavendo.de> wrote:
>> >The proposal is mapping one WS Message to one HEADERS + multiple DATAs.
>> >For large messages, it can achieve less overhead than WS-TUNNEL at the
>> cost of some RFC6455 compatibility.
>> How can it achieve less overhead?
>>
>> A SPDY HEADER with information from first WS frame seems to be always
>> bigger than a SPDY DATA header (8 octets) + WS frame header.
>>
>> The only situation I can think of: a large, _fragmented_ WS message is
>> reframed into a SPDY HEADER+DATAs sequence.
>>
>> Effectively, you do away with the individual WS frame information. And
>> this exactly breaks full RFC6455 compatibility.
>>
>> /Tobias
>> _______________________________________________
>> hybi mailing list
>> hybi@ietf.org
>> https://www.ietf.org/mailman/listinfo/hybi
>>
>>
>>
>
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi
>
>
>
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi
>
>

--001a11c16cccdcc50a04f186aba1
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><br><div class=3D"gmail_extra"><br><br><div class=3D"gmail=
_quote">On Mon, Feb 3, 2014 at 12:36 PM, Peter Thorson <span dir=3D"ltr">&l=
t;<a href=3D"mailto:webmaster@zaphoyd.com" target=3D"_blank">webmaster@zaph=
oyd.com</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div style=3D"word-wrap:break-word"><div>My =
priorities:</div><div><br></div><div>Important:</div><div>1. W3C WebSocket =
API and muxing go without saying.</div>
<div>2. Dedicated RFC6455 connections remain a viable option in modern brow=
sers.</div><div>3. Efficient and clearly defined method for detecting HTTP/=
2.0 messaging support, choosing HTTP/2.0 vs raw RFC6455, when to fall back,=
 etc.</div>
<div>4. Payload compression with appropriate configurability (either via pe=
rmessage-deflate support or a native mechanism)</div><div><br></div><div><d=
iv>Nice to haves:</div><div>5. Avoid significant regressions from RFC6455 i=
n wire overhead, latency, and per message fixed costs for very small (&lt;1=
00byte) messages.</div>
</div><div>6. Clean spec/avoiding redundant and unnecessary features of RFC=
6455.</div><div><br></div><div>Only if it doesn=E2=80=99t conflict with 1-6=
:</div><div>7. Options for future compatibility with simple RFC6455 extensi=
ons</div>
<div><br></div><div>As long as #2 and #3 are satisfied I am less concerned =
by performance details or RFC6455 compatibility.</div><div><br></div><div>Q=
uestion for Roberto:</div><div class=3D"im"><div><blockquote type=3D"cite">
<div dir=3D"ltr"><div style=3D"font-family:arial,sans-serif;font-size:13px"=
>I&#39;m not at all interested in needing to implement multiple separate pr=
otocols in order to implement the WS API.</div></div></blockquote></div><di=
v>
<br></div></div><div>Can you clarify under what context you are referring h=
ere? Is your meaning that Chrome.app should not have to implement both HTTP=
/2.0 and RFC6455? Or that one should be able to create an HTTP/2.0-only pro=
duct without needing to pull in an RFC6455 library? Something else?</div>
</div></blockquote><div><br></div><div>Yup. One shouldn&#39;t have to write=
 parsers for both 6455 and HTTP/2 if one is only interested in supporting W=
S via HTTP/2.</div><div><br></div><div>Separately, I wouldn&#39;t at all mi=
nd having an ALPN identifier for RFC6455 WS on port 443. I view that as a s=
ide-benefit of the HTTP/2 work, but not HTTP/2 itself.</div>
<div><br></div><div>-=3DR</div><div>=C2=A0<br></div><blockquote class=3D"gm=
ail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-le=
ft:1ex"><div style=3D"word-wrap:break-word"><div><br></div><div>Question fo=
r Yutaka:</div>
<div class=3D"im"><div><div class=3D"gmail_extra"><div class=3D"gmail_quote=
"><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;borde=
r-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid=
;padding-left:1ex">
<div lang=3D"DE" link=3D"blue" vlink=3D"purple"><div style=3D"border-style:=
none none none solid;border-left-color:blue;border-left-width:1.5pt;padding=
:0cm 0cm 0cm 4pt"><div><p class=3D"MsoNormal">- A mechanism that enables in=
termediaries to transfer WS over HTTP/2.0 without fully understanding WS ov=
er HTTP/2.0.</p>
</div></div></div></blockquote></div></div></div></div><div>Does the introd=
uction of a &quot;message end=E2=80=9D flag that HTTP/2.0 intermediaries wo=
uld have to understand as =E2=80=9Cdo not coalesce across=E2=80=9D violate =
this constraint?</div>
<div><div class=3D"h5"><br><div><div>On Feb 3, 2014, at 11:15 , Roberto Peo=
n &lt;<a href=3D"mailto:fenix@google.com" target=3D"_blank">fenix@google.co=
m</a>&gt; wrote:</div><br><blockquote type=3D"cite"><div dir=3D"ltr"><span =
style=3D"font-family:arial,sans-serif;font-size:13px">I&#39;m interested in=
 supporting the W3C WS API.</span><div style=3D"font-family:arial,sans-seri=
f;font-size:13px">
I&#39;m interested in correctly muxing, demuxing, and loadbalancing of stre=
ams-of-messages without needing to understand application-layer semantics.<=
/div>
<div style=3D"font-family:arial,sans-serif;font-size:13px">I&#39;m interest=
ed in enabling a single transport-layer session for both HTTP, WS, and pote=
ntially other content.</div><div style=3D"font-family:arial,sans-serif;font=
-size:13px">

<br></div><div style=3D"font-family:arial,sans-serif;font-size:13px">I&#39;=
m not at all interested in needing to implement multiple separate protocols=
 in order to implement the WS API.</div><div style=3D"font-family:arial,san=
s-serif;font-size:13px">

I&#39;m uninterested in small-number-of-byte savings.</div><div style=3D"fo=
nt-family:arial,sans-serif;font-size:13px">I am uninterested in any extensi=
ons or extension frameworks that require any extension-specific knowledge i=
n intermediaries.=C2=A0</div>

</div><div class=3D"gmail_extra"><br><br><div class=3D"gmail_quote">On Fri,=
 Jan 31, 2014 at 4:13 AM, Tobias Oberstein <span dir=3D"ltr">&lt;<a href=3D=
"mailto:tobias.oberstein@tavendo.de" target=3D"_blank">tobias.oberstein@tav=
endo.de</a>&gt;</span> wrote:<br>

<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div lang=3D"DE" link=3D"blue" vlink=3D"purp=
le"><p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quo=
t;Calibri&quot;,&quot;sans-serif&quot;;color:#1f497d">Hello Yutaka,<u></u><=
u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></spa=
n></p><p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&q=
uot;Calibri&quot;,&quot;sans-serif&quot;;color:#1f497d">my priorities:<u></=
u><u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></spa=
n></p><p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&q=
uot;Calibri&quot;,&quot;sans-serif&quot;;color:#1f497d">1. allows to provid=
e W3C WebSocket API<u></u><u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">2. no &quot;80%-WebSocket=
-protocol-compatible&quot; design (either compatible with full RFC6455 incl=
. any extensions or outright independent of WS)<u></u><u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">3. does not put servers t=
hat provide both native WS and SPDY (but not any SPDY/WS) on same origin at=
 a disadvantage<u></u><u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">4. minimizes code couplin=
g between WS and SPDY within implementations<u></u><u></u></span></p><p cla=
ss=3D"MsoNormal">

<span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-=
serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></span></p><p class=3D"MsoNo=
rmal"><span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot=
;sans-serif&quot;;color:#1f497d">/Tobias<u></u><u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></spa=
n></p><div style=3D"border-style:none none none solid;border-left-color:blu=
e;border-left-width:1.5pt;padding:0cm 0cm 0cm 4pt">

<div><div style=3D"border:none;border-top:solid #e1e1e1 1.0pt;padding:3.0pt=
 0cm 0cm 0cm"><p class=3D"MsoNormal"><b><span style=3D"font-size:11.0pt;fon=
t-family:&quot;Calibri&quot;,&quot;sans-serif&quot;">Von:</span></b><span s=
tyle=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif&q=
uot;"> <a href=3D"mailto:yhirano@google.com" target=3D"_blank">yhirano@goog=
le.com</a> [mailto:<a href=3D"mailto:yhirano@google.com" target=3D"_blank">=
yhirano@google.com</a>] <b>Im Auftrag von </b>Yutaka Hirano<br>

<b>Gesendet:</b> Freitag, 31. Januar 2014 11:41<br><b>An:</b> Tobias Oberst=
ein<br><b>Cc:</b> Roberto Peon; Peter Thorson; <a href=3D"mailto:hybi@ietf.=
org" target=3D"_blank">hybi@ietf.org</a></span></p><div><br><b>Betreff:</b>=
 Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0<u=
></u><u></u></div>
<div><br></div></div></div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p><=
div><blockquote style=3D"border:none;border-left:solid #cccccc 1.0pt;paddin=
g:0cm 0cm 0cm 6.0pt;margin-left:4.8pt;margin-right:0cm"><p class=3D"MsoNorm=
al"><span style=3D"font-size:10.0pt;font-family:&quot;Arial&quot;,&quot;san=
s-serif&quot;">Seriously: _what_ is the goal of this effort?</span><u></u><=
u></u></p>

</blockquote><div><div><div><p class=3D"MsoNormal">My motivation is, specif=
ying a multiplexed bidirectional message protocol conforming to the W3C Web=
Socket API on HTTP/2.0.<u></u><u></u></p></div><div><p class=3D"MsoNormal">

<u></u>=C2=A0<u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- The W3C W=
ebSocket API<u></u><u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- Mul=
tiplexing<u></u><u></u></p></div><div><p class=3D"MsoNormal">is mandatory f=
or me.<u></u><u></u></p>

</div><div><p class=3D"MsoNormal">Since It is built on HTTP/2.0, I prefer s=
trongly the following:<u></u><u></u></p></div><div><p class=3D"MsoNormal">=
=C2=A0- A mechanism that enables intermediaries to transfer WS over HTTP/2.=
0 without fully understanding WS over HTTP/2.0.<u></u><u></u></p>

</div><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div><div><p cla=
ss=3D"MsoNormal">The remaining, though they are preferable, are not mandato=
ry for me.<u></u><u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- RFC64=
55 conformance<u></u><u></u></p>

</div><div><p class=3D"MsoNormal">=C2=A0- small amount of bandwidth efficie=
ncy<u></u><u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- the spec cle=
anness<u></u><u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- less-inva=
siveness<u></u><u></u></p>

</div><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div><div><p cla=
ss=3D"MsoNormal">Note: It does not mean I can&#39;t accept WS-TUNNEL.<u></u=
><u></u></p></div><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div=
><div><p class=3D"MsoNormal">

Tobias, Roberto, Peter, can you share your priority?<u></u><u></u></p></div=
><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div></div></div></di=
v><div><div><p class=3D"MsoNormal" style=3D"margin-bottom:12.0pt">
<u></u>=C2=A0<u></u></p><div><p class=3D"MsoNormal">On Thu, Jan 30, 2014 at=
 8:54 PM, Tobias Oberstein &lt;<a href=3D"mailto:tobias.oberstein@tavendo.d=
e" target=3D"_blank">tobias.oberstein@tavendo.de</a>&gt; wrote:<u></u><u></=
u></p>
<blockquote style=3D"border:none;border-left:solid #cccccc 1.0pt;padding:0c=
m 0cm 0cm 6.0pt;margin-left:4.8pt;margin-right:0cm">
<div><p class=3D"MsoNormal" style=3D"margin-bottom:12.0pt">&gt;I don&#39;t =
find the frame-size arguments very compelling since I haven&#39;t seen or m=
et anyone with a substantial deployment who cares significantly about the o=
verhead difference given packet, TLS, etc. overheads.<br>

&gt;There is no deployment that we know about that uses frame-based extensi=
bility- we can (and should) remove it from the standard since it hasn&#39;t=
 proven to be interesting.<u></u><u></u></p></div><p class=3D"MsoNormal">

There is a draft for a WebSocket extension (&quot;permessage-priority&quot;=
) being currently developed that makes use of this and we will start an exp=
erimental implementation in Autobahn. It _is_ interesting.<br><br>In any ca=
se, the ability for framing, and the ability for WS extensions to interleav=
e WS messages is part of RFC6455.<br>

<br>Why remove something from RFC6455 just to fit the limited capabilities =
that SPDY offers?<u></u><u></u></p><div><p class=3D"MsoNormal" style=3D"mar=
gin-bottom:12.0pt"><br>&gt;I&#39;m not sure I agree with your statement tha=
t libraries supporting HTTP/2 will include necessarily also have WS framing=
. It seems unlikely to me-- I know of at least one large deployment where t=
his isn&#39;t at all true.<u></u><u></u></p>

</div><p class=3D"MsoNormal">I&#39;d put it like this: any networking libra=
ry that wants to support &quot;messaging&quot; on the Web will need to supp=
ort native WebSocket, plus optionally any SPDY-based messaging.<br><br>Sinc=
e, WebSocket is deployed _today_ and HTTP2 won&#39;t be everywhere anytime =
soon.<br>

<br>E.g. HTTP/2 support on Windows seems to be coming only on Windows 8+ (s=
ince they have HTTP in-kernel). So native WebSocket is the only messaging t=
hat works on any IE version on Windows &lt;8.<u></u><u></u></p><div><p clas=
s=3D"MsoNormal" style=3D"margin-bottom:12.0pt">

<br>&gt;The HTTP/2 standard has (or shortly will have, as soon as the pull =
request goes in) all support necessary to handle any stream-of-message prot=
ocol given support of END_MESSAGE.<br>&gt;I&#39;d strongly prefer that we m=
ap WS to those primitives so it, and any other stream-of-message protocols =
can be handled using loadbalancer/routing infrastructure which is and can r=
emain protocol agnostic.<u></u><u></u></p>

</div><p class=3D"MsoNormal">That&#39;s fine, let&#39;s just not call it We=
bSocket-over-SPDY then;)<br><br>Seriously: _what_ is the goal of this effor=
t?<br><span style=3D"color:#888888"><br><span>/Tobias</span></span><u></u><=
u></u></p>

<div><p class=3D"MsoNormal" style=3D"margin-bottom:12.0pt"><br><br>-=3DR<br=
><br>On Wed, Jan 29, 2014 at 5:56 AM, Tobias Oberstein &lt;<a href=3D"mailt=
o:tobias.oberstein@tavendo.de" target=3D"_blank">tobias.oberstein@tavendo.d=
e</a>&gt; wrote:<br>

&gt;The proposal is mapping one WS Message to one HEADERS + multiple DATAs.=
<br>&gt;For large messages, it can achieve less overhead than WS-TUNNEL at =
the cost of some RFC6455 compatibility.<br>How can it achieve less overhead=
?<br>

<br>A SPDY HEADER with information from first WS frame seems to be always b=
igger than a SPDY DATA header (8 octets) + WS frame header.<br><br>The only=
 situation I can think of: a large, _fragmented_ WS message is reframed int=
o a SPDY HEADER+DATAs sequence.<br>

<br>Effectively, you do away with the individual WS frame information. And =
this exactly breaks full RFC6455 compatibility.<br><br>/Tobias<br>_________=
______________________________________<br>hybi mailing list<br><a href=3D"m=
ailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><br>

<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><u></u><u></u></p></div></block=
quote></div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div></div>
</div></div></blockquote></div><br></div>
_______________________________________________<br>hybi mailing list<br><a =
href=3D"mailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><br><a hre=
f=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">https://=
www.ietf.org/mailman/listinfo/hybi</a><br>
</blockquote></div><br></div></div></div><br>______________________________=
_________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
<br></blockquote></div><br></div></div>

--001a11c16cccdcc50a04f186aba1--

From yhirano@google.com  Mon Feb  3 22:12:05 2014
Return-Path: <yhirano@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 447D21A0366 for <hybi@ietfa.amsl.com>; Mon,  3 Feb 2014 22:12:05 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.313
X-Spam-Level: 
X-Spam-Status: No, score=-1.313 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, J_CHICKENPOX_65=0.6, RP_MATCHES_RCVD=-0.535, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id k3st_8wsSymy for <hybi@ietfa.amsl.com>; Mon,  3 Feb 2014 22:12:02 -0800 (PST)
Received: from mail-yk0-x22f.google.com (mail-yk0-x22f.google.com [IPv6:2607:f8b0:4002:c07::22f]) by ietfa.amsl.com (Postfix) with ESMTP id 6A7CB1A036A for <hybi@ietf.org>; Mon,  3 Feb 2014 22:12:02 -0800 (PST)
Received: by mail-yk0-f175.google.com with SMTP id q200so45218760ykb.6 for <hybi@ietf.org>; Mon, 03 Feb 2014 22:12:02 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=VFUGczEoZ6hm74DdjFKODIPCEczWRtHV4FIs1zfGxSI=; b=CdpeU9p8cNgeoWnq0MqCNMxW8sVPgD7RG0Q5NKzzyvqPUJ/B4mTN/YniDLUDLdGQ0t 0IRiSEFElWeV25ZpE3f6M55YOIxExxVTOqlYMbOYF6GZghK+P9ls0fALyIiLDhiPvlDY nnRjoBqrxbmb/Cn6z8IgIGRZqu4uuU0gmOBTHbCB6WfF2K1DmIK2leACXsxiw1eS797e a/nUATWp4K0tjdPFE+RsEhPqA2m5Yf/EdCOFIcOPw2mf2HcLRxZzKem/ZWSNByK0reWl uS6H3f5iRW9sFX16QwEYOfCD+M21IW2TG+GG20y8bM7JsPHSZUIvft74YzKDE1bpaIzh Q7gQ==
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=chromium.org; s=google; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=VFUGczEoZ6hm74DdjFKODIPCEczWRtHV4FIs1zfGxSI=; b=gWDtPjSp5O69FpCy0qnxF874AY3W7fWk4+pI+ercm+SpOCciqRKOY6eN+Fp4LOSZNf 8T+/6fc7IGpfISmhy+kgl2BJFgHX9YLyjJymHYXZ90Be7n+uT2OMqmsFBLHQ1VG9eHyC Ghlg+8VQGexshi6MSJlBjj1/ywVkMKXy/3NE8=
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:sender:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=VFUGczEoZ6hm74DdjFKODIPCEczWRtHV4FIs1zfGxSI=; b=Zpjezuxz3VJTbMeJXaqGMP4ZaMw4ydvxG2nTKF2TVoZxtKKBvM2G3Ns7s6VE4JOwXe s8qjldMWvjFePV8z3KoU4j7xZFDwv7r6CSGe/H7NsaKsWAZTZqKwe6IqM2zmlyhF54M0 HULpTDKC15xP0KosA9UKR4/ju1Ql+bwdot3lfpLKi0ofVK3UJLGrZganA73MRDbPARG1 n7BchNl5JNUXH7ztIMAeHvibJyKcCKruOMOwAESsJya3Cnuv+C34ujbteEYoFXSiTYjd W0z4bWbFTh0/BtrOaniabq4jmKD0aqS1Vj7wqvGgvUQKHP7iZT4kUF7N2Bkl8mnKikd5 a/Og==
X-Gm-Message-State: ALoCoQkuHgiKUObOHZOB0yQgeIVofDn42x0Ny412v6588zKNvY9/utDZroaenE3qVuv/50RA80ui17Fv1kfp9rcIVgliPb/DxPU/Rwcmcb5LZu0O371EDWMpfdClFsJdWbh/SeyeJc94EkysaX7YfcGet2S3+6vgOqSwjJWlhx89o8oowqlUNLqw+wGZrn+0Wz9Ygs20ntxX
MIME-Version: 1.0
X-Received: by 10.236.121.162 with SMTP id r22mr9496711yhh.47.1391494321963; Mon, 03 Feb 2014 22:12:01 -0800 (PST)
Sender: yhirano@google.com
Received: by 10.170.135.22 with HTTP; Mon, 3 Feb 2014 22:12:01 -0800 (PST)
In-Reply-To: <3CF0431C-3AF2-4F6D-82C6-3680976C0FA7@zaphoyd.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <CAGzyod6_9H=a+uejDTEd_6HfbD2e2ynkA2frCnhk0TboricgfQ@mail.gmail.com> <3CF0431C-3AF2-4F6D-82C6-3680976C0FA7@zaphoyd.com>
Date: Tue, 4 Feb 2014 15:12:01 +0900
X-Google-Sender-Auth: G484yG2_NxnEK7GboU3jgb9ugEU
Message-ID: <CABihn6G2ViYrt-yTLcsZPbewREshVV8LStvvsB4Gi8H+_M=_ow@mail.gmail.com>
From: Yutaka Hirano <yhirano@chromium.org>
To: Peter Thorson <webmaster@zaphoyd.com>
Content-Type: multipart/alternative; boundary=20cf301af34fc49f2204f18e84a7
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 04 Feb 2014 06:12:05 -0000

--20cf301af34fc49f2204f18e84a7
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

>
> Question for Yutaka:
>
>> - A mechanism that enables intermediaries to transfer WS over HTTP/2.0
>> without fully understanding WS over HTTP/2.0.
>
> Does the introduction of a "message end=E2=80=9D flag that HTTP/2.0 inter=
mediaries
> would have to understand as =E2=80=9Cdo not coalesce across=E2=80=9D viol=
ate this
> constraint?

I think it does not.
I would like to introduce MSG_DONE in the HTTP/2.0 spec so that all
intermediaries conforming to the spec can transfer WS over HTTP/2.0
messages.
The point is they don't have to read the WebSocket over HTTP/2.0 spec if
they don't interested in it.

Thanks,



On Tue, Feb 4, 2014 at 5:36 AM, Peter Thorson <webmaster@zaphoyd.com> wrote=
:

> My priorities:
>
> Important:
> 1. W3C WebSocket API and muxing go without saying.
> 2. Dedicated RFC6455 connections remain a viable option in modern browser=
s.
> 3. Efficient and clearly defined method for detecting HTTP/2.0 messaging
> support, choosing HTTP/2.0 vs raw RFC6455, when to fall back, etc.
> 4. Payload compression with appropriate configurability (either via
> permessage-deflate support or a native mechanism)
>
> Nice to haves:
> 5. Avoid significant regressions from RFC6455 in wire overhead, latency,
> and per message fixed costs for very small (<100byte) messages.
> 6. Clean spec/avoiding redundant and unnecessary features of RFC6455.
>
> Only if it doesn=E2=80=99t conflict with 1-6:
> 7. Options for future compatibility with simple RFC6455 extensions
>
> As long as #2 and #3 are satisfied I am less concerned by performance
> details or RFC6455 compatibility.
>
> Question for Roberto:
>
> I'm not at all interested in needing to implement multiple separate
> protocols in order to implement the WS API.
>
>
> Can you clarify under what context you are referring here? Is your meanin=
g
> that Chrome.app should not have to implement both HTTP/2.0 and RFC6455? O=
r
> that one should be able to create an HTTP/2.0-only product without needin=
g
> to pull in an RFC6455 library? Something else?
>
> Question for Yutaka:
>
>> - A mechanism that enables intermediaries to transfer WS over HTTP/2.0
>> without fully understanding WS over HTTP/2.0.
>>
> Does the introduction of a "message end=E2=80=9D flag that HTTP/2.0 inter=
mediaries
> would have to understand as =E2=80=9Cdo not coalesce across=E2=80=9D viol=
ate this
> constraint?
>
> On Feb 3, 2014, at 11:15 , Roberto Peon <fenix@google.com> wrote:
>
> I'm interested in supporting the W3C WS API.
> I'm interested in correctly muxing, demuxing, and loadbalancing of
> streams-of-messages without needing to understand application-layer
> semantics.
> I'm interested in enabling a single transport-layer session for both HTTP=
,
> WS, and potentially other content.
>
> I'm not at all interested in needing to implement multiple separate
> protocols in order to implement the WS API.
> I'm uninterested in small-number-of-byte savings.
> I am uninterested in any extensions or extension frameworks that require
> any extension-specific knowledge in intermediaries.
>
>
> On Fri, Jan 31, 2014 at 4:13 AM, Tobias Oberstein <
> tobias.oberstein@tavendo.de> wrote:
>
>> Hello Yutaka,
>>
>>
>>
>> my priorities:
>>
>>
>>
>> 1. allows to provide W3C WebSocket API
>>
>> 2. no "80%-WebSocket-protocol-compatible" design (either compatible with
>> full RFC6455 incl. any extensions or outright independent of WS)
>>
>> 3. does not put servers that provide both native WS and SPDY (but not an=
y
>> SPDY/WS) on same origin at a disadvantage
>>
>> 4. minimizes code coupling between WS and SPDY within implementations
>>
>>
>>
>> /Tobias
>>
>>
>>
>> *Von:* yhirano@google.com [mailto:yhirano@google.com] *Im Auftrag von *Y=
utaka
>> Hirano
>> *Gesendet:* Freitag, 31. Januar 2014 11:41
>> *An:* Tobias Oberstein
>> *Cc:* Roberto Peon; Peter Thorson; hybi@ietf.org
>>
>> *Betreff:* Re: [hybi] Discontinuation of mux standardizaton in favor of
>> WS/HTTP/2.0
>>
>>
>>
>> Seriously: _what_ is the goal of this effort?
>>
>> My motivation is, specifying a multiplexed bidirectional message protoco=
l
>> conforming to the W3C WebSocket API on HTTP/2.0.
>>
>>
>>
>>  - The W3C WebSocket API
>>
>>  - Multiplexing
>>
>> is mandatory for me.
>>
>> Since It is built on HTTP/2.0, I prefer strongly the following:
>>
>>  - A mechanism that enables intermediaries to transfer WS over HTTP/2.0
>> without fully understanding WS over HTTP/2.0.
>>
>>
>>
>> The remaining, though they are preferable, are not mandatory for me.
>>
>>  - RFC6455 conformance
>>
>>  - small amount of bandwidth efficiency
>>
>>  - the spec cleanness
>>
>>  - less-invasiveness
>>
>>
>>
>> Note: It does not mean I can't accept WS-TUNNEL.
>>
>>
>>
>> Tobias, Roberto, Peter, can you share your priority?
>>
>>
>>
>>
>>
>> On Thu, Jan 30, 2014 at 8:54 PM, Tobias Oberstein <
>> tobias.oberstein@tavendo.de> wrote:
>>
>> >I don't find the frame-size arguments very compelling since I haven't
>> seen or met anyone with a substantial deployment who cares significantly
>> about the overhead difference given packet, TLS, etc. overheads.
>> >There is no deployment that we know about that uses frame-based
>> extensibility- we can (and should) remove it from the standard since it
>> hasn't proven to be interesting.
>>
>> There is a draft for a WebSocket extension ("permessage-priority") being
>> currently developed that makes use of this and we will start an
>> experimental implementation in Autobahn. It _is_ interesting.
>>
>> In any case, the ability for framing, and the ability for WS extensions
>> to interleave WS messages is part of RFC6455.
>>
>> Why remove something from RFC6455 just to fit the limited capabilities
>> that SPDY offers?
>>
>>
>> >I'm not sure I agree with your statement that libraries supporting
>> HTTP/2 will include necessarily also have WS framing. It seems unlikely =
to
>> me-- I know of at least one large deployment where this isn't at all tru=
e.
>>
>> I'd put it like this: any networking library that wants to support
>> "messaging" on the Web will need to support native WebSocket, plus
>> optionally any SPDY-based messaging.
>>
>> Since, WebSocket is deployed _today_ and HTTP2 won't be everywhere
>> anytime soon.
>>
>> E.g. HTTP/2 support on Windows seems to be coming only on Windows 8+
>> (since they have HTTP in-kernel). So native WebSocket is the only messag=
ing
>> that works on any IE version on Windows <8.
>>
>>
>> >The HTTP/2 standard has (or shortly will have, as soon as the pull
>> request goes in) all support necessary to handle any stream-of-message
>> protocol given support of END_MESSAGE.
>> >I'd strongly prefer that we map WS to those primitives so it, and any
>> other stream-of-message protocols can be handled using loadbalancer/rout=
ing
>> infrastructure which is and can remain protocol agnostic.
>>
>> That's fine, let's just not call it WebSocket-over-SPDY then;)
>>
>> Seriously: _what_ is the goal of this effort?
>>
>> /Tobias
>>
>>
>>
>> -=3DR
>>
>> On Wed, Jan 29, 2014 at 5:56 AM, Tobias Oberstein <
>> tobias.oberstein@tavendo.de> wrote:
>> >The proposal is mapping one WS Message to one HEADERS + multiple DATAs.
>> >For large messages, it can achieve less overhead than WS-TUNNEL at the
>> cost of some RFC6455 compatibility.
>> How can it achieve less overhead?
>>
>> A SPDY HEADER with information from first WS frame seems to be always
>> bigger than a SPDY DATA header (8 octets) + WS frame header.
>>
>> The only situation I can think of: a large, _fragmented_ WS message is
>> reframed into a SPDY HEADER+DATAs sequence.
>>
>> Effectively, you do away with the individual WS frame information. And
>> this exactly breaks full RFC6455 compatibility.
>>
>> /Tobias
>> _______________________________________________
>> hybi mailing list
>> hybi@ietf.org
>> https://www.ietf.org/mailman/listinfo/hybi
>>
>>
>>
>
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi
>
>
>

--20cf301af34fc49f2204f18e84a7
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px =
0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-l=
eft-style:solid;padding-left:1ex">Question for Yutaka:<br><blockquote class=
=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;bo=
rder-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">
- A mechanism that enables intermediaries to transfer WS over HTTP/2.0 with=
out fully understanding WS over HTTP/2.0.</blockquote>Does the introduction=
 of a &quot;message end=E2=80=9D flag that HTTP/2.0 intermediaries would ha=
ve to understand as =E2=80=9Cdo not coalesce across=E2=80=9D violate this c=
onstraint?</blockquote>
<div>I think it does not.</div><div>I would like to introduce MSG_DONE in t=
he HTTP/2.0 spec so that all intermediaries conforming to the spec can tran=
sfer WS over HTTP/2.0 messages.</div><div>The point is they don&#39;t have =
to read the WebSocket over HTTP/2.0 spec if they don&#39;t interested in it=
.</div>
<div><br></div><div>Thanks,</div><div><br></div><div class=3D"" style=3D"fo=
nt-family:arial,sans-serif;font-size:13px"></div></div><div class=3D"gmail_=
extra"><br><br><div class=3D"gmail_quote">On Tue, Feb 4, 2014 at 5:36 AM, P=
eter Thorson <span dir=3D"ltr">&lt;<a href=3D"mailto:webmaster@zaphoyd.com"=
 target=3D"_blank">webmaster@zaphoyd.com</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div style=3D"word-wrap:break-word"><div>My =
priorities:</div><div><br></div><div>Important:</div><div>1. W3C WebSocket =
API and muxing go without saying.</div>
<div>2. Dedicated RFC6455 connections remain a viable option in modern brow=
sers.</div><div>3. Efficient and clearly defined method for detecting HTTP/=
2.0 messaging support, choosing HTTP/2.0 vs raw RFC6455, when to fall back,=
 etc.</div>
<div>4. Payload compression with appropriate configurability (either via pe=
rmessage-deflate support or a native mechanism)</div><div><br></div><div><d=
iv>Nice to haves:</div><div>5. Avoid significant regressions from RFC6455 i=
n wire overhead, latency, and per message fixed costs for very small (&lt;1=
00byte) messages.</div>
</div><div>6. Clean spec/avoiding redundant and unnecessary features of RFC=
6455.</div><div><br></div><div>Only if it doesn=E2=80=99t conflict with 1-6=
:</div><div>7. Options for future compatibility with simple RFC6455 extensi=
ons</div>
<div><br></div><div>As long as #2 and #3 are satisfied I am less concerned =
by performance details or RFC6455 compatibility.</div><div><br></div><div>Q=
uestion for Roberto:</div><div class=3D"im"><div><blockquote type=3D"cite">
<div dir=3D"ltr"><div style=3D"font-family:arial,sans-serif;font-size:13px"=
>I&#39;m not at all interested in needing to implement multiple separate pr=
otocols in order to implement the WS API.</div></div></blockquote></div><di=
v>
<br></div></div><div>Can you clarify under what context you are referring h=
ere? Is your meaning that Chrome.app should not have to implement both HTTP=
/2.0 and RFC6455? Or that one should be able to create an HTTP/2.0-only pro=
duct without needing to pull in an RFC6455 library? Something else?</div>
<div><br></div><div>Question for Yutaka:</div><div class=3D"im"><div><div c=
lass=3D"gmail_extra"><div class=3D"gmail_quote"><blockquote class=3D"gmail_=
quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-=
color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">
<div lang=3D"DE" link=3D"blue" vlink=3D"purple"><div style=3D"border-style:=
none none none solid;border-left-color:blue;border-left-width:1.5pt;padding=
:0cm 0cm 0cm 4pt"><div><p class=3D"MsoNormal">- A mechanism that enables in=
termediaries to transfer WS over HTTP/2.0 without fully understanding WS ov=
er HTTP/2.0.</p>
</div></div></div></blockquote></div></div></div></div><div>Does the introd=
uction of a &quot;message end=E2=80=9D flag that HTTP/2.0 intermediaries wo=
uld have to understand as =E2=80=9Cdo not coalesce across=E2=80=9D violate =
this constraint?</div>
<div><div class=3D"h5"><br><div><div>On Feb 3, 2014, at 11:15 , Roberto Peo=
n &lt;<a href=3D"mailto:fenix@google.com" target=3D"_blank">fenix@google.co=
m</a>&gt; wrote:</div><br><blockquote type=3D"cite"><div dir=3D"ltr"><span =
style=3D"font-family:arial,sans-serif;font-size:13px">I&#39;m interested in=
 supporting the W3C WS API.</span><div style=3D"font-family:arial,sans-seri=
f;font-size:13px">
I&#39;m interested in correctly muxing, demuxing, and loadbalancing of stre=
ams-of-messages without needing to understand application-layer semantics.<=
/div>
<div style=3D"font-family:arial,sans-serif;font-size:13px">I&#39;m interest=
ed in enabling a single transport-layer session for both HTTP, WS, and pote=
ntially other content.</div><div style=3D"font-family:arial,sans-serif;font=
-size:13px">

<br></div><div style=3D"font-family:arial,sans-serif;font-size:13px">I&#39;=
m not at all interested in needing to implement multiple separate protocols=
 in order to implement the WS API.</div><div style=3D"font-family:arial,san=
s-serif;font-size:13px">

I&#39;m uninterested in small-number-of-byte savings.</div><div style=3D"fo=
nt-family:arial,sans-serif;font-size:13px">I am uninterested in any extensi=
ons or extension frameworks that require any extension-specific knowledge i=
n intermediaries.=C2=A0</div>

</div><div class=3D"gmail_extra"><br><br><div class=3D"gmail_quote">On Fri,=
 Jan 31, 2014 at 4:13 AM, Tobias Oberstein <span dir=3D"ltr">&lt;<a href=3D=
"mailto:tobias.oberstein@tavendo.de" target=3D"_blank">tobias.oberstein@tav=
endo.de</a>&gt;</span> wrote:<br>

<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div lang=3D"DE" link=3D"blue" vlink=3D"purp=
le"><p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quo=
t;Calibri&quot;,&quot;sans-serif&quot;;color:#1f497d">Hello Yutaka,<u></u><=
u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></spa=
n></p><p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&q=
uot;Calibri&quot;,&quot;sans-serif&quot;;color:#1f497d">my priorities:<u></=
u><u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></spa=
n></p><p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&q=
uot;Calibri&quot;,&quot;sans-serif&quot;;color:#1f497d">1. allows to provid=
e W3C WebSocket API<u></u><u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">2. no &quot;80%-WebSocket=
-protocol-compatible&quot; design (either compatible with full RFC6455 incl=
. any extensions or outright independent of WS)<u></u><u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">3. does not put servers t=
hat provide both native WS and SPDY (but not any SPDY/WS) on same origin at=
 a disadvantage<u></u><u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">4. minimizes code couplin=
g between WS and SPDY within implementations<u></u><u></u></span></p><p cla=
ss=3D"MsoNormal">

<span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-=
serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></span></p><p class=3D"MsoNo=
rmal"><span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot=
;sans-serif&quot;;color:#1f497d">/Tobias<u></u><u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></spa=
n></p><div style=3D"border-style:none none none solid;border-left-color:blu=
e;border-left-width:1.5pt;padding:0cm 0cm 0cm 4pt">

<div><div style=3D"border:none;border-top:solid #e1e1e1 1.0pt;padding:3.0pt=
 0cm 0cm 0cm"><p class=3D"MsoNormal"><b><span style=3D"font-size:11.0pt;fon=
t-family:&quot;Calibri&quot;,&quot;sans-serif&quot;">Von:</span></b><span s=
tyle=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif&q=
uot;"> <a href=3D"mailto:yhirano@google.com" target=3D"_blank">yhirano@goog=
le.com</a> [mailto:<a href=3D"mailto:yhirano@google.com" target=3D"_blank">=
yhirano@google.com</a>] <b>Im Auftrag von </b>Yutaka Hirano<br>

<b>Gesendet:</b> Freitag, 31. Januar 2014 11:41<br><b>An:</b> Tobias Oberst=
ein<br><b>Cc:</b> Roberto Peon; Peter Thorson; <a href=3D"mailto:hybi@ietf.=
org" target=3D"_blank">hybi@ietf.org</a></span></p><div><br><b>Betreff:</b>=
 Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0<u=
></u><u></u></div>
<div><br></div></div></div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p><=
div><blockquote style=3D"border:none;border-left:solid #cccccc 1.0pt;paddin=
g:0cm 0cm 0cm 6.0pt;margin-left:4.8pt;margin-right:0cm"><p class=3D"MsoNorm=
al"><span style=3D"font-size:10.0pt;font-family:&quot;Arial&quot;,&quot;san=
s-serif&quot;">Seriously: _what_ is the goal of this effort?</span><u></u><=
u></u></p>

</blockquote><div><div><div><p class=3D"MsoNormal">My motivation is, specif=
ying a multiplexed bidirectional message protocol conforming to the W3C Web=
Socket API on HTTP/2.0.<u></u><u></u></p></div><div><p class=3D"MsoNormal">

<u></u>=C2=A0<u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- The W3C W=
ebSocket API<u></u><u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- Mul=
tiplexing<u></u><u></u></p></div><div><p class=3D"MsoNormal">is mandatory f=
or me.<u></u><u></u></p>

</div><div><p class=3D"MsoNormal">Since It is built on HTTP/2.0, I prefer s=
trongly the following:<u></u><u></u></p></div><div><p class=3D"MsoNormal">=
=C2=A0- A mechanism that enables intermediaries to transfer WS over HTTP/2.=
0 without fully understanding WS over HTTP/2.0.<u></u><u></u></p>

</div><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div><div><p cla=
ss=3D"MsoNormal">The remaining, though they are preferable, are not mandato=
ry for me.<u></u><u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- RFC64=
55 conformance<u></u><u></u></p>

</div><div><p class=3D"MsoNormal">=C2=A0- small amount of bandwidth efficie=
ncy<u></u><u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- the spec cle=
anness<u></u><u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- less-inva=
siveness<u></u><u></u></p>

</div><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div><div><p cla=
ss=3D"MsoNormal">Note: It does not mean I can&#39;t accept WS-TUNNEL.<u></u=
><u></u></p></div><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div=
><div><p class=3D"MsoNormal">

Tobias, Roberto, Peter, can you share your priority?<u></u><u></u></p></div=
><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div></div></div></di=
v><div><div><p class=3D"MsoNormal" style=3D"margin-bottom:12.0pt">
<u></u>=C2=A0<u></u></p><div><p class=3D"MsoNormal">On Thu, Jan 30, 2014 at=
 8:54 PM, Tobias Oberstein &lt;<a href=3D"mailto:tobias.oberstein@tavendo.d=
e" target=3D"_blank">tobias.oberstein@tavendo.de</a>&gt; wrote:<u></u><u></=
u></p>
<blockquote style=3D"border:none;border-left:solid #cccccc 1.0pt;padding:0c=
m 0cm 0cm 6.0pt;margin-left:4.8pt;margin-right:0cm">
<div><p class=3D"MsoNormal" style=3D"margin-bottom:12.0pt">&gt;I don&#39;t =
find the frame-size arguments very compelling since I haven&#39;t seen or m=
et anyone with a substantial deployment who cares significantly about the o=
verhead difference given packet, TLS, etc. overheads.<br>

&gt;There is no deployment that we know about that uses frame-based extensi=
bility- we can (and should) remove it from the standard since it hasn&#39;t=
 proven to be interesting.<u></u><u></u></p></div><p class=3D"MsoNormal">

There is a draft for a WebSocket extension (&quot;permessage-priority&quot;=
) being currently developed that makes use of this and we will start an exp=
erimental implementation in Autobahn. It _is_ interesting.<br><br>In any ca=
se, the ability for framing, and the ability for WS extensions to interleav=
e WS messages is part of RFC6455.<br>

<br>Why remove something from RFC6455 just to fit the limited capabilities =
that SPDY offers?<u></u><u></u></p><div><p class=3D"MsoNormal" style=3D"mar=
gin-bottom:12.0pt"><br>&gt;I&#39;m not sure I agree with your statement tha=
t libraries supporting HTTP/2 will include necessarily also have WS framing=
. It seems unlikely to me-- I know of at least one large deployment where t=
his isn&#39;t at all true.<u></u><u></u></p>

</div><p class=3D"MsoNormal">I&#39;d put it like this: any networking libra=
ry that wants to support &quot;messaging&quot; on the Web will need to supp=
ort native WebSocket, plus optionally any SPDY-based messaging.<br><br>Sinc=
e, WebSocket is deployed _today_ and HTTP2 won&#39;t be everywhere anytime =
soon.<br>

<br>E.g. HTTP/2 support on Windows seems to be coming only on Windows 8+ (s=
ince they have HTTP in-kernel). So native WebSocket is the only messaging t=
hat works on any IE version on Windows &lt;8.<u></u><u></u></p><div><p clas=
s=3D"MsoNormal" style=3D"margin-bottom:12.0pt">

<br>&gt;The HTTP/2 standard has (or shortly will have, as soon as the pull =
request goes in) all support necessary to handle any stream-of-message prot=
ocol given support of END_MESSAGE.<br>&gt;I&#39;d strongly prefer that we m=
ap WS to those primitives so it, and any other stream-of-message protocols =
can be handled using loadbalancer/routing infrastructure which is and can r=
emain protocol agnostic.<u></u><u></u></p>

</div><p class=3D"MsoNormal">That&#39;s fine, let&#39;s just not call it We=
bSocket-over-SPDY then;)<br><br>Seriously: _what_ is the goal of this effor=
t?<br><span style=3D"color:#888888"><br><span>/Tobias</span></span><u></u><=
u></u></p>

<div><p class=3D"MsoNormal" style=3D"margin-bottom:12.0pt"><br><br>-=3DR<br=
><br>On Wed, Jan 29, 2014 at 5:56 AM, Tobias Oberstein &lt;<a href=3D"mailt=
o:tobias.oberstein@tavendo.de" target=3D"_blank">tobias.oberstein@tavendo.d=
e</a>&gt; wrote:<br>

&gt;The proposal is mapping one WS Message to one HEADERS + multiple DATAs.=
<br>&gt;For large messages, it can achieve less overhead than WS-TUNNEL at =
the cost of some RFC6455 compatibility.<br>How can it achieve less overhead=
?<br>

<br>A SPDY HEADER with information from first WS frame seems to be always b=
igger than a SPDY DATA header (8 octets) + WS frame header.<br><br>The only=
 situation I can think of: a large, _fragmented_ WS message is reframed int=
o a SPDY HEADER+DATAs sequence.<br>

<br>Effectively, you do away with the individual WS frame information. And =
this exactly breaks full RFC6455 compatibility.<br><br>/Tobias<br>_________=
______________________________________<br>hybi mailing list<br><a href=3D"m=
ailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><br>

<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><u></u><u></u></p></div></block=
quote></div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div></div>
</div></div></blockquote></div><br></div>
_______________________________________________<br>hybi mailing list<br><a =
href=3D"mailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><br><a hre=
f=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">https://=
www.ietf.org/mailman/listinfo/hybi</a><br>
</blockquote></div><br></div></div></div></blockquote></div><br></div>

--20cf301af34fc49f2204f18e84a7--

From yhirano@google.com  Mon Feb  3 23:16:48 2014
Return-Path: <yhirano@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 5E4891A0382 for <hybi@ietfa.amsl.com>; Mon,  3 Feb 2014 23:16:48 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.913
X-Spam-Level: 
X-Spam-Status: No, score=-1.913 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.535, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id xE2XCXu6Ngg5 for <hybi@ietfa.amsl.com>; Mon,  3 Feb 2014 23:16:44 -0800 (PST)
Received: from mail-yk0-x232.google.com (mail-yk0-x232.google.com [IPv6:2607:f8b0:4002:c07::232]) by ietfa.amsl.com (Postfix) with ESMTP id EF26C1A0381 for <hybi@ietf.org>; Mon,  3 Feb 2014 23:16:43 -0800 (PST)
Received: by mail-yk0-f178.google.com with SMTP id 79so45428937ykr.9 for <hybi@ietf.org>; Mon, 03 Feb 2014 23:16:43 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=+XO+WFE9le50AuqyfCdb/LR3mGLVnrkw94OtPBPM3IY=; b=jY8ly3TiPYE/2KXrJaEj59xpMjtyRZ+cxnrzOMLDZ0GenADooqATJ7Gnt2G1Wn+M6B wamAyrS74NrBZ79bgTBqkNYnsXTZGbyjv7hr5N5pMbIbBO2stNfOgqgdHY9QUigROpE6 p7l2PtQO3PfhXLgnWQC1tGoTPry2qHe8U5XfSPY6jCGu3imad7lIArScYrjxxHit8BKC OQueZbWY7BZZcsBF49/iyBUc+lVvv4Kfzj1RG6KAqSfDvWVPPuPGDqSW/wvvdmVHieEf IYJYWJJKYLbj6IdKraRh7Wexc4k4h6TMwNk7rC4A8Rmw6B2lk3RAijx6/4gRxbLC6iai sHFQ==
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=chromium.org; s=google; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=+XO+WFE9le50AuqyfCdb/LR3mGLVnrkw94OtPBPM3IY=; b=HWkzd4Y284mF7LG4vu565JNXV0l6w8DQZbqqHqIAjYS02mK4b5J+9uFM7t3uig29lz Nmkv5i0Ugr8CQayv0m0ZI3ahukZxm2/Hqcy8EVNyqvD+IhLgDQiOFBWhNKGiArKnXPIM bIre1e2r3EKWhwbyZorv94eENULswNQFTiBv0=
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:sender:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=+XO+WFE9le50AuqyfCdb/LR3mGLVnrkw94OtPBPM3IY=; b=XMpaZVr+hRij46+y7xj6HxLQtdHd8sr0qtYiB+BH6C1WPOzSYVGvDIqjXLgJyJiN7c dNxFLHoXsUC7ZxCPkOjP+EWYv2hoMxfl7KkLfqOegyBpvpmmeVUBVJvQ7R0/lt94jEOK EnoHlhLlQCAJdFTFQzPhMsO/6vRylkQwHRAfE+WmWaXXywYrrftC1eKhsqk/RBuigziN Ixl5CPry9nXqcjLYeKVVTNJ6NNdaElbzEn/JAqC7Vm3MYC2rqokzyG6UTPH1eFzx6fps Q0UDzFLEEE1xOG7GG8ASr3kAEynD4pwOmxoqa4G1fKYAwZhhw0vbECkSyjLbi9AwP6Qg 9juA==
X-Gm-Message-State: ALoCoQmB7gyufu9gA/2rOtgQ28r1r4/jLjImPGv7SL5SfCNptp+Qxlt454eL8PIDAbEAP8kd7XX+kdgCw11BrxikEyULqYQ7QonegKaoeEdjQKjzFmcv01iVzxRuoONL3LI7zj/qDwcT3twT06m+JXGZ59gtDIiQ3wOzLqR72+ov3xoPiCqySXMXYYNfWDIUxCtnrILs1RfZ
MIME-Version: 1.0
X-Received: by 10.236.134.139 with SMTP id s11mr36544yhi.133.1391498203573; Mon, 03 Feb 2014 23:16:43 -0800 (PST)
Sender: yhirano@google.com
Received: by 10.170.135.22 with HTTP; Mon, 3 Feb 2014 23:16:43 -0800 (PST)
In-Reply-To: <634914A010D0B943A035D226786325D4446C1E9261@EXVMBX020-12.exch020.serverdata.net>
References: <CABihn6EtrUQvCjRXnjmEYHq_3z4KrPzX5e_JeRev0mAX-Od+fQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446C1E9261@EXVMBX020-12.exch020.serverdata.net>
Date: Tue, 4 Feb 2014 16:16:43 +0900
X-Google-Sender-Auth: y2N8Dz5lPI4kxaB9wT084gywJJ8
Message-ID: <CABihn6HAiNeVNoSFsHqMo_pccxwbWs26kFFXysOGxT+yOCcrxg@mail.gmail.com>
From: Yutaka Hirano <yhirano@chromium.org>
To: Tobias Oberstein <tobias.oberstein@tavendo.de>
Content-Type: multipart/alternative; boundary=20cf303a2e852167a704f18f6c75
Cc: "hybi@ietf.org" <hybi@ietf.org>, agl@chromium.org
Subject: Re: [hybi] WebSocket over HTTP/2.0 protocol negotiation
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 04 Feb 2014 07:16:48 -0000

--20cf303a2e852167a704f18f6c75
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

Hi Tobias,

In other words: I think a server should actively announce support for any
> SPDY/WS, and only then
> should SPDY/WS be tried first when a JS opens a WebSocket connection to a=
n
> origin that has an
> existing SPDY connection and which already announced it's support and
> willingness to speak
> any SPDY/WS.

I see. I propose a new negotiation method here and I think it addresses
your concern.

*Negotiation Decision Process:*
1: If the client has an existing HTTP/2.0 session and it knows that the
server supports WebSocket over HTTP/2.0, create a WebSocket stream on the
session. Both of WebSocket (ws) and secure WebSocket (wss) could share the
same HTTP/2.0 session.
2: If the client is going to establish a secure WebSocket connection,
create a WebSocket connection over TLS.
3: If the client is going to establish a non-secure WebSocket connection,
try the WebSocket protocol negotiation.

*Negotiation Cases:*

*Create WebSocket Stream on Existing HTTP/2.0 Session*
The client SHOULD create a new HTTP/2.0 stream in the existing HTTP/2.0
session.

*Create WebSocket Connection over TLS*
HTTP/2.0 uses TLS Application Level Protocol Negotiation (ALPN) for the
protocol negotiation. The client sends the application protocols it wants
to use, "HTTP/1.1" for the native WebSocket over TLS, "HTTP/2.0[ws]" for
non secure WebSocket over HTTP/2.0, "HTTP/2.0[wss]" for secure WebSocket
over HTTP/2.0.
If the server selects "HTTP/1.1", the client SHOULD start the native
WebSocket opening handshake on TLS.
If the server selects "HTTP/2.0[ws]", the client SHOULD create an HTTP/2.0
session and start the WebSocket opening handshake on it with "ws" scheme.
If the server selects "HTTP/2.0[wss]", the client SHOULD create an HTTP/2.0
session and start the WebSocket opening handshake on it with "wss" scheme.

*The WebSocket Protocol Negotiation*
(same as the document)

*Supporting scheme negotiation using SETTINGS frame*
By default, a client can send HEADERS frames on schemes that server
supports based on the protocol server selected in ALPN. A server SHOULD
send a SETTINGS frame as soon as possible whether it supports other schemes
or not.
Here is a proposed ID and value which should be defined in the HTTP/2.0
specification. If the endpoint received a HEADERS on unsupporting schemes,
the endpoint MUST reset the stream.

Defined IDs: (same as the document)


No additional RTTs are introduced when the server supports the native
WebSocket and HTTP/2.0 but does not support WebSocket over HTTP/2.0.
We should introduce the following in the HTTP/2.0 specification.
 - Supporting scheme advertisement using SETTINGS frame
 - HTTP/2.0[xx] in the ALPN stands for the HTTP/2.0 protocol with scheme xx=
.

Thanks,



On Thu, Jan 30, 2014 at 9:24 PM, Tobias Oberstein <
tobias.oberstein@tavendo.de> wrote:

> Hi Yutaka,
>
>
>
> the point I was trying to raise is this:
>
>
>
> For a JavaScript in a browser that opens a "ws://=E2=80=A6" connection, t=
here
> should never be an additional roundtrip
>
> incurred by first probing for any SPDY/WS and then falling back to native
> WS, but the native WS should be
>
> tried first.
>
>
>
> In other words: the current, immediate connection establishment for nativ=
e
> WS should not be put
>
> at an disadvantage just because of introducing any SPDY/WS.
>
>
>
> A server should be able to provide both SPDY and native WebSocket on the
> _same_ origin without
>
> being required to implement SPDY/WS also, and without incurring an
> additional roundtrip for
>
> probing SPDY/WS while establishing a WS connection.
>
>
>
> In other words: I think a server should actively announce support for any
> SPDY/WS, and only then
>
> should SPDY/WS be tried first when a JS opens a WebSocket connection to a=
n
> origin that has an
>
> existing SPDY connection and which already announced it's support and
> willingness to speak
>
> any SPDY/WS.
>
>
>
> Probably I misunderstood the draft.
>
>
>
> Are browsers supposed to work like above when they implement any SPDY/WS?
>
>
>
> /Tobias
>
>
>
>
>
> *Von:* hybi [mailto:hybi-bounces@ietf.org] *Im Auftrag von *Yutaka Hirano
> *Gesendet:* Donnerstag, 30. Januar 2014 07:45
> *An:* hybi@ietf.org
> *Betreff:* [hybi] WebSocket over HTTP/2.0 protocol negotiation
>
>
>
> Hi,
>
>
>
> As I announced, I've updated the draft and requested for comments.
>
> Tobias gave a comment on protocol negotiation (thanks!) and I think it is
> good to discuss here, not at the draft comment fields.
>
>
>
> The draft says:
>
>    1. If the client has a HTTP/2.0 session established by previous
>    requests and it doesn=E2=80=99t know that the session does NOT support=
 WebSocket
>    layering over HTTP/2.0, create a WebSocket stream on the session. Both=
 of
>    WebSocket (ws) and secure WebSocket (wss) could share the same HTTP/2.=
0
>    session.
>    2. If the client is going to establish a secure WebSocket connection,
>    create a WebSocket connection over TLS.
>    3. If the client is going to establish a non-secure WebSocket
>    connection, try the WebSocket protocol negotiation.
>
> Tobias doesn't like the first step. He says:
>
> I think native WS should be tried first. Native WS has less overhead than
> any WS-over-SPDY layering, and hence should be preferred.
>
>
>
> I have some comments about it.
>
>  - WebSocket over HTTP/2.0 is multiplexed, so it is not so obvious which
> is preferable. I think it depends on the popularity of the services:
>
>    - WebSocket over HTTP/2.0
>
>    - Native WebSocket and HTTP/2.0
>
>    - Native WebSocket and WebSocket over HTTP/2.0
>
>    Note that servers which doesn't support HTTP/2.0 don't matter, because
> they don't meet the precondition of the first step.
>
>  -  There are two scheme negotiation hints described below.
>
>     If we choose Hint X, before receiving a SETTINGS frame from the
> server, the client doesn't know which service is preferable. It is rare,
> but trying connecting to the wrong service can occur.
>
>     If we choose Hint Y, whether the server supports WebSocket over
> HTTP/2.0 or not is detected by the client prior to HTTP/2.0 session
> establishment and hence the client always understands the server's
> preference.
>
>     So, the actual negotiation decision process depends on the hint we
> choose.
>
>
>
> I'm sorry that the draft is not clear at this point, but it will be when
> the hint is chosen.
>
>
>
> Thanks,
>
>
>

--20cf303a2e852167a704f18f6c75
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Hi Tobias,<div><br></div><blockquote class=3D"gmail_quote"=
 style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:=
rgb(204,204,204);border-left-style:solid;padding-left:1ex">In other words: =
I think a server should actively announce support for any SPDY/WS, and only=
 then<br>
should SPDY/WS be tried first when a JS opens a WebSocket connection to an =
origin that has an<br>existing SPDY connection and which already announced =
it&#39;s support and willingness to speak<br>any SPDY/WS.</blockquote><div>
I see. I propose a new negotiation method here and I think it addresses you=
r concern.</div><div><br></div><blockquote style=3D"margin:0 0 0 40px;borde=
r:none;padding:0px"><div><b>Negotiation Decision Process:</b></div><div><di=
v>
1: If the client has an existing HTTP/2.0 session and it knows that the ser=
ver supports WebSocket over HTTP/2.0, create a WebSocket stream on the sess=
ion. Both of WebSocket (ws) and secure WebSocket (wss) could share the same=
 HTTP/2.0 session.</div>
</div><div>2: If the client is going to establish a secure WebSocket connec=
tion, create a WebSocket connection over TLS.</div><div>3: If the client is=
 going to establish a non-secure WebSocket connection, try the WebSocket pr=
otocol negotiation.</div>
<div><br></div><div><b>Negotiation Cases:</b></div><div><br></div><div><b>C=
reate WebSocket Stream on Existing HTTP/2.0 Session</b></div>The client SHO=
ULD create a new HTTP/2.0 stream in the existing HTTP/2.0 session.<br><br>
</blockquote><blockquote style=3D"margin:0 0 0 40px;border:none;padding:0px=
"><b>Create WebSocket Connection over TLS</b><div>HTTP/2.0 uses TLS Applica=
tion Level Protocol Negotiation (ALPN) for the protocol negotiation. The cl=
ient sends the application protocols it wants to use, &quot;HTTP/1.1&quot; =
for the native WebSocket over TLS, &quot;HTTP/2.0[ws]&quot; for non secure =
WebSocket over HTTP/2.0, &quot;HTTP/2.0[wss]&quot; for secure WebSocket ove=
r HTTP/2.0.</div>
<div>If the server selects &quot;HTTP/1.1&quot;, the client SHOULD start th=
e native WebSocket opening handshake on TLS.</div><div><div>If the server s=
elects &quot;HTTP/2.0[ws]&quot;, the client SHOULD create an HTTP/2.0 sessi=
on and start the WebSocket opening handshake on it with &quot;ws&quot; sche=
me.</div>
</div><div><div>If the server selects &quot;HTTP/2.0[wss]&quot;, the client=
 SHOULD create an HTTP/2.0 session and start the WebSocket opening handshak=
e on it with &quot;wss&quot; scheme.</div></div><div><div><br></div></div>
<div><div><b>The WebSocket Protocol Negotiation</b></div></div><div><div>(s=
ame as the document)</div></div><div><div><br></div></div><div><b>Supportin=
g scheme negotiation using SETTINGS frame</b></div><div>By default, a clien=
t can send HEADERS frames on schemes that server supports based on the prot=
ocol server selected in ALPN. A server SHOULD send a SETTINGS frame as soon=
 as possible whether it supports other schemes or not.</div>
<div>Here is a proposed ID and value which should be defined in the HTTP/2.=
0 specification. If the endpoint received a HEADERS on unsupporting schemes=
, the endpoint MUST reset the stream.</div><div><br></div><div>Defined IDs:=
 (same as the document)</div>
</blockquote><div><div></div></div><div><br>No additional RTTs are introduc=
ed when the server supports the native WebSocket and HTTP/2.0 but does not =
support WebSocket over HTTP/2.0.</div><div>We should introduce the followin=
g in the HTTP/2.0 specification.</div>
<div>=C2=A0- Supporting scheme advertisement using SETTINGS frame</div><div=
>=C2=A0- HTTP/2.0[xx] in the ALPN stands for the HTTP/2.0 protocol with sch=
eme xx.</div><div><br>Thanks,<br><div><br></div></div></div><div class=3D"g=
mail_extra">
<br><br><div class=3D"gmail_quote">On Thu, Jan 30, 2014 at 9:24 PM, Tobias =
Oberstein <span dir=3D"ltr">&lt;<a href=3D"mailto:tobias.oberstein@tavendo.=
de" target=3D"_blank">tobias.oberstein@tavendo.de</a>&gt;</span> wrote:<br>=
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
<div lang=3D"DE" link=3D"#0563C1" vlink=3D"#954F72"><div><p class=3D"MsoNor=
mal"><span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;=
sans-serif&quot;;color:#1f497d">Hi Yutaka,<u></u><u></u></span></p><p class=
=3D"MsoNormal">
<span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-=
serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></span></p><p class=3D"MsoNo=
rmal"><span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot=
;sans-serif&quot;;color:#1f497d">the point I was trying to raise is this:<u=
></u><u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></spa=
n></p><p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&q=
uot;Calibri&quot;,&quot;sans-serif&quot;;color:#1f497d">For a JavaScript in=
 a browser that opens a &quot;ws://=E2=80=A6&quot; connection, there should=
 never be an additional roundtrip<u></u><u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">incurred by first probing=
 for any SPDY/WS and then falling back to native WS, but the native WS shou=
ld be<u></u><u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">tried first.<u></u><u></u=
></span></p><p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-fam=
ily:&quot;Calibri&quot;,&quot;sans-serif&quot;;color:#1f497d"><u></u>=C2=A0=
<u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">In other words: the curre=
nt, immediate connection establishment for native WS should not be put<u></=
u><u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">at an disadvantage just b=
ecause of introducing any SPDY/WS.<u></u><u></u></span></p><p class=3D"MsoN=
ormal">
<span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-=
serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></span></p><p class=3D"MsoNo=
rmal"><span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot=
;sans-serif&quot;;color:#1f497d">A server should be able to provide both SP=
DY and native WebSocket on the _same_ origin without<u></u><u></u></span></=
p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">being required to impleme=
nt SPDY/WS also, and without incurring an additional roundtrip for<u></u><u=
></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">probing SPDY/WS while est=
ablishing a WS connection.<u></u><u></u></span></p><p class=3D"MsoNormal"><=
span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-s=
erif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">In other words: I think a=
 server should actively announce support for any SPDY/WS, and only then<u><=
/u><u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">should SPDY/WS be tried f=
irst when a JS opens a WebSocket connection to an origin that has an<u></u>=
<u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">existing SPDY connection =
and which already announced it&#39;s support and willingness to speak<u></u=
><u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">any SPDY/WS.<u></u><u></u=
></span></p><p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-fam=
ily:&quot;Calibri&quot;,&quot;sans-serif&quot;;color:#1f497d"><u></u>=C2=A0=
<u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">Probably I misunderstood =
the draft.<u></u><u></u></span></p><p class=3D"MsoNormal"><span style=3D"fo=
nt-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;;color=
:#1f497d"><u></u>=C2=A0<u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">Are browsers supposed to =
work like above when they implement any SPDY/WS?<u></u><u></u></span></p><p=
 class=3D"MsoNormal">
<span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-=
serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></span></p><p class=3D"MsoNo=
rmal"><span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot=
;sans-serif&quot;;color:#1f497d">/Tobias<u></u><u></u></span></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></spa=
n></p><p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&q=
uot;Calibri&quot;,&quot;sans-serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u=
></span></p>
<div style=3D"border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm =
4.0pt"><div><div style=3D"border:none;border-top:solid #e1e1e1 1.0pt;paddin=
g:3.0pt 0cm 0cm 0cm"><p class=3D"MsoNormal"><b><span style=3D"font-size:11.=
0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;">Von:</span></b>=
<span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-=
serif&quot;"> hybi [mailto:<a href=3D"mailto:hybi-bounces@ietf.org" target=
=3D"_blank">hybi-bounces@ietf.org</a>] <b>Im Auftrag von </b>Yutaka Hirano<=
br>
<b>Gesendet:</b> Donnerstag, 30. Januar 2014 07:45<br><b>An:</b> <a href=3D=
"mailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><br><b>Betreff:</=
b> [hybi] WebSocket over HTTP/2.0 protocol negotiation<u></u><u></u></span>=
</p>
</div></div><div><div class=3D"h5"><p class=3D"MsoNormal"><u></u>=C2=A0<u><=
/u></p><div><p class=3D"MsoNormal">Hi,<u></u><u></u></p><div><p class=3D"Ms=
oNormal"><u></u>=C2=A0<u></u></p></div><div><p class=3D"MsoNormal">As I ann=
ounced, I&#39;ve updated the draft and requested for comments.<u></u><u></u=
></p>
</div><div><p class=3D"MsoNormal">Tobias gave a comment on protocol negotia=
tion (thanks!) and I think it is good to discuss here, not at the draft com=
ment fields.<u></u><u></u></p></div><div><p class=3D"MsoNormal"><u></u>=C2=
=A0<u></u></p>
</div><div><p class=3D"MsoNormal">The draft says:<u></u><u></u></p></div><o=
l start=3D"1" type=3D"1"><li class=3D"MsoNormal">If the client has a HTTP/2=
.0 session established by previous requests and it doesn=E2=80=99t know tha=
t the session does NOT support WebSocket layering over HTTP/2.0, create a W=
ebSocket stream on the session. Both of WebSocket (ws) and secure WebSocket=
 (wss) could share the same HTTP/2.0 session.<u></u><u></u></li>
<li class=3D"MsoNormal">If the client is going to establish a secure WebSoc=
ket connection, create a WebSocket connection over TLS.<u></u><u></u></li><=
li class=3D"MsoNormal">If the client is going to establish a non-secure Web=
Socket connection, try the WebSocket protocol negotiation.<u></u><u></u></l=
i>
</ol><p class=3D"MsoNormal">Tobias doesn&#39;t like the first step. He says=
:<u></u><u></u></p><blockquote style=3D"border:none;border-left:solid #cccc=
cc 1.0pt;padding:0cm 0cm 0cm 6.0pt;margin-left:4.8pt;margin-right:0cm"><p c=
lass=3D"MsoNormal">
I think native WS should be tried first. Native WS has less overhead than a=
ny WS-over-SPDY layering, and hence should be preferred.<u></u><u></u></p><=
/blockquote><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div><div>=
<p class=3D"MsoNormal">
I have some comments about it.<u></u><u></u></p></div><div><p class=3D"MsoN=
ormal">=C2=A0- WebSocket over HTTP/2.0 is multiplexed, so it is not so obvi=
ous which is preferable. I think it depends on the popularity of the servic=
es:<u></u><u></u></p>
</div><div><p class=3D"MsoNormal">=C2=A0 =C2=A0- WebSocket over HTTP/2.0<u>=
</u><u></u></p></div><div><p class=3D"MsoNormal">=C2=A0 =C2=A0- Native WebS=
ocket and HTTP/2.0<u></u><u></u></p></div><div><p class=3D"MsoNormal">=C2=
=A0 =C2=A0- Native WebSocket and WebSocket over HTTP/2.0<u></u><u></u></p>
</div><div><p class=3D"MsoNormal">=C2=A0 =C2=A0Note that servers which does=
n&#39;t support HTTP/2.0 don&#39;t matter, because they don&#39;t meet the =
precondition of the first step.<u></u><u></u></p></div><div><p class=3D"Mso=
Normal">=C2=A0- =C2=A0There are two scheme negotiation hints described belo=
w.<u></u><u></u></p>
</div><div><p class=3D"MsoNormal">=C2=A0 =C2=A0 If we choose Hint X, before=
 receiving a SETTINGS frame from the server, the client doesn&#39;t know wh=
ich service is preferable. It is rare, but trying connecting to the wrong s=
ervice can occur.<u></u><u></u></p>
</div><div><p class=3D"MsoNormal">=C2=A0 =C2=A0 If we choose Hint Y, whethe=
r the server supports WebSocket over HTTP/2.0 or not is detected by the cli=
ent prior to HTTP/2.0 session establishment and hence the client always und=
erstands the server&#39;s preference.<u></u><u></u></p>
</div><div><p class=3D"MsoNormal">=C2=A0 =C2=A0 So, the actual negotiation =
decision process depends on the hint we choose.<u></u><u></u></p></div><div=
><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div><div><p class=3D"MsoN=
ormal">I&#39;m sorry that the draft is not clear at this point, but it will=
 be when the hint is chosen.<u></u><u></u></p>
</div><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div><div><p cla=
ss=3D"MsoNormal">Thanks,<u></u><u></u></p></div><div><p class=3D"MsoNormal"=
><u></u>=C2=A0<u></u></p></div></div></div></div></div></div></div></blockq=
uote></div><br>
</div>

--20cf303a2e852167a704f18f6c75--

From tobias.oberstein@tavendo.de  Tue Feb  4 02:32:15 2014
Return-Path: <tobias.oberstein@tavendo.de>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 0A1691A03F0 for <hybi@ietfa.amsl.com>; Tue,  4 Feb 2014 02:32:15 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.899
X-Spam-Level: 
X-Spam-Status: No, score=-1.899 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_NONE=-0.0001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 8Z5VpKJFQAd8 for <hybi@ietfa.amsl.com>; Tue,  4 Feb 2014 02:32:08 -0800 (PST)
Received: from EXHUB020-5.exch020.serverdata.net (exhub020-5.exch020.serverdata.net [206.225.164.32]) by ietfa.amsl.com (Postfix) with ESMTP id B09B81A03ED for <hybi@ietf.org>; Tue,  4 Feb 2014 02:32:07 -0800 (PST)
Received: from EXVMBX020-12.exch020.serverdata.net ([169.254.3.11]) by EXHUB020-5.exch020.serverdata.net ([206.225.164.32]) with mapi; Tue, 4 Feb 2014 02:32:06 -0800
From: Tobias Oberstein <tobias.oberstein@tavendo.de>
To: Yutaka Hirano <yhirano@chromium.org>
Date: Tue, 4 Feb 2014 02:32:04 -0800
Thread-Topic: [hybi] WebSocket over HTTP/2.0 protocol negotiation
Thread-Index: Ac8heQ+FNwgYfYGdRtuskl0fWhB4KQAFHX2g
Message-ID: <634914A010D0B943A035D226786325D4446C2954EF@EXVMBX020-12.exch020.serverdata.net>
References: <CABihn6EtrUQvCjRXnjmEYHq_3z4KrPzX5e_JeRev0mAX-Od+fQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446C1E9261@EXVMBX020-12.exch020.serverdata.net> <CABihn6HAiNeVNoSFsHqMo_pccxwbWs26kFFXysOGxT+yOCcrxg@mail.gmail.com>
In-Reply-To: <CABihn6HAiNeVNoSFsHqMo_pccxwbWs26kFFXysOGxT+yOCcrxg@mail.gmail.com>
Accept-Language: de-DE, en-US
Content-Language: de-DE
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
acceptlanguage: de-DE, en-US
Content-Type: multipart/alternative; boundary="_000_634914A010D0B943A035D226786325D4446C2954EFEXVMBX02012ex_"
MIME-Version: 1.0
Cc: "hybi@ietf.org" <hybi@ietf.org>, "agl@chromium.org" <agl@chromium.org>
Subject: Re: [hybi] WebSocket over HTTP/2.0 protocol negotiation
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 04 Feb 2014 10:32:15 -0000

--_000_634914A010D0B943A035D226786325D4446C2954EFEXVMBX02012ex_
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: base64
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--_000_634914A010D0B943A035D226786325D4446C2954EFEXVMBX02012ex_
Content-Type: text/html; charset="utf-8"
Content-Transfer-Encoding: base64
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--_000_634914A010D0B943A035D226786325D4446C2954EFEXVMBX02012ex_--


From yhirano@google.com  Tue Feb  4 06:24:57 2014
Return-Path: <yhirano@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 857381A00E2 for <hybi@ietfa.amsl.com>; Tue,  4 Feb 2014 06:24:57 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.313
X-Spam-Level: 
X-Spam-Status: No, score=-1.313 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, J_CHICKENPOX_65=0.6, RP_MATCHES_RCVD=-0.535, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Kv1eCB6-V3VR for <hybi@ietfa.amsl.com>; Tue,  4 Feb 2014 06:24:54 -0800 (PST)
Received: from mail-yk0-x22a.google.com (mail-yk0-x22a.google.com [IPv6:2607:f8b0:4002:c07::22a]) by ietfa.amsl.com (Postfix) with ESMTP id 2E47A1A00FD for <hybi@ietf.org>; Tue,  4 Feb 2014 06:24:54 -0800 (PST)
Received: by mail-yk0-f170.google.com with SMTP id 9so47500817ykp.1 for <hybi@ietf.org>; Tue, 04 Feb 2014 06:24:53 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=cti0/Y14QoiPi4XreSddYOKqY6ESPCsGNoxmKJptRLs=; b=RdM2vv50zM9bTC71Bav/x5RGOd0ogvLsbtI75cgtRR9oo4l1RO+lLDQnKiXFx4vdbj p64hyVwtaqirYdYJLKjYhfAR6C6s5TFD49/dcm4FRLGKlzQTkaLc7wqAWXet44b0BfjP BKIRPcfinUI2qQrIuAatGwKDXPlvrhreZZNqTltWom+uy1VdzSV5VFDYBfayWXTYTRU1 UlLgrMtKUG0irXqpPk+hRBIfbAJDwuq5NCPvLfGYxSvw3HNtYYGOe0g5LD0u7KhCiqe5 VBVczPAIjnhlT6M5+mEg+Hto4+KfTQ8NxGpKjy4S9x7wOT1whpJ9/GFV0iI3ufh/oKnj 3h0Q==
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=chromium.org; s=google; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=cti0/Y14QoiPi4XreSddYOKqY6ESPCsGNoxmKJptRLs=; b=Rlj4M9RTMcMhbqenUO2OS+4xnXoa2anSmaRVXAektKMJGW2+sLB4nfjsvukjVwNEWC 5/FvVf/0dPdp+peuUkbBtr0cP3Dq33kPpzTcd5WTZFWg/1WA7zDta2HseKdZlr9+FdM+ a8FQeecL2jI6PpXuYvB5YBeJxhdSNjAcA14Yk=
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:sender:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=cti0/Y14QoiPi4XreSddYOKqY6ESPCsGNoxmKJptRLs=; b=Sf531H1xrriw+Wi8IirrpPv4jn3NQY7llyrOoPaKjcuTk8+YCtn23pkzTg0bFOp+g0 yxJp6T2gBbWKwjsnZmQBKmuARHYVEXOstGkWksqAXgWYuawPq1QsrlYjzRhfY1tjVktj OzcusZTaFQfkqoq+q5TFrGrC+BUvmmzTTBXpUT02+EunoeOnTl7zR0tpSzeJd4XsEE8T gsFBB0RxOd4QVeY6xMce/xd8Q8Md5CbNvvwfriRA5abcB4I8x5ItKUgl6bhv8wp8tIt5 kgf2GiqwwAapa5RwbYbecMl06q6SNBxpND3MCUimGvGP8LAtln7zvyLy0bVi6f42//hQ MvjQ==
X-Gm-Message-State: ALoCoQnWPpk/sgLqTj9yrnXizUVcOgRhdxxSNA5RYLAWTSiZEQGW8tp159/E5/NUioWEgz+/uPzp45rHyPa5KBsHYNkEDn8gLZZJZoEVmg5mhGYH/dcDRWKlz2thl7ZD4a/pLBE2VCHYzq6sBKXTg8PZYxNa6rzenLgTowDgVaPhdh26tPHSBcLVo1GS0odKNCMhG9UG7vhK
MIME-Version: 1.0
X-Received: by 10.236.131.163 with SMTP id m23mr11038096yhi.61.1391523893521;  Tue, 04 Feb 2014 06:24:53 -0800 (PST)
Sender: yhirano@google.com
Received: by 10.170.135.22 with HTTP; Tue, 4 Feb 2014 06:24:53 -0800 (PST)
In-Reply-To: <CAGzyod6_9H=a+uejDTEd_6HfbD2e2ynkA2frCnhk0TboricgfQ@mail.gmail.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <CABihn6GSrU8oJa4nbMJiDDo2Z+-c4fT69roVFxcHYM4XYFToRQ@mail.gmail.com> <F1F115D4-29EB-457B-B02C-71E46D3E1C81@zaphoyd.com> <CABihn6GXxUwHwLd+926VPEkZxHRZUUi-3dw0PoNaO6aK8X4sEQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446C1E8D2C@EXVMBX020-12.exch020.serverdata.net> <CAGzyod7pt_PzjexmnZj8dSMg3BooUmuMb6xB5P0sG1t6S5FXZw@mail.gmail.com> <634914A010D0B943A035D226786325D4446C1E925F@EXVMBX020-12.exch020.serverdata.net> <CABihn6GrDwu=74WEACTrd8RMwDe78G-Q3X+e_OUrYbPcFm8gZQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446C1E9735@EXVMBX020-12.exch020.serverdata.net> <CAGzyod6_9H=a+uejDTEd_6HfbD2e2ynkA2frCnhk0TboricgfQ@mail.gmail.com>
Date: Tue, 4 Feb 2014 23:24:53 +0900
X-Google-Sender-Auth: sH1_sR7ficirav-kiuhMC1JWK_E
Message-ID: <CABihn6Gc34oSR70kkYUxpxQOMb2ZepR9STkLoe-SrZPTP8voYg@mail.gmail.com>
From: Yutaka Hirano <yhirano@chromium.org>
To: Roberto Peon <fenix@google.com>
Content-Type: multipart/alternative; boundary=20cf301af53f5ebe7904f195671c
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 04 Feb 2014 14:24:57 -0000

--20cf301af53f5ebe7904f195671c
Content-Type: text/plain; charset=UTF-8

Hi Roberto,

> I'm interested in correctly muxing, demuxing, and loadbalancing of
> streams-of-messages without needing to understand application-layer
> semantics.

IIUC None of Plan A, B and C provide the ability to detect exact message
boundary without understanding WebSocket level semantics even if we give up
some extensions, because of control message injection.
Introducing START_MESSAGE flag in addition to MSG_DONE enables it with Plan
A or Plan C. Is it acceptable for you?



On Tue, Feb 4, 2014 at 2:15 AM, Roberto Peon <fenix@google.com> wrote:

> I'm interested in supporting the W3C WS API.
> I'm interested in correctly muxing, demuxing, and loadbalancing of
> streams-of-messages without needing to understand application-layer
> semantics.
> I'm interested in enabling a single transport-layer session for both HTTP,
> WS, and potentially other content.
>
> I'm not at all interested in needing to implement multiple separate
> protocols in order to implement the WS API.
> I'm uninterested in small-number-of-byte savings.
> I am uninterested in any extensions or extension frameworks that require
> any extension-specific knowledge in intermediaries.
>
>
> On Fri, Jan 31, 2014 at 4:13 AM, Tobias Oberstein <
> tobias.oberstein@tavendo.de> wrote:
>
>> Hello Yutaka,
>>
>>
>>
>> my priorities:
>>
>>
>>
>> 1. allows to provide W3C WebSocket API
>>
>> 2. no "80%-WebSocket-protocol-compatible" design (either compatible with
>> full RFC6455 incl. any extensions or outright independent of WS)
>>
>> 3. does not put servers that provide both native WS and SPDY (but not any
>> SPDY/WS) on same origin at a disadvantage
>>
>> 4. minimizes code coupling between WS and SPDY within implementations
>>
>>
>>
>> /Tobias
>>
>>
>>
>> *Von:* yhirano@google.com [mailto:yhirano@google.com] *Im Auftrag von *Yutaka
>> Hirano
>> *Gesendet:* Freitag, 31. Januar 2014 11:41
>> *An:* Tobias Oberstein
>> *Cc:* Roberto Peon; Peter Thorson; hybi@ietf.org
>>
>> *Betreff:* Re: [hybi] Discontinuation of mux standardizaton in favor of
>> WS/HTTP/2.0
>>
>>
>>
>> Seriously: _what_ is the goal of this effort?
>>
>> My motivation is, specifying a multiplexed bidirectional message protocol
>> conforming to the W3C WebSocket API on HTTP/2.0.
>>
>>
>>
>>  - The W3C WebSocket API
>>
>>  - Multiplexing
>>
>> is mandatory for me.
>>
>> Since It is built on HTTP/2.0, I prefer strongly the following:
>>
>>  - A mechanism that enables intermediaries to transfer WS over HTTP/2.0
>> without fully understanding WS over HTTP/2.0.
>>
>>
>>
>> The remaining, though they are preferable, are not mandatory for me.
>>
>>  - RFC6455 conformance
>>
>>  - small amount of bandwidth efficiency
>>
>>  - the spec cleanness
>>
>>  - less-invasiveness
>>
>>
>>
>> Note: It does not mean I can't accept WS-TUNNEL.
>>
>>
>>
>> Tobias, Roberto, Peter, can you share your priority?
>>
>>
>>
>>
>>
>> On Thu, Jan 30, 2014 at 8:54 PM, Tobias Oberstein <
>> tobias.oberstein@tavendo.de> wrote:
>>
>> >I don't find the frame-size arguments very compelling since I haven't
>> seen or met anyone with a substantial deployment who cares significantly
>> about the overhead difference given packet, TLS, etc. overheads.
>> >There is no deployment that we know about that uses frame-based
>> extensibility- we can (and should) remove it from the standard since it
>> hasn't proven to be interesting.
>>
>> There is a draft for a WebSocket extension ("permessage-priority") being
>> currently developed that makes use of this and we will start an
>> experimental implementation in Autobahn. It _is_ interesting.
>>
>> In any case, the ability for framing, and the ability for WS extensions
>> to interleave WS messages is part of RFC6455.
>>
>> Why remove something from RFC6455 just to fit the limited capabilities
>> that SPDY offers?
>>
>>
>> >I'm not sure I agree with your statement that libraries supporting
>> HTTP/2 will include necessarily also have WS framing. It seems unlikely to
>> me-- I know of at least one large deployment where this isn't at all true.
>>
>> I'd put it like this: any networking library that wants to support
>> "messaging" on the Web will need to support native WebSocket, plus
>> optionally any SPDY-based messaging.
>>
>> Since, WebSocket is deployed _today_ and HTTP2 won't be everywhere
>> anytime soon.
>>
>> E.g. HTTP/2 support on Windows seems to be coming only on Windows 8+
>> (since they have HTTP in-kernel). So native WebSocket is the only messaging
>> that works on any IE version on Windows <8.
>>
>>
>> >The HTTP/2 standard has (or shortly will have, as soon as the pull
>> request goes in) all support necessary to handle any stream-of-message
>> protocol given support of END_MESSAGE.
>> >I'd strongly prefer that we map WS to those primitives so it, and any
>> other stream-of-message protocols can be handled using loadbalancer/routing
>> infrastructure which is and can remain protocol agnostic.
>>
>> That's fine, let's just not call it WebSocket-over-SPDY then;)
>>
>> Seriously: _what_ is the goal of this effort?
>>
>> /Tobias
>>
>>
>>
>> -=R
>>
>> On Wed, Jan 29, 2014 at 5:56 AM, Tobias Oberstein <
>> tobias.oberstein@tavendo.de> wrote:
>> >The proposal is mapping one WS Message to one HEADERS + multiple DATAs.
>> >For large messages, it can achieve less overhead than WS-TUNNEL at the
>> cost of some RFC6455 compatibility.
>> How can it achieve less overhead?
>>
>> A SPDY HEADER with information from first WS frame seems to be always
>> bigger than a SPDY DATA header (8 octets) + WS frame header.
>>
>> The only situation I can think of: a large, _fragmented_ WS message is
>> reframed into a SPDY HEADER+DATAs sequence.
>>
>> Effectively, you do away with the individual WS frame information. And
>> this exactly breaks full RFC6455 compatibility.
>>
>> /Tobias
>> _______________________________________________
>> hybi mailing list
>> hybi@ietf.org
>> https://www.ietf.org/mailman/listinfo/hybi
>>
>>
>>
>
>

--20cf301af53f5ebe7904f195671c
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div>Hi Roberto,=C2=A0</div><blockquote class=3D"gmail_quo=
te" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-col=
or:rgb(204,204,204);border-left-style:solid;padding-left:1ex"><span style=
=3D"font-family:arial,sans-serif;font-size:13px">I&#39;m interested in corr=
ectly muxing, demuxing, and loadbalancing of streams-of-messages without ne=
eding to understand application-layer semantics.</span></blockquote>

<div>IIUC None of Plan A, B and C provide the ability to detect exact messa=
ge boundary without understanding WebSocket level semantics even if we give=
 up some extensions, because of control message injection.</div><div>Introd=
ucing START_MESSAGE flag in addition to MSG_DONE enables it with Plan A or =
Plan C. Is it acceptable for you?</div>
<div><br></div></div><div class=3D"gmail_extra"><br><br><div class=3D"gmail=
_quote">On Tue, Feb 4, 2014 at 2:15 AM, Roberto Peon <span dir=3D"ltr">&lt;=
<a href=3D"mailto:fenix@google.com" target=3D"_blank">fenix@google.com</a>&=
gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div dir=3D"ltr"><span style=3D"font-family:=
arial,sans-serif;font-size:13px">I&#39;m interested in supporting the W3C W=
S API.</span><div style=3D"font-family:arial,sans-serif;font-size:13px">
I&#39;m interested in correctly muxing, demuxing, and loadbalancing of stre=
ams-of-messages without needing to understand application-layer semantics.<=
/div>
<div style=3D"font-family:arial,sans-serif;font-size:13px">I&#39;m interest=
ed in enabling a single transport-layer session for both HTTP, WS, and pote=
ntially other content.</div><div style=3D"font-family:arial,sans-serif;font=
-size:13px">

<br></div><div style=3D"font-family:arial,sans-serif;font-size:13px">I&#39;=
m not at all interested in needing to implement multiple separate protocols=
 in order to implement the WS API.</div><div style=3D"font-family:arial,san=
s-serif;font-size:13px">

I&#39;m uninterested in small-number-of-byte savings.</div><div style=3D"fo=
nt-family:arial,sans-serif;font-size:13px">I am uninterested in any extensi=
ons or extension frameworks that require any extension-specific knowledge i=
n intermediaries.=C2=A0</div>

</div><div class=3D"HOEnZb"><div class=3D"h5"><div class=3D"gmail_extra"><b=
r><br><div class=3D"gmail_quote">On Fri, Jan 31, 2014 at 4:13 AM, Tobias Ob=
erstein <span dir=3D"ltr">&lt;<a href=3D"mailto:tobias.oberstein@tavendo.de=
" target=3D"_blank">tobias.oberstein@tavendo.de</a>&gt;</span> wrote:<br>

<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div lang=3D"DE" link=3D"blue" vlink=3D"purp=
le"><div><p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family=
:&quot;Calibri&quot;,&quot;sans-serif&quot;;color:#1f497d">Hello Yutaka,<u>=
</u><u></u></span></p>

<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></spa=
n></p><p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&q=
uot;Calibri&quot;,&quot;sans-serif&quot;;color:#1f497d">my priorities:<u></=
u><u></u></span></p>

<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></spa=
n></p><p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&q=
uot;Calibri&quot;,&quot;sans-serif&quot;;color:#1f497d">1. allows to provid=
e W3C WebSocket API<u></u><u></u></span></p>

<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">2. no &quot;80%-WebSocket=
-protocol-compatible&quot; design (either compatible with full RFC6455 incl=
. any extensions or outright independent of WS)<u></u><u></u></span></p>

<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">3. does not put servers t=
hat provide both native WS and SPDY (but not any SPDY/WS) on same origin at=
 a disadvantage<u></u><u></u></span></p>

<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">4. minimizes code couplin=
g between WS and SPDY within implementations<u></u><u></u></span></p><p cla=
ss=3D"MsoNormal">

<span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-=
serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></span></p><p class=3D"MsoNo=
rmal"><span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot=
;sans-serif&quot;;color:#1f497d">/Tobias<u></u><u></u></span></p>

<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></spa=
n></p><div style=3D"border:none;border-left:solid blue 1.5pt;padding:0cm 0c=
m 0cm 4.0pt">

<div><div style=3D"border:none;border-top:solid #e1e1e1 1.0pt;padding:3.0pt=
 0cm 0cm 0cm"><p class=3D"MsoNormal"><b><span style=3D"font-size:11.0pt;fon=
t-family:&quot;Calibri&quot;,&quot;sans-serif&quot;">Von:</span></b><span s=
tyle=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif&q=
uot;"> <a href=3D"mailto:yhirano@google.com" target=3D"_blank">yhirano@goog=
le.com</a> [mailto:<a href=3D"mailto:yhirano@google.com" target=3D"_blank">=
yhirano@google.com</a>] <b>Im Auftrag von </b>Yutaka Hirano<br>

<b>Gesendet:</b> Freitag, 31. Januar 2014 11:41<br><b>An:</b> Tobias Oberst=
ein<br><b>Cc:</b> Roberto Peon; Peter Thorson; <a href=3D"mailto:hybi@ietf.=
org" target=3D"_blank">hybi@ietf.org</a></span></p><div><br><b>Betreff:</b>=
 Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0<u=
></u><u></u></div>

<p></p></div></div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p><div><blo=
ckquote style=3D"border:none;border-left:solid #cccccc 1.0pt;padding:0cm 0c=
m 0cm 6.0pt;margin-left:4.8pt;margin-right:0cm"><p class=3D"MsoNormal"><spa=
n style=3D"font-size:10.0pt;font-family:&quot;Arial&quot;,&quot;sans-serif&=
quot;">Seriously: _what_ is the goal of this effort?</span><u></u><u></u></=
p>

</blockquote><div><div><div><p class=3D"MsoNormal">My motivation is, specif=
ying a multiplexed bidirectional message protocol conforming to the W3C Web=
Socket API on HTTP/2.0.<u></u><u></u></p></div><div><p class=3D"MsoNormal">

<u></u>=C2=A0<u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- The W3C W=
ebSocket API<u></u><u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- Mul=
tiplexing<u></u><u></u></p></div><div><p class=3D"MsoNormal">is mandatory f=
or me.<u></u><u></u></p>

</div><div><p class=3D"MsoNormal">Since It is built on HTTP/2.0, I prefer s=
trongly the following:<u></u><u></u></p></div><div><p class=3D"MsoNormal">=
=C2=A0- A mechanism that enables intermediaries to transfer WS over HTTP/2.=
0 without fully understanding WS over HTTP/2.0.<u></u><u></u></p>

</div><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div><div><p cla=
ss=3D"MsoNormal">The remaining, though they are preferable, are not mandato=
ry for me.<u></u><u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- RFC64=
55 conformance<u></u><u></u></p>

</div><div><p class=3D"MsoNormal">=C2=A0- small amount of bandwidth efficie=
ncy<u></u><u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- the spec cle=
anness<u></u><u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- less-inva=
siveness<u></u><u></u></p>

</div><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div><div><p cla=
ss=3D"MsoNormal">Note: It does not mean I can&#39;t accept WS-TUNNEL.<u></u=
><u></u></p></div><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div=
><div><p class=3D"MsoNormal">

Tobias, Roberto, Peter, can you share your priority?<u></u><u></u></p></div=
><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div></div></div></di=
v><div><div><div><p class=3D"MsoNormal" style=3D"margin-bottom:12.0pt">
<u></u>=C2=A0<u></u></p><div><p class=3D"MsoNormal">On Thu, Jan 30, 2014 at=
 8:54 PM, Tobias Oberstein &lt;<a href=3D"mailto:tobias.oberstein@tavendo.d=
e" target=3D"_blank">tobias.oberstein@tavendo.de</a>&gt; wrote:<u></u><u></=
u></p>
<blockquote style=3D"border:none;border-left:solid #cccccc 1.0pt;padding:0c=
m 0cm 0cm 6.0pt;margin-left:4.8pt;margin-right:0cm">
<div><p class=3D"MsoNormal" style=3D"margin-bottom:12.0pt">&gt;I don&#39;t =
find the frame-size arguments very compelling since I haven&#39;t seen or m=
et anyone with a substantial deployment who cares significantly about the o=
verhead difference given packet, TLS, etc. overheads.<br>

&gt;There is no deployment that we know about that uses frame-based extensi=
bility- we can (and should) remove it from the standard since it hasn&#39;t=
 proven to be interesting.<u></u><u></u></p></div><p class=3D"MsoNormal">

There is a draft for a WebSocket extension (&quot;permessage-priority&quot;=
) being currently developed that makes use of this and we will start an exp=
erimental implementation in Autobahn. It _is_ interesting.<br><br>In any ca=
se, the ability for framing, and the ability for WS extensions to interleav=
e WS messages is part of RFC6455.<br>

<br>Why remove something from RFC6455 just to fit the limited capabilities =
that SPDY offers?<u></u><u></u></p><div><p class=3D"MsoNormal" style=3D"mar=
gin-bottom:12.0pt"><br>&gt;I&#39;m not sure I agree with your statement tha=
t libraries supporting HTTP/2 will include necessarily also have WS framing=
. It seems unlikely to me-- I know of at least one large deployment where t=
his isn&#39;t at all true.<u></u><u></u></p>

</div><p class=3D"MsoNormal">I&#39;d put it like this: any networking libra=
ry that wants to support &quot;messaging&quot; on the Web will need to supp=
ort native WebSocket, plus optionally any SPDY-based messaging.<br><br>Sinc=
e, WebSocket is deployed _today_ and HTTP2 won&#39;t be everywhere anytime =
soon.<br>

<br>E.g. HTTP/2 support on Windows seems to be coming only on Windows 8+ (s=
ince they have HTTP in-kernel). So native WebSocket is the only messaging t=
hat works on any IE version on Windows &lt;8.<u></u><u></u></p><div><p clas=
s=3D"MsoNormal" style=3D"margin-bottom:12.0pt">

<br>&gt;The HTTP/2 standard has (or shortly will have, as soon as the pull =
request goes in) all support necessary to handle any stream-of-message prot=
ocol given support of END_MESSAGE.<br>&gt;I&#39;d strongly prefer that we m=
ap WS to those primitives so it, and any other stream-of-message protocols =
can be handled using loadbalancer/routing infrastructure which is and can r=
emain protocol agnostic.<u></u><u></u></p>

</div><p class=3D"MsoNormal">That&#39;s fine, let&#39;s just not call it We=
bSocket-over-SPDY then;)<br><br>Seriously: _what_ is the goal of this effor=
t?<br><span style=3D"color:#888888"><br><span>/Tobias</span></span><u></u><=
u></u></p>

<div><div><p class=3D"MsoNormal" style=3D"margin-bottom:12.0pt"><br><br>-=
=3DR<br><br>On Wed, Jan 29, 2014 at 5:56 AM, Tobias Oberstein &lt;<a href=
=3D"mailto:tobias.oberstein@tavendo.de" target=3D"_blank">tobias.oberstein@=
tavendo.de</a>&gt; wrote:<br>

&gt;The proposal is mapping one WS Message to one HEADERS + multiple DATAs.=
<br>&gt;For large messages, it can achieve less overhead than WS-TUNNEL at =
the cost of some RFC6455 compatibility.<br>How can it achieve less overhead=
?<br>

<br>A SPDY HEADER with information from first WS frame seems to be always b=
igger than a SPDY DATA header (8 octets) + WS frame header.<br><br>The only=
 situation I can think of: a large, _fragmented_ WS message is reframed int=
o a SPDY HEADER+DATAs sequence.<br>

<br>Effectively, you do away with the individual WS frame information. And =
this exactly breaks full RFC6455 compatibility.<br><br>/Tobias<br>_________=
______________________________________<br>hybi mailing list<br><a href=3D"m=
ailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><br>

<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><u></u><u></u></p></div></div><=
/blockquote></div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div></di=
v></div>

</div></div></div></blockquote></div><br></div>
</div></div></blockquote></div><br></div>

--20cf301af53f5ebe7904f195671c--

From tobias.oberstein@tavendo.de  Tue Feb  4 07:02:19 2014
Return-Path: <tobias.oberstein@tavendo.de>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 50B431A011A for <hybi@ietfa.amsl.com>; Tue,  4 Feb 2014 07:02:19 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.3
X-Spam-Level: 
X-Spam-Status: No, score=-1.3 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, J_CHICKENPOX_65=0.6, RCVD_IN_DNSWL_NONE=-0.0001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id CO58LOKCAaJK for <hybi@ietfa.amsl.com>; Tue,  4 Feb 2014 07:02:17 -0800 (PST)
Received: from EXHUB020-2.exch020.serverdata.net (exhub020-2.exch020.serverdata.net [206.225.164.29]) by ietfa.amsl.com (Postfix) with ESMTP id 606281A0124 for <hybi@ietf.org>; Tue,  4 Feb 2014 07:02:17 -0800 (PST)
Received: from EXVMBX020-12.exch020.serverdata.net ([169.254.3.11]) by EXHUB020-2.exch020.serverdata.net ([206.225.164.29]) with mapi; Tue, 4 Feb 2014 07:02:16 -0800
From: Tobias Oberstein <tobias.oberstein@tavendo.de>
To: Roberto Peon <fenix@google.com>
Date: Tue, 4 Feb 2014 07:02:13 -0800
Thread-Topic: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
Thread-Index: Ac8hBIs9kI8Dqb8gSk+trRmWF1zVKAAs9NRg
Message-ID: <634914A010D0B943A035D226786325D4446C2955D9@EXVMBX020-12.exch020.serverdata.net>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <CABihn6GSrU8oJa4nbMJiDDo2Z+-c4fT69roVFxcHYM4XYFToRQ@mail.gmail.com> <F1F115D4-29EB-457B-B02C-71E46D3E1C81@zaphoyd.com> <CABihn6GXxUwHwLd+926VPEkZxHRZUUi-3dw0PoNaO6aK8X4sEQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446C1E8D2C@EXVMBX020-12.exch020.serverdata.net> <CAGzyod7pt_PzjexmnZj8dSMg3BooUmuMb6xB5P0sG1t6S5FXZw@mail.gmail.com> <634914A010D0B943A035D226786325D4446C1E925F@EXVMBX020-12.exch020.serverdata.net> <CABihn6GrDwu=74WEACTrd8RMwDe78G-Q3X+e_OUrYbPcFm8gZQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446C1E9735@EXVMBX020-12.exch020.serverdata.net> <CAGzyod6_9H=a+uejDTEd_6HfbD2e2ynkA2frCnhk0TboricgfQ@mail.gmail.com>
In-Reply-To: <CAGzyod6_9H=a+uejDTEd_6HfbD2e2ynkA2frCnhk0TboricgfQ@mail.gmail.com>
Accept-Language: de-DE, en-US
Content-Language: de-DE
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
acceptlanguage: de-DE, en-US
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: base64
MIME-Version: 1.0
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 04 Feb 2014 15:02:19 -0000
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From webmaster@zaphoyd.com  Tue Feb  4 09:08:32 2014
Return-Path: <webmaster@zaphoyd.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 36BE41A01CB for <hybi@ietfa.amsl.com>; Tue,  4 Feb 2014 09:08:32 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 0.801
X-Spam-Level: 
X-Spam-Status: No, score=0.801 tagged_above=-999 required=5 tests=[BAYES_50=0.8, HTML_MESSAGE=0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id o2cM4vgRt6jf for <hybi@ietfa.amsl.com>; Tue,  4 Feb 2014 09:08:27 -0800 (PST)
Received: from leaf103.mfilter.dimenoc.com (tree.mfilter.dimenoc.com [72.29.89.3]) by ietfa.amsl.com (Postfix) with ESMTP id 99A621A016A for <hybi@ietf.org>; Tue,  4 Feb 2014 09:08:27 -0800 (PST)
Received: from localhost (localhost [127.0.0.1]) by leaf103.mfilter.dimenoc.com (Postfix) with ESMTP id B42341CC760 for <hybi@ietf.org>; Tue,  4 Feb 2014 12:08:26 -0500 (EST)
X-Virus-Scanned: amavisd-new at leaf.mfilter.dimenoc.com
X-Amavis-Alert: BAD HEADER SECTION, Header line longer than 998 characters: References: <CA[...]
Received: from leaf103.mfilter.dimenoc.com ([IPv6:::ffff:127.0.0.1]) by localhost (leaf103.mfilter.dimenoc.com [IPv6:::ffff:127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 3sV6lTobc48p for <hybi@ietf.org>; Tue,  4 Feb 2014 12:08:25 -0500 (EST)
Received: from sh78.surpasshosting.com (sh78.surpasshosting.com [72.29.64.142]) by leaf103.mfilter.dimenoc.com (Postfix) with ESMTP for <hybi@ietf.org>; Tue,  4 Feb 2014 12:08:25 -0500 (EST)
Received: from ranna.uchicago.edu ([128.135.45.206]:59538) by sh78.surpasshosting.com with esmtpsa (TLSv1:AES128-SHA:128) (Exim 4.80.1) (envelope-from <webmaster@zaphoyd.com>) id 1WAjU0-0001SN-1C; Tue, 04 Feb 2014 12:08:24 -0500
Content-Type: multipart/alternative; boundary="Apple-Mail=_61703D8B-882B-4661-A5E1-C9F9F69B22BD"
Mime-Version: 1.0 (Mac OS X Mail 7.1 \(1827\))
From: Peter Thorson <webmaster@zaphoyd.com>
In-Reply-To: <CABihn6Gc34oSR70kkYUxpxQOMb2ZepR9STkLoe-SrZPTP8voYg@mail.gmail.com>
Date: Tue, 4 Feb 2014 11:08:17 -0600
Message-Id: <3274C7C2-41D4-49AD-971E-DA3D85898BC2@zaphoyd.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <CABihn6GSr...
To: Yutaka Hirano <yhirano@chromium.org>
X-Mailer: Apple Mail (2.1827)
X-AntiAbuse: This header was added to track abuse, please include it with any abuse report
X-AntiAbuse: Primary Hostname - sh78.surpasshosting.com
X-AntiAbuse: Original Domain - ietf.org
X-AntiAbuse: Originator/Caller UID/GID - [47 12] / [47 12]
X-AntiAbuse: Sender Address Domain - zaphoyd.com
X-Get-Message-Sender-Via: sh78.surpasshosting.com: authenticated_id: webmaster@zaphoyd.com
X-Source: 
X-Source-Args: 
X-Source-Dir: 
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 04 Feb 2014 17:08:32 -0000

--Apple-Mail=_61703D8B-882B-4661-A5E1-C9F9F69B22BD
Content-Transfer-Encoding: quoted-printable
Content-Type: text/plain;
	charset=windows-1252


On Feb 4, 2014, at 08:24 , Yutaka Hirano <yhirano@chromium.org> wrote:

> Hi Roberto,=20
> I'm interested in correctly muxing, demuxing, and loadbalancing of =
streams-of-messages without needing to understand application-layer =
semantics.
> IIUC None of Plan A, B and C provide the ability to detect exact =
message boundary without understanding WebSocket level semantics even if =
we give up some extensions, because of control message injection.
> Introducing START_MESSAGE flag in addition to MSG_DONE enables it with =
Plan A or Plan C. Is it acceptable for you?

Can we clarify again what plan A, B, and C are?

I don=92t see any issue with message boundary detection using only =
MSG_DONE in either a native HTTP/2.0 message plan or a purely tunneled =
plan.

Under tunneled plan the only information that is necessary for HTTP/2.0 =
intermediaries to understand in order to mux/demux/load balance/etc is =
the stream end bit (unless you mean to balance messages within a stream =
to different backends). All information about message boundaries and =
control messages is contained in the stream data itself and the RFC6455 =
parser eventually reading the stream will be able to pick all that up =
without any additional help.

Under the native HTTP/2.0 message plan there is no need for control =
messages at all except implicitly a =93close=94 frame (i.e. a frame with =
stream end set) which as the last frame in the stream should not =
conflict with message boundary detection.=

--Apple-Mail=_61703D8B-882B-4661-A5E1-C9F9F69B22BD
Content-Transfer-Encoding: quoted-printable
Content-Type: text/html;
	charset=windows-1252

<html><head><meta http-equiv=3D"Content-Type" content=3D"text/html =
charset=3Dwindows-1252"></head><body style=3D"word-wrap: break-word; =
-webkit-nbsp-mode: space; -webkit-line-break: =
after-white-space;"><br><div><div>On Feb 4, 2014, at 08:24 , Yutaka =
Hirano &lt;<a =
href=3D"mailto:yhirano@chromium.org">yhirano@chromium.org</a>&gt; =
wrote:</div><br class=3D"Apple-interchange-newline"><blockquote =
type=3D"cite"><div dir=3D"ltr"><div>Hi Roberto,&nbsp;</div><blockquote =
class=3D"gmail_quote" style=3D"margin:0px 0px 0px =
0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left=
-style:solid;padding-left:1ex"><span =
style=3D"font-family:arial,sans-serif;font-size:13px">I'm interested in =
correctly muxing, demuxing, and loadbalancing of streams-of-messages =
without needing to understand application-layer =
semantics.</span></blockquote>

<div>IIUC None of Plan A, B and C provide the ability to detect exact =
message boundary without understanding WebSocket level semantics even if =
we give up some extensions, because of control message =
injection.</div><div>Introducing START_MESSAGE flag in addition to =
MSG_DONE enables it with Plan A or Plan C. Is it acceptable for =
you?</div></div></blockquote><br></div><div>Can we clarify again what =
plan A, B, and C are?</div><div><br></div><div>I don=92t see any issue =
with message boundary detection using only MSG_DONE in either a native =
HTTP/2.0 message plan or a purely tunneled =
plan.</div><div><br></div><div>Under tunneled plan the only information =
that is necessary for HTTP/2.0 intermediaries to understand in order to =
mux/demux/load balance/etc is the stream end bit (unless you mean to =
balance messages within a stream to different backends). All information =
about message boundaries and control messages is contained in the stream =
data itself and the RFC6455 parser eventually reading the stream will be =
able to pick all that up without any additional =
help.</div><div><br></div><div>Under the native HTTP/2.0 message plan =
there is no need for control messages at all except implicitly a =93close=94=
 frame (i.e. a frame with stream end set) which as the last frame in the =
stream should not conflict with message boundary =
detection.</div></body></html>=

--Apple-Mail=_61703D8B-882B-4661-A5E1-C9F9F69B22BD--

From fenix@google.com  Tue Feb  4 09:38:28 2014
Return-Path: <fenix@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 6E8731A0100 for <hybi@ietfa.amsl.com>; Tue,  4 Feb 2014 09:38:28 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.313
X-Spam-Level: 
X-Spam-Status: No, score=-1.313 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, J_CHICKENPOX_65=0.6, RP_MATCHES_RCVD=-0.535, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Bt350rY6fzM6 for <hybi@ietfa.amsl.com>; Tue,  4 Feb 2014 09:38:26 -0800 (PST)
Received: from mail-qc0-x22b.google.com (mail-qc0-x22b.google.com [IPv6:2607:f8b0:400d:c01::22b]) by ietfa.amsl.com (Postfix) with ESMTP id B829B1A0035 for <hybi@ietf.org>; Tue,  4 Feb 2014 09:38:25 -0800 (PST)
Received: by mail-qc0-f171.google.com with SMTP id n7so14159020qcx.16 for <hybi@ietf.org>; Tue, 04 Feb 2014 09:38:25 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=IQg56ZH//gwXeRmRr01CIYPIAEQgr+qBSwjbIXR6qj8=; b=Ix7xfhI+hFEVEzOvHHnACcO6GmlgWFIyzLf54xzZRteUb4O69CL0O3ktCY/p5Cn+PL QFEAgmVv5mNj4M+VdSRWi00aZKqw1blEwDn6xvEf1O1X6vpAQcXxGzY8dCqYNWyLINGX TK+G6Cdba1kf7a8WLl0dL4x2bCq0DeHXkjppa0ixO21sTKf1W5EXgxQ/6aSn5rGnqtH3 OMsu8y8LLPVbErd/1YAzUMqvy9lRIHZP+s6G7whJwTAm/CUWXV4Pm2Jn5+t9JLd/Oqbm R6FmQql+8fk9X/zjOCeaYrnKj2+9IAN+achynNqun3hc97RjTfUBLZZX+jUGypSVPJZt th6w==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=IQg56ZH//gwXeRmRr01CIYPIAEQgr+qBSwjbIXR6qj8=; b=lGrTFLtywhGnEgpb1Rja30w4bnMb7ntbxX5j5k7u4F36BffZ7bU2semGsGRugeHPhz maJfDkWq6JoT+ndKgigd9Q1PSAwqBJ8qRZ1jMdNUkolQXd2Ml0nkY31PrS90U/xyOYVO Qafm+Ehji9b3Kc5hHw4I9YI8SSVIJ+G1AW2/zBKGVsx0xH/YgkQhobM2Gk8EdJoc9ikP dIGXrq7t1XlRwoE4sm4PxyCD0n3AGQEsqHbvwKI83+omEnsJ+diEh8GBsNoEVdP1F6or 95SvnziEh2INbMHpozjPAeOaT3PxLD6GwDAcHr0DwI4qPWae+/5Pmim/VWdMCA0nOIR/ rBEw==
X-Gm-Message-State: ALoCoQnMo3icQlLaP719r/Fgc/pPg8vHfHHlZ1LRJNZcxotWN3I8pKsDt/mi7PsDKloOt+79oC0IpM03M/ISD1aOxLg8d/uFK/vEPNQeUVwHhKf+LopHdLTabQJPSZ4UkqL/P6NLiUlundRVjAtUWe//ctGNDSFzLxNHTMKotNsiiFqmum8NQ8zMctrVRQH2zmUTDp2tI4Jk
MIME-Version: 1.0
X-Received: by 10.224.66.66 with SMTP id m2mr67454416qai.67.1391535502699; Tue, 04 Feb 2014 09:38:22 -0800 (PST)
Received: by 10.229.13.7 with HTTP; Tue, 4 Feb 2014 09:38:22 -0800 (PST)
In-Reply-To: <CABihn6Gc34oSR70kkYUxpxQOMb2ZepR9STkLoe-SrZPTP8voYg@mail.gmail.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <CABihn6GSrU8oJa4nbMJiDDo2Z+-c4fT69roVFxcHYM4XYFToRQ@mail.gmail.com> <F1F115D4-29EB-457B-B02C-71E46D3E1C81@zaphoyd.com> <CABihn6GXxUwHwLd+926VPEkZxHRZUUi-3dw0PoNaO6aK8X4sEQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446C1E8D2C@EXVMBX020-12.exch020.serverdata.net> <CAGzyod7pt_PzjexmnZj8dSMg3BooUmuMb6xB5P0sG1t6S5FXZw@mail.gmail.com> <634914A010D0B943A035D226786325D4446C1E925F@EXVMBX020-12.exch020.serverdata.net> <CABihn6GrDwu=74WEACTrd8RMwDe78G-Q3X+e_OUrYbPcFm8gZQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446C1E9735@EXVMBX020-12.exch020.serverdata.net> <CAGzyod6_9H=a+uejDTEd_6HfbD2e2ynkA2frCnhk0TboricgfQ@mail.gmail.com> <CABihn6Gc34oSR70kkYUxpxQOMb2ZepR9STkLoe-SrZPTP8voYg@mail.gmail.com>
Date: Tue, 4 Feb 2014 09:38:22 -0800
Message-ID: <CAGzyod6dmqyu2n-0=YzSK_xkeS3hYFEJsCcERgKjvknWUXD5yA@mail.gmail.com>
From: Roberto Peon <fenix@google.com>
To: Yutaka Hirano <yhirano@chromium.org>
Content-Type: multipart/alternative; boundary=001a11c2c9a854c4fd04f1981b84
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 04 Feb 2014 17:38:28 -0000

--001a11c2c9a854c4fd04f1981b84
Content-Type: text/plain; charset=UTF-8

>From my viewpoint, that depends on how you define it.
If you say that everything until MSG_DONE is part of a message, you're
achieved what I want. If there is metadata in there, that is fine. HTTP/2
is spec'd to ensure that metadata and data are always synchronized.
Metadata 10283 bytes into a message will always be 10283 bytes into a
message.
-=R


On Tue, Feb 4, 2014 at 6:24 AM, Yutaka Hirano <yhirano@chromium.org> wrote:

> Hi Roberto,
>
>> I'm interested in correctly muxing, demuxing, and loadbalancing of
>> streams-of-messages without needing to understand application-layer
>> semantics.
>
> IIUC None of Plan A, B and C provide the ability to detect exact message
> boundary without understanding WebSocket level semantics even if we give up
> some extensions, because of control message injection.
> Introducing START_MESSAGE flag in addition to MSG_DONE enables it with
> Plan A or Plan C. Is it acceptable for you?
>
>
>
> On Tue, Feb 4, 2014 at 2:15 AM, Roberto Peon <fenix@google.com> wrote:
>
>> I'm interested in supporting the W3C WS API.
>> I'm interested in correctly muxing, demuxing, and loadbalancing of
>> streams-of-messages without needing to understand application-layer
>> semantics.
>> I'm interested in enabling a single transport-layer session for both
>> HTTP, WS, and potentially other content.
>>
>> I'm not at all interested in needing to implement multiple separate
>> protocols in order to implement the WS API.
>> I'm uninterested in small-number-of-byte savings.
>> I am uninterested in any extensions or extension frameworks that require
>> any extension-specific knowledge in intermediaries.
>>
>>
>> On Fri, Jan 31, 2014 at 4:13 AM, Tobias Oberstein <
>> tobias.oberstein@tavendo.de> wrote:
>>
>>> Hello Yutaka,
>>>
>>>
>>>
>>> my priorities:
>>>
>>>
>>>
>>> 1. allows to provide W3C WebSocket API
>>>
>>> 2. no "80%-WebSocket-protocol-compatible" design (either compatible with
>>> full RFC6455 incl. any extensions or outright independent of WS)
>>>
>>> 3. does not put servers that provide both native WS and SPDY (but not
>>> any SPDY/WS) on same origin at a disadvantage
>>>
>>> 4. minimizes code coupling between WS and SPDY within implementations
>>>
>>>
>>>
>>> /Tobias
>>>
>>>
>>>
>>> *Von:* yhirano@google.com [mailto:yhirano@google.com] *Im Auftrag von *Yutaka
>>> Hirano
>>> *Gesendet:* Freitag, 31. Januar 2014 11:41
>>> *An:* Tobias Oberstein
>>> *Cc:* Roberto Peon; Peter Thorson; hybi@ietf.org
>>>
>>> *Betreff:* Re: [hybi] Discontinuation of mux standardizaton in favor of
>>> WS/HTTP/2.0
>>>
>>>
>>>
>>> Seriously: _what_ is the goal of this effort?
>>>
>>> My motivation is, specifying a multiplexed bidirectional message
>>> protocol conforming to the W3C WebSocket API on HTTP/2.0.
>>>
>>>
>>>
>>>  - The W3C WebSocket API
>>>
>>>  - Multiplexing
>>>
>>> is mandatory for me.
>>>
>>> Since It is built on HTTP/2.0, I prefer strongly the following:
>>>
>>>  - A mechanism that enables intermediaries to transfer WS over HTTP/2.0
>>> without fully understanding WS over HTTP/2.0.
>>>
>>>
>>>
>>> The remaining, though they are preferable, are not mandatory for me.
>>>
>>>  - RFC6455 conformance
>>>
>>>  - small amount of bandwidth efficiency
>>>
>>>  - the spec cleanness
>>>
>>>  - less-invasiveness
>>>
>>>
>>>
>>> Note: It does not mean I can't accept WS-TUNNEL.
>>>
>>>
>>>
>>> Tobias, Roberto, Peter, can you share your priority?
>>>
>>>
>>>
>>>
>>>
>>> On Thu, Jan 30, 2014 at 8:54 PM, Tobias Oberstein <
>>> tobias.oberstein@tavendo.de> wrote:
>>>
>>> >I don't find the frame-size arguments very compelling since I haven't
>>> seen or met anyone with a substantial deployment who cares significantly
>>> about the overhead difference given packet, TLS, etc. overheads.
>>> >There is no deployment that we know about that uses frame-based
>>> extensibility- we can (and should) remove it from the standard since it
>>> hasn't proven to be interesting.
>>>
>>> There is a draft for a WebSocket extension ("permessage-priority") being
>>> currently developed that makes use of this and we will start an
>>> experimental implementation in Autobahn. It _is_ interesting.
>>>
>>> In any case, the ability for framing, and the ability for WS extensions
>>> to interleave WS messages is part of RFC6455.
>>>
>>> Why remove something from RFC6455 just to fit the limited capabilities
>>> that SPDY offers?
>>>
>>>
>>> >I'm not sure I agree with your statement that libraries supporting
>>> HTTP/2 will include necessarily also have WS framing. It seems unlikely to
>>> me-- I know of at least one large deployment where this isn't at all true.
>>>
>>> I'd put it like this: any networking library that wants to support
>>> "messaging" on the Web will need to support native WebSocket, plus
>>> optionally any SPDY-based messaging.
>>>
>>> Since, WebSocket is deployed _today_ and HTTP2 won't be everywhere
>>> anytime soon.
>>>
>>> E.g. HTTP/2 support on Windows seems to be coming only on Windows 8+
>>> (since they have HTTP in-kernel). So native WebSocket is the only messaging
>>> that works on any IE version on Windows <8.
>>>
>>>
>>> >The HTTP/2 standard has (or shortly will have, as soon as the pull
>>> request goes in) all support necessary to handle any stream-of-message
>>> protocol given support of END_MESSAGE.
>>> >I'd strongly prefer that we map WS to those primitives so it, and any
>>> other stream-of-message protocols can be handled using loadbalancer/routing
>>> infrastructure which is and can remain protocol agnostic.
>>>
>>> That's fine, let's just not call it WebSocket-over-SPDY then;)
>>>
>>> Seriously: _what_ is the goal of this effort?
>>>
>>> /Tobias
>>>
>>>
>>>
>>> -=R
>>>
>>> On Wed, Jan 29, 2014 at 5:56 AM, Tobias Oberstein <
>>> tobias.oberstein@tavendo.de> wrote:
>>> >The proposal is mapping one WS Message to one HEADERS + multiple DATAs.
>>> >For large messages, it can achieve less overhead than WS-TUNNEL at the
>>> cost of some RFC6455 compatibility.
>>> How can it achieve less overhead?
>>>
>>> A SPDY HEADER with information from first WS frame seems to be always
>>> bigger than a SPDY DATA header (8 octets) + WS frame header.
>>>
>>> The only situation I can think of: a large, _fragmented_ WS message is
>>> reframed into a SPDY HEADER+DATAs sequence.
>>>
>>> Effectively, you do away with the individual WS frame information. And
>>> this exactly breaks full RFC6455 compatibility.
>>>
>>> /Tobias
>>> _______________________________________________
>>> hybi mailing list
>>> hybi@ietf.org
>>> https://www.ietf.org/mailman/listinfo/hybi
>>>
>>>
>>>
>>
>>
>

--001a11c2c9a854c4fd04f1981b84
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">From my viewpoint, that depends on how you define it.<div>=
If you say that everything until MSG_DONE is part of a message, you&#39;re =
achieved what I want. If there is metadata in there, that is fine. HTTP/2 i=
s spec&#39;d to ensure that metadata and data are always synchronized. Meta=
data 10283 bytes into a message will always be 10283 bytes into a message.<=
/div>
<div>-=3DR</div></div><div class=3D"gmail_extra"><br><br><div class=3D"gmai=
l_quote">On Tue, Feb 4, 2014 at 6:24 AM, Yutaka Hirano <span dir=3D"ltr">&l=
t;<a href=3D"mailto:yhirano@chromium.org" target=3D"_blank">yhirano@chromiu=
m.org</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div dir=3D"ltr"><div>Hi Roberto,=C2=A0</div=
><div class=3D"im"><blockquote class=3D"gmail_quote" style=3D"margin:0px 0p=
x 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border=
-left-style:solid;padding-left:1ex">
<span style=3D"font-family:arial,sans-serif;font-size:13px">I&#39;m interes=
ted in correctly muxing, demuxing, and loadbalancing of streams-of-messages=
 without needing to understand application-layer semantics.</span></blockqu=
ote>


</div><div>IIUC None of Plan A, B and C provide the ability to detect exact=
 message boundary without understanding WebSocket level semantics even if w=
e give up some extensions, because of control message injection.</div>
<div>Introducing START_MESSAGE flag in addition to MSG_DONE enables it with=
 Plan A or Plan C. Is it acceptable for you?</div>
<div><br></div></div><div class=3D"HOEnZb"><div class=3D"h5"><div class=3D"=
gmail_extra"><br><br><div class=3D"gmail_quote">On Tue, Feb 4, 2014 at 2:15=
 AM, Roberto Peon <span dir=3D"ltr">&lt;<a href=3D"mailto:fenix@google.com"=
 target=3D"_blank">fenix@google.com</a>&gt;</span> wrote:<br>

<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div dir=3D"ltr"><span style=3D"font-family:=
arial,sans-serif;font-size:13px">I&#39;m interested in supporting the W3C W=
S API.</span><div style=3D"font-family:arial,sans-serif;font-size:13px">

I&#39;m interested in correctly muxing, demuxing, and loadbalancing of stre=
ams-of-messages without needing to understand application-layer semantics.<=
/div>
<div style=3D"font-family:arial,sans-serif;font-size:13px">I&#39;m interest=
ed in enabling a single transport-layer session for both HTTP, WS, and pote=
ntially other content.</div><div style=3D"font-family:arial,sans-serif;font=
-size:13px">


<br></div><div style=3D"font-family:arial,sans-serif;font-size:13px">I&#39;=
m not at all interested in needing to implement multiple separate protocols=
 in order to implement the WS API.</div><div style=3D"font-family:arial,san=
s-serif;font-size:13px">


I&#39;m uninterested in small-number-of-byte savings.</div><div style=3D"fo=
nt-family:arial,sans-serif;font-size:13px">I am uninterested in any extensi=
ons or extension frameworks that require any extension-specific knowledge i=
n intermediaries.=C2=A0</div>


</div><div><div><div class=3D"gmail_extra"><br><br><div class=3D"gmail_quot=
e">On Fri, Jan 31, 2014 at 4:13 AM, Tobias Oberstein <span dir=3D"ltr">&lt;=
<a href=3D"mailto:tobias.oberstein@tavendo.de" target=3D"_blank">tobias.obe=
rstein@tavendo.de</a>&gt;</span> wrote:<br>


<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div lang=3D"DE" link=3D"blue" vlink=3D"purp=
le"><div><p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family=
:&quot;Calibri&quot;,&quot;sans-serif&quot;;color:#1f497d">Hello Yutaka,<u>=
</u><u></u></span></p>


<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></spa=
n></p><p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&q=
uot;Calibri&quot;,&quot;sans-serif&quot;;color:#1f497d">my priorities:<u></=
u><u></u></span></p>


<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></spa=
n></p><p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&q=
uot;Calibri&quot;,&quot;sans-serif&quot;;color:#1f497d">1. allows to provid=
e W3C WebSocket API<u></u><u></u></span></p>


<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">2. no &quot;80%-WebSocket=
-protocol-compatible&quot; design (either compatible with full RFC6455 incl=
. any extensions or outright independent of WS)<u></u><u></u></span></p>


<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">3. does not put servers t=
hat provide both native WS and SPDY (but not any SPDY/WS) on same origin at=
 a disadvantage<u></u><u></u></span></p>


<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">4. minimizes code couplin=
g between WS and SPDY within implementations<u></u><u></u></span></p><p cla=
ss=3D"MsoNormal">


<span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-=
serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></span></p><p class=3D"MsoNo=
rmal"><span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot=
;sans-serif&quot;;color:#1f497d">/Tobias<u></u><u></u></span></p>


<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></spa=
n></p><div style=3D"border:none;border-left:solid blue 1.5pt;padding:0cm 0c=
m 0cm 4.0pt">


<div><div style=3D"border:none;border-top:solid #e1e1e1 1.0pt;padding:3.0pt=
 0cm 0cm 0cm"><p class=3D"MsoNormal"><b><span style=3D"font-size:11.0pt;fon=
t-family:&quot;Calibri&quot;,&quot;sans-serif&quot;">Von:</span></b><span s=
tyle=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif&q=
uot;"> <a href=3D"mailto:yhirano@google.com" target=3D"_blank">yhirano@goog=
le.com</a> [mailto:<a href=3D"mailto:yhirano@google.com" target=3D"_blank">=
yhirano@google.com</a>] <b>Im Auftrag von </b>Yutaka Hirano<br>


<b>Gesendet:</b> Freitag, 31. Januar 2014 11:41<br><b>An:</b> Tobias Oberst=
ein<br><b>Cc:</b> Roberto Peon; Peter Thorson; <a href=3D"mailto:hybi@ietf.=
org" target=3D"_blank">hybi@ietf.org</a></span></p><div><br><b>Betreff:</b>=
 Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0<u=
></u><u></u></div>


<p></p></div></div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p><div><blo=
ckquote style=3D"border:none;border-left:solid #cccccc 1.0pt;padding:0cm 0c=
m 0cm 6.0pt;margin-left:4.8pt;margin-right:0cm"><p class=3D"MsoNormal"><spa=
n style=3D"font-size:10.0pt;font-family:&quot;Arial&quot;,&quot;sans-serif&=
quot;">Seriously: _what_ is the goal of this effort?</span><u></u><u></u></=
p>


</blockquote><div><div><div><p class=3D"MsoNormal">My motivation is, specif=
ying a multiplexed bidirectional message protocol conforming to the W3C Web=
Socket API on HTTP/2.0.<u></u><u></u></p></div><div><p class=3D"MsoNormal">


<u></u>=C2=A0<u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- The W3C W=
ebSocket API<u></u><u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- Mul=
tiplexing<u></u><u></u></p></div><div><p class=3D"MsoNormal">is mandatory f=
or me.<u></u><u></u></p>


</div><div><p class=3D"MsoNormal">Since It is built on HTTP/2.0, I prefer s=
trongly the following:<u></u><u></u></p></div><div><p class=3D"MsoNormal">=
=C2=A0- A mechanism that enables intermediaries to transfer WS over HTTP/2.=
0 without fully understanding WS over HTTP/2.0.<u></u><u></u></p>


</div><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div><div><p cla=
ss=3D"MsoNormal">The remaining, though they are preferable, are not mandato=
ry for me.<u></u><u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- RFC64=
55 conformance<u></u><u></u></p>


</div><div><p class=3D"MsoNormal">=C2=A0- small amount of bandwidth efficie=
ncy<u></u><u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- the spec cle=
anness<u></u><u></u></p></div><div><p class=3D"MsoNormal">=C2=A0- less-inva=
siveness<u></u><u></u></p>


</div><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div><div><p cla=
ss=3D"MsoNormal">Note: It does not mean I can&#39;t accept WS-TUNNEL.<u></u=
><u></u></p></div><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div=
><div><p class=3D"MsoNormal">


Tobias, Roberto, Peter, can you share your priority?<u></u><u></u></p></div=
><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div></div></div></di=
v><div><div><div><p class=3D"MsoNormal" style=3D"margin-bottom:12.0pt">
<u></u>=C2=A0<u></u></p><div><p class=3D"MsoNormal">On Thu, Jan 30, 2014 at=
 8:54 PM, Tobias Oberstein &lt;<a href=3D"mailto:tobias.oberstein@tavendo.d=
e" target=3D"_blank">tobias.oberstein@tavendo.de</a>&gt; wrote:<u></u><u></=
u></p>

<blockquote style=3D"border:none;border-left:solid #cccccc 1.0pt;padding:0c=
m 0cm 0cm 6.0pt;margin-left:4.8pt;margin-right:0cm">
<div><p class=3D"MsoNormal" style=3D"margin-bottom:12.0pt">&gt;I don&#39;t =
find the frame-size arguments very compelling since I haven&#39;t seen or m=
et anyone with a substantial deployment who cares significantly about the o=
verhead difference given packet, TLS, etc. overheads.<br>


&gt;There is no deployment that we know about that uses frame-based extensi=
bility- we can (and should) remove it from the standard since it hasn&#39;t=
 proven to be interesting.<u></u><u></u></p></div><p class=3D"MsoNormal">


There is a draft for a WebSocket extension (&quot;permessage-priority&quot;=
) being currently developed that makes use of this and we will start an exp=
erimental implementation in Autobahn. It _is_ interesting.<br><br>In any ca=
se, the ability for framing, and the ability for WS extensions to interleav=
e WS messages is part of RFC6455.<br>


<br>Why remove something from RFC6455 just to fit the limited capabilities =
that SPDY offers?<u></u><u></u></p><div><p class=3D"MsoNormal" style=3D"mar=
gin-bottom:12.0pt"><br>&gt;I&#39;m not sure I agree with your statement tha=
t libraries supporting HTTP/2 will include necessarily also have WS framing=
. It seems unlikely to me-- I know of at least one large deployment where t=
his isn&#39;t at all true.<u></u><u></u></p>


</div><p class=3D"MsoNormal">I&#39;d put it like this: any networking libra=
ry that wants to support &quot;messaging&quot; on the Web will need to supp=
ort native WebSocket, plus optionally any SPDY-based messaging.<br><br>Sinc=
e, WebSocket is deployed _today_ and HTTP2 won&#39;t be everywhere anytime =
soon.<br>


<br>E.g. HTTP/2 support on Windows seems to be coming only on Windows 8+ (s=
ince they have HTTP in-kernel). So native WebSocket is the only messaging t=
hat works on any IE version on Windows &lt;8.<u></u><u></u></p><div><p clas=
s=3D"MsoNormal" style=3D"margin-bottom:12.0pt">


<br>&gt;The HTTP/2 standard has (or shortly will have, as soon as the pull =
request goes in) all support necessary to handle any stream-of-message prot=
ocol given support of END_MESSAGE.<br>&gt;I&#39;d strongly prefer that we m=
ap WS to those primitives so it, and any other stream-of-message protocols =
can be handled using loadbalancer/routing infrastructure which is and can r=
emain protocol agnostic.<u></u><u></u></p>


</div><p class=3D"MsoNormal">That&#39;s fine, let&#39;s just not call it We=
bSocket-over-SPDY then;)<br><br>Seriously: _what_ is the goal of this effor=
t?<br><span style=3D"color:#888888"><br><span>/Tobias</span></span><u></u><=
u></u></p>


<div><div><p class=3D"MsoNormal" style=3D"margin-bottom:12.0pt"><br><br>-=
=3DR<br><br>On Wed, Jan 29, 2014 at 5:56 AM, Tobias Oberstein &lt;<a href=
=3D"mailto:tobias.oberstein@tavendo.de" target=3D"_blank">tobias.oberstein@=
tavendo.de</a>&gt; wrote:<br>


&gt;The proposal is mapping one WS Message to one HEADERS + multiple DATAs.=
<br>&gt;For large messages, it can achieve less overhead than WS-TUNNEL at =
the cost of some RFC6455 compatibility.<br>How can it achieve less overhead=
?<br>


<br>A SPDY HEADER with information from first WS frame seems to be always b=
igger than a SPDY DATA header (8 octets) + WS frame header.<br><br>The only=
 situation I can think of: a large, _fragmented_ WS message is reframed int=
o a SPDY HEADER+DATAs sequence.<br>


<br>Effectively, you do away with the individual WS frame information. And =
this exactly breaks full RFC6455 compatibility.<br><br>/Tobias<br>_________=
______________________________________<br>hybi mailing list<br><a href=3D"m=
ailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><br>


<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><u></u><u></u></p></div></div><=
/blockquote></div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div></di=
v></div>


</div></div></div></blockquote></div><br></div>
</div></div></blockquote></div><br></div>
</div></div></blockquote></div><br></div>

--001a11c2c9a854c4fd04f1981b84--

From fenix@google.com  Tue Feb  4 09:40:28 2014
Return-Path: <fenix@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 5E9E61A01AB for <hybi@ietfa.amsl.com>; Tue,  4 Feb 2014 09:40:28 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.313
X-Spam-Level: 
X-Spam-Status: No, score=-1.313 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, J_CHICKENPOX_65=0.6, RP_MATCHES_RCVD=-0.535, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id b1sShaXk3PGH for <hybi@ietfa.amsl.com>; Tue,  4 Feb 2014 09:40:25 -0800 (PST)
Received: from mail-qa0-x22d.google.com (mail-qa0-x22d.google.com [IPv6:2607:f8b0:400d:c00::22d]) by ietfa.amsl.com (Postfix) with ESMTP id 3CE7A1A01A9 for <hybi@ietf.org>; Tue,  4 Feb 2014 09:40:25 -0800 (PST)
Received: by mail-qa0-f45.google.com with SMTP id ii20so12730441qab.32 for <hybi@ietf.org>; Tue, 04 Feb 2014 09:40:24 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=FI2Wh2gl0LBECfwuGlXntkSKCTs/ggi5MadDCZbJNmo=; b=jWhb9My4yYOa++zITGjS/jT5nWkpYPrTe48UqFtViNn76m0oaXTuex4V3MU5QCcgh0 wF8n6u/Zgr1A9zfeDImTLmp+/oZjfjUE7e6RVDetUyBrlV/0jwbMGpjTtLmPLpIXdaXl q2qDzth36Vp710WfPqarHBBwRGawOEPO95Hza8/+7jRKdVHoIWT7sbEcnYcCXaCX+y0R utBH4wDKrVJWzpVC125oAkLLuuX0YOeo+d7vv88CKG3/bUVXloTdUeDuklj9R9JP/WDA lhGq46TRVWzTVqAR46ZQ0tMVU+F2JyHKm40RVTSd4cW8jEjFuMcgjJr6Re9sP+OtAd18 Yzbw==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=FI2Wh2gl0LBECfwuGlXntkSKCTs/ggi5MadDCZbJNmo=; b=YnZD+SSoV97kvkPVFHBplL69/ib6QPxRo+GejzzeDeIcgjG6phR1TEdlg/PwiSSk5g /S1WeT94ppxFyOw7H5fP11K2hQDnlnROD45wU6Wf7sUj1A1vZewv7f1rY2Krvu+EtH1k KynSn0jabMkosWUaF/esgK7E74kHPo1r7lUcfRdx5CBuqfq7E55qNeRwEJqzHEotvQpU TXBowbYr5XviXbVMG8TMcqmo9UoGcP3P2JtA590VcsDMatNgABIJg6/qkU/RPCrhM31U uk3Oj50HpgTuaDyAo2SBp/9asJMiFOcOTXmXQD8mUISno4ykIMGVTWDr0GE37Zn0hGPa pxFw==
X-Gm-Message-State: ALoCoQkRqGcKJKeyEXdGSJZoJPELWj+zbm0iMBFZZIlOXVTbfKGBf16uPAckgr2R45rZtDzljOErctmuO2P9OmdcR1HZZ9IAlEI2F7vULUQjj/bDKWGQcK+3Xq5Gtb04raLbTAh3ItWXvMAMj6jlb/F76cMGz1l77RtcE5copacb5AG+LhfOZ5/x3MqQ6+X0/kxzim1967Ss
MIME-Version: 1.0
X-Received: by 10.140.101.162 with SMTP id u31mr65084101qge.107.1391535624613;  Tue, 04 Feb 2014 09:40:24 -0800 (PST)
Received: by 10.229.13.7 with HTTP; Tue, 4 Feb 2014 09:40:24 -0800 (PST)
In-Reply-To: <634914A010D0B943A035D226786325D4446C2955D9@EXVMBX020-12.exch020.serverdata.net>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <CABihn6GSrU8oJa4nbMJiDDo2Z+-c4fT69roVFxcHYM4XYFToRQ@mail.gmail.com> <F1F115D4-29EB-457B-B02C-71E46D3E1C81@zaphoyd.com> <CABihn6GXxUwHwLd+926VPEkZxHRZUUi-3dw0PoNaO6aK8X4sEQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446C1E8D2C@EXVMBX020-12.exch020.serverdata.net> <CAGzyod7pt_PzjexmnZj8dSMg3BooUmuMb6xB5P0sG1t6S5FXZw@mail.gmail.com> <634914A010D0B943A035D226786325D4446C1E925F@EXVMBX020-12.exch020.serverdata.net> <CABihn6GrDwu=74WEACTrd8RMwDe78G-Q3X+e_OUrYbPcFm8gZQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446C1E9735@EXVMBX020-12.exch020.serverdata.net> <CAGzyod6_9H=a+uejDTEd_6HfbD2e2ynkA2frCnhk0TboricgfQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446C2955D9@EXVMBX020-12.exch020.serverdata.net>
Date: Tue, 4 Feb 2014 09:40:24 -0800
Message-ID: <CAGzyod6H-To9EhaPqW5C0c9xyy5XnRg_851GuT3q=X+KoXe7Wg@mail.gmail.com>
From: Roberto Peon <fenix@google.com>
To: Tobias Oberstein <tobias.oberstein@tavendo.de>
Content-Type: multipart/alternative; boundary=001a11c1640a990cd904f1982213
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 04 Feb 2014 17:40:28 -0000

--001a11c1640a990cd904f1982213
Content-Type: text/plain; charset=UTF-8

On Tue, Feb 4, 2014 at 7:02 AM, Tobias Oberstein <
tobias.oberstein@tavendo.de> wrote:

> >I'm interested in correctly muxing, demuxing, and loadbalancing of
> streams-of-messages without needing to understand application-layer
> semantics.
>
> What is the unit of muxing/demuxing/loadbalancing?
>
> If the unit is the SPDY _stream_, then we can tunnel WebSocket with any
> extensions by simply transporting complete
> WebSocket frames inside the payload of SPDY data frames - with no flags
> whatsoever at SPDY header level.
>
> If the unit is the individual application _message_, then understanding of
> application message boundaries is required at SPDY level.
>
> Which unit did you think of?
>

Messages, since I have to worry about other message-based protocols.


>
> >I'm not at all interested in needing to implement multiple separate
> protocols in order to implement the WS API.
>
> Sure, but it looks like a practical necessity. RFC6455 is out on the
> street and widely implemented.
>

Well, I'm not interested in implementing it :)


>
> >I am uninterested in any extensions or extension frameworks that require
> any extension-specific knowledge in intermediaries.
>
> Not sure what this means ..
>
> WebSocket extensions are allowed to interleave messages explicitly in
> RFC6455. Intermediaries are not required
> to _understand_ the interleaving. They are required to _not touch_ the
> fragmentation in this case.
>
> A HTTP2.0 intermediary that does it's muxing/demuxing/loadbalancing at the
> SPDY stream level does not need
> to understand anything about tunneled WS.
>

s/stream/message, and I agree.
-=R


>
> /Tobias
>
>
> On Fri, Jan 31, 2014 at 4:13 AM, Tobias Oberstein <
> tobias.oberstein@tavendo.de> wrote:
> Hello Yutaka,
>
> my priorities:
>
> 1. allows to provide W3C WebSocket API
> 2. no "80%-WebSocket-protocol-compatible" design (either compatible with
> full RFC6455 incl. any extensions or outright independent of WS)
> 3. does not put servers that provide both native WS and SPDY (but not any
> SPDY/WS) on same origin at a disadvantage
> 4. minimizes code coupling between WS and SPDY within implementations
>
> /Tobias
>
> Von: yhirano@google.com [mailto:yhirano@google.com] Im Auftrag von Yutaka
> Hirano
> Gesendet: Freitag, 31. Januar 2014 11:41
> An: Tobias Oberstein
> Cc: Roberto Peon; Peter Thorson; hybi@ietf.org
>
> Betreff: Re: [hybi] Discontinuation of mux standardizaton in favor of
> WS/HTTP/2.0
>
> Seriously: _what_ is the goal of this effort?
> My motivation is, specifying a multiplexed bidirectional message protocol
> conforming to the W3C WebSocket API on HTTP/2.0.
>
>  - The W3C WebSocket API
>  - Multiplexing
> is mandatory for me.
> Since It is built on HTTP/2.0, I prefer strongly the following:
>  - A mechanism that enables intermediaries to transfer WS over HTTP/2.0
> without fully understanding WS over HTTP/2.0.
>
> The remaining, though they are preferable, are not mandatory for me.
>  - RFC6455 conformance
>  - small amount of bandwidth efficiency
>  - the spec cleanness
>  - less-invasiveness
>
> Note: It does not mean I can't accept WS-TUNNEL.
>
> Tobias, Roberto, Peter, can you share your priority?
>
>
> On Thu, Jan 30, 2014 at 8:54 PM, Tobias Oberstein <
> tobias.oberstein@tavendo.de> wrote:
> >I don't find the frame-size arguments very compelling since I haven't
> seen or met anyone with a substantial deployment who cares significantly
> about the overhead difference given packet, TLS, etc. overheads.
> >There is no deployment that we know about that uses frame-based
> extensibility- we can (and should) remove it from the standard since it
> hasn't proven to be interesting.
> There is a draft for a WebSocket extension ("permessage-priority") being
> currently developed that makes use of this and we will start an
> experimental implementation in Autobahn. It _is_ interesting.
>
> In any case, the ability for framing, and the ability for WS extensions to
> interleave WS messages is part of RFC6455.
>
> Why remove something from RFC6455 just to fit the limited capabilities
> that SPDY offers?
>
> >I'm not sure I agree with your statement that libraries supporting HTTP/2
> will include necessarily also have WS framing. It seems unlikely to me-- I
> know of at least one large deployment where this isn't at all true.
> I'd put it like this: any networking library that wants to support
> "messaging" on the Web will need to support native WebSocket, plus
> optionally any SPDY-based messaging.
>
> Since, WebSocket is deployed _today_ and HTTP2 won't be everywhere anytime
> soon.
>
> E.g. HTTP/2 support on Windows seems to be coming only on Windows 8+
> (since they have HTTP in-kernel). So native WebSocket is the only messaging
> that works on any IE version on Windows <8.
>
> >The HTTP/2 standard has (or shortly will have, as soon as the pull
> request goes in) all support necessary to handle any stream-of-message
> protocol given support of END_MESSAGE.
> >I'd strongly prefer that we map WS to those primitives so it, and any
> other stream-of-message protocols can be handled using loadbalancer/routing
> infrastructure which is and can remain protocol agnostic.
> That's fine, let's just not call it WebSocket-over-SPDY then;)
>
> Seriously: _what_ is the goal of this effort?
>
> /Tobias
>
>
> -=R
>
> On Wed, Jan 29, 2014 at 5:56 AM, Tobias Oberstein <
> tobias.oberstein@tavendo.de> wrote:
> >The proposal is mapping one WS Message to one HEADERS + multiple DATAs.
> >For large messages, it can achieve less overhead than WS-TUNNEL at the
> cost of some RFC6455 compatibility.
> How can it achieve less overhead?
>
> A SPDY HEADER with information from first WS frame seems to be always
> bigger than a SPDY DATA header (8 octets) + WS frame header.
>
> The only situation I can think of: a large, _fragmented_ WS message is
> reframed into a SPDY HEADER+DATAs sequence.
>
> Effectively, you do away with the individual WS frame information. And
> this exactly breaks full RFC6455 compatibility.
>
> /Tobias
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi
>
>
>

--001a11c1640a990cd904f1982213
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><br><div class=3D"gmail_extra"><br><br><div class=3D"gmail=
_quote">On Tue, Feb 4, 2014 at 7:02 AM, Tobias Oberstein <span dir=3D"ltr">=
&lt;<a href=3D"mailto:tobias.oberstein@tavendo.de" target=3D"_blank">tobias=
.oberstein@tavendo.de</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div class=3D"im">&gt;I&#39;m interested in =
correctly muxing, demuxing, and loadbalancing of streams-of-messages withou=
t needing to understand application-layer semantics.<br>

<br>
</div>What is the unit of muxing/demuxing/loadbalancing?<br>
<br>
If the unit is the SPDY _stream_, then we can tunnel WebSocket with any ext=
ensions by simply transporting complete<br>
WebSocket frames inside the payload of SPDY data frames - with no flags wha=
tsoever at SPDY header level.<br>
<br>
If the unit is the individual application _message_, then understanding of =
application message boundaries is required at SPDY level.<br>
<br>
Which unit did you think of?<br></blockquote><div><br></div><div>Messages, =
since I have to worry about other message-based protocols.</div><div>=C2=A0=
</div><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-l=
eft:1px #ccc solid;padding-left:1ex">

<div class=3D"im"><br>
&gt;I&#39;m not at all interested in needing to implement multiple separate=
 protocols in order to implement the WS API.<br>
<br>
</div>Sure, but it looks like a practical necessity. RFC6455 is out on the =
street and widely implemented.<br></blockquote><div><br></div><div>Well, I&=
#39;m not interested in implementing it :)</div><div>=C2=A0<br></div><block=
quote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc=
 solid;padding-left:1ex">

<div class=3D"im"><br>
&gt;I am uninterested in any extensions or extension frameworks that requir=
e any extension-specific knowledge in intermediaries.=C2=A0<br>
<br>
</div>Not sure what this means ..<br>
<br>
WebSocket extensions are allowed to interleave messages explicitly in RFC64=
55. Intermediaries are not required<br>
to _understand_ the interleaving. They are required to _not touch_ the frag=
mentation in this case.<br>
<br>
A HTTP2.0 intermediary that does it&#39;s muxing/demuxing/loadbalancing at =
the SPDY stream level does not need<br>
to understand anything about tunneled WS.<br></blockquote><div><br></div><d=
iv>s/stream/message, and I agree.</div><div>-=3DR</div><div>=C2=A0</div><bl=
ockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #=
ccc solid;padding-left:1ex">

<span class=3D"HOEnZb"><font color=3D"#888888"><br>
/Tobias<br>
</font></span><div class=3D"HOEnZb"><div class=3D"h5"><br>
<br>
On Fri, Jan 31, 2014 at 4:13 AM, Tobias Oberstein &lt;<a href=3D"mailto:tob=
ias.oberstein@tavendo.de">tobias.oberstein@tavendo.de</a>&gt; wrote:<br>
Hello Yutaka,<br>
=C2=A0<br>
my priorities:<br>
=C2=A0<br>
1. allows to provide W3C WebSocket API<br>
2. no &quot;80%-WebSocket-protocol-compatible&quot; design (either compatib=
le with full RFC6455 incl. any extensions or outright independent of WS)<br=
>
3. does not put servers that provide both native WS and SPDY (but not any S=
PDY/WS) on same origin at a disadvantage<br>
4. minimizes code coupling between WS and SPDY within implementations<br>
=C2=A0<br>
/Tobias<br>
=C2=A0<br>
Von: <a href=3D"mailto:yhirano@google.com">yhirano@google.com</a> [mailto:<=
a href=3D"mailto:yhirano@google.com">yhirano@google.com</a>] Im Auftrag von=
 Yutaka Hirano<br>
Gesendet: Freitag, 31. Januar 2014 11:41<br>
An: Tobias Oberstein<br>
Cc: Roberto Peon; Peter Thorson; <a href=3D"mailto:hybi@ietf.org">hybi@ietf=
.org</a><br>
<br>
Betreff: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HT=
TP/2.0<br>
=C2=A0<br>
Seriously: _what_ is the goal of this effort?<br>
My motivation is, specifying a multiplexed bidirectional message protocol c=
onforming to the W3C WebSocket API on HTTP/2.0.<br>
=C2=A0<br>
=C2=A0- The W3C WebSocket API<br>
=C2=A0- Multiplexing<br>
is mandatory for me.<br>
Since It is built on HTTP/2.0, I prefer strongly the following:<br>
=C2=A0- A mechanism that enables intermediaries to transfer WS over HTTP/2.=
0 without fully understanding WS over HTTP/2.0.<br>
=C2=A0<br>
The remaining, though they are preferable, are not mandatory for me.<br>
=C2=A0- RFC6455 conformance<br>
=C2=A0- small amount of bandwidth efficiency<br>
=C2=A0- the spec cleanness<br>
=C2=A0- less-invasiveness<br>
=C2=A0<br>
Note: It does not mean I can&#39;t accept WS-TUNNEL.<br>
=C2=A0<br>
Tobias, Roberto, Peter, can you share your priority?<br>
=C2=A0<br>
=C2=A0<br>
On Thu, Jan 30, 2014 at 8:54 PM, Tobias Oberstein &lt;<a href=3D"mailto:tob=
ias.oberstein@tavendo.de">tobias.oberstein@tavendo.de</a>&gt; wrote:<br>
&gt;I don&#39;t find the frame-size arguments very compelling since I haven=
&#39;t seen or met anyone with a substantial deployment who cares significa=
ntly about the overhead difference given packet, TLS, etc. overheads.<br>

&gt;There is no deployment that we know about that uses frame-based extensi=
bility- we can (and should) remove it from the standard since it hasn&#39;t=
 proven to be interesting.<br>
There is a draft for a WebSocket extension (&quot;permessage-priority&quot;=
) being currently developed that makes use of this and we will start an exp=
erimental implementation in Autobahn. It _is_ interesting.<br>
<br>
In any case, the ability for framing, and the ability for WS extensions to =
interleave WS messages is part of RFC6455.<br>
<br>
Why remove something from RFC6455 just to fit the limited capabilities that=
 SPDY offers?<br>
<br>
&gt;I&#39;m not sure I agree with your statement that libraries supporting =
HTTP/2 will include necessarily also have WS framing. It seems unlikely to =
me-- I know of at least one large deployment where this isn&#39;t at all tr=
ue.<br>

I&#39;d put it like this: any networking library that wants to support &quo=
t;messaging&quot; on the Web will need to support native WebSocket, plus op=
tionally any SPDY-based messaging.<br>
<br>
Since, WebSocket is deployed _today_ and HTTP2 won&#39;t be everywhere anyt=
ime soon.<br>
<br>
E.g. HTTP/2 support on Windows seems to be coming only on Windows 8+ (since=
 they have HTTP in-kernel). So native WebSocket is the only messaging that =
works on any IE version on Windows &lt;8.<br>
<br>
&gt;The HTTP/2 standard has (or shortly will have, as soon as the pull requ=
est goes in) all support necessary to handle any stream-of-message protocol=
 given support of END_MESSAGE.<br>
&gt;I&#39;d strongly prefer that we map WS to those primitives so it, and a=
ny other stream-of-message protocols can be handled using loadbalancer/rout=
ing infrastructure which is and can remain protocol agnostic.<br>
That&#39;s fine, let&#39;s just not call it WebSocket-over-SPDY then;)<br>
<br>
Seriously: _what_ is the goal of this effort?<br>
<br>
/Tobias<br>
<br>
<br>
-=3DR<br>
<br>
On Wed, Jan 29, 2014 at 5:56 AM, Tobias Oberstein &lt;<a href=3D"mailto:tob=
ias.oberstein@tavendo.de">tobias.oberstein@tavendo.de</a>&gt; wrote:<br>
&gt;The proposal is mapping one WS Message to one HEADERS + multiple DATAs.=
<br>
&gt;For large messages, it can achieve less overhead than WS-TUNNEL at the =
cost of some RFC6455 compatibility.<br>
How can it achieve less overhead?<br>
<br>
A SPDY HEADER with information from first WS frame seems to be always bigge=
r than a SPDY DATA header (8 octets) + WS frame header.<br>
<br>
The only situation I can think of: a large, _fragmented_ WS message is refr=
amed into a SPDY HEADER+DATAs sequence.<br>
<br>
Effectively, you do away with the individual WS frame information. And this=
 exactly breaks full RFC6455 compatibility.<br>
<br>
/Tobias<br>
_______________________________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
=C2=A0<br>
<br>
</div></div></blockquote></div><br></div></div>

--001a11c1640a990cd904f1982213--

From tobias.oberstein@tavendo.de  Tue Feb  4 10:11:29 2014
Return-Path: <tobias.oberstein@tavendo.de>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id A45491A01AB for <hybi@ietfa.amsl.com>; Tue,  4 Feb 2014 10:11:29 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.3
X-Spam-Level: 
X-Spam-Status: No, score=-1.3 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, J_CHICKENPOX_65=0.6, RCVD_IN_DNSWL_NONE=-0.0001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id hME4HESX686i for <hybi@ietfa.amsl.com>; Tue,  4 Feb 2014 10:11:23 -0800 (PST)
Received: from EXHUB020-1.exch020.serverdata.net (exhub020-1.exch020.serverdata.net [206.225.164.28]) by ietfa.amsl.com (Postfix) with ESMTP id 405AF1A01B3 for <hybi@ietf.org>; Tue,  4 Feb 2014 10:11:22 -0800 (PST)
Received: from EXVMBX020-12.exch020.serverdata.net ([169.254.3.11]) by EXHUB020-1.exch020.serverdata.net ([206.225.164.28]) with mapi; Tue, 4 Feb 2014 10:11:21 -0800
From: Tobias Oberstein <tobias.oberstein@tavendo.de>
To: Roberto Peon <fenix@google.com>
Date: Tue, 4 Feb 2014 10:11:20 -0800
Thread-Topic: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
Thread-Index: Ac8h0DCBX5QMtZVPS5iLvdFaOURFQwAAHuIA
Message-ID: <634914A010D0B943A035D226786325D4446C295789@EXVMBX020-12.exch020.serverdata.net>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <CABihn6GSrU8oJa4nbMJiDDo2Z+-c4fT69roVFxcHYM4XYFToRQ@mail.gmail.com> <F1F115D4-29EB-457B-B02C-71E46D3E1C81@zaphoyd.com> <CABihn6GXxUwHwLd+926VPEkZxHRZUUi-3dw0PoNaO6aK8X4sEQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446C1E8D2C@EXVMBX020-12.exch020.serverdata.net> <CAGzyod7pt_PzjexmnZj8dSMg3BooUmuMb6xB5P0sG1t6S5FXZw@mail.gmail.com> <634914A010D0B943A035D226786325D4446C1E925F@EXVMBX020-12.exch020.serverdata.net> <CABihn6GrDwu=74WEACTrd8RMwDe78G-Q3X+e_OUrYbPcFm8gZQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446C1E9735@EXVMBX020-12.exch020.serverdata.net> <CAGzyod6_9H=a+uejDTEd_6HfbD2e2ynkA2frCnhk0TboricgfQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446C2955D9@EXVMBX020-12.exch020.serverdata.net> <CAGzyod6H-To9EhaPqW5C0c9xyy5XnRg_851GuT3q=X+KoXe7Wg@mail.gmail.com>
In-Reply-To: <CAGzyod6H-To9EhaPqW5C0c9xyy5XnRg_851GuT3q=X+KoXe7Wg@mail.gmail.com>
Accept-Language: de-DE, en-US
Content-Language: de-DE
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
acceptlanguage: de-DE, en-US
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: base64
MIME-Version: 1.0
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 04 Feb 2014 18:11:29 -0000
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From webmaster@zaphoyd.com  Tue Feb  4 10:12:07 2014
Return-Path: <webmaster@zaphoyd.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 695751A0125 for <hybi@ietfa.amsl.com>; Tue,  4 Feb 2014 10:12:07 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.899
X-Spam-Level: 
X-Spam-Status: No, score=-1.899 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HTML_MESSAGE=0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 9ZTgQZ0JBEu4 for <hybi@ietfa.amsl.com>; Tue,  4 Feb 2014 10:11:58 -0800 (PST)
Received: from leaf101.mfilter.dimenoc.com (tree.mfilter.dimenoc.com [72.29.89.3]) by ietfa.amsl.com (Postfix) with ESMTP id B295B1A00EC for <hybi@ietf.org>; Tue,  4 Feb 2014 10:11:55 -0800 (PST)
Received: from localhost (localhost [127.0.0.1]) by leaf101.mfilter.dimenoc.com (Postfix) with ESMTP id 4460C6BC2D3 for <hybi@ietf.org>; Tue,  4 Feb 2014 13:11:51 -0500 (EST)
X-Virus-Scanned: amavisd-new at leaf.mfilter.dimenoc.com
X-Amavis-Alert: BAD HEADER SECTION, Header line longer than 998 characters: References: <CA[...]
Received: from leaf101.mfilter.dimenoc.com ([IPv6:::ffff:127.0.0.1]) by localhost (leaf101.mfilter.dimenoc.com [IPv6:::ffff:127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 15Ta5LlqJL9v for <hybi@ietf.org>; Tue,  4 Feb 2014 13:11:49 -0500 (EST)
Received: from sh78.surpasshosting.com (sh78.surpasshosting.com [72.29.64.142]) by leaf101.mfilter.dimenoc.com (Postfix) with ESMTP for <hybi@ietf.org>; Tue,  4 Feb 2014 13:11:49 -0500 (EST)
Received: from ranna.uchicago.edu ([128.135.45.206]:60271) by sh78.surpasshosting.com with esmtpsa (TLSv1:AES128-SHA:128) (Exim 4.80.1) (envelope-from <webmaster@zaphoyd.com>) id 1WAkTN-0005PX-Ow; Tue, 04 Feb 2014 13:11:50 -0500
Content-Type: multipart/alternative; boundary="Apple-Mail=_99DAE372-B9CB-4983-B36F-59AC98D83979"
Mime-Version: 1.0 (Mac OS X Mail 7.1 \(1827\))
From: Peter Thorson <webmaster@zaphoyd.com>
In-Reply-To: <CAGzyod6H-To9EhaPqW5C0c9xyy5XnRg_851GuT3q=X+KoXe7Wg@mail.gmail.com>
Date: Tue, 4 Feb 2014 12:11:42 -0600
Message-Id: <0533B942-C1DA-4BFE-BE80-4722C0BD740D@zaphoyd.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <CABihn6GSr...
To: Roberto Peon <fenix@google.com>
X-Mailer: Apple Mail (2.1827)
X-AntiAbuse: This header was added to track abuse, please include it with any abuse report
X-AntiAbuse: Primary Hostname - sh78.surpasshosting.com
X-AntiAbuse: Original Domain - ietf.org
X-AntiAbuse: Originator/Caller UID/GID - [47 12] / [47 12]
X-AntiAbuse: Sender Address Domain - zaphoyd.com
X-Get-Message-Sender-Via: sh78.surpasshosting.com: authenticated_id: webmaster@zaphoyd.com
X-Source: 
X-Source-Args: 
X-Source-Dir: 
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 04 Feb 2014 18:12:07 -0000

--Apple-Mail=_99DAE372-B9CB-4983-B36F-59AC98D83979
Content-Transfer-Encoding: quoted-printable
Content-Type: text/plain;
	charset=us-ascii


On Feb 4, 2014, at 11:40 , Roberto Peon <fenix@google.com> wrote:

> On Tue, Feb 4, 2014 at 7:02 AM, Tobias Oberstein =
<tobias.oberstein@tavendo.de> wrote:
> >I'm interested in correctly muxing, demuxing, and loadbalancing of =
streams-of-messages without needing to understand application-layer =
semantics.
>=20
> What is the unit of muxing/demuxing/loadbalancing?
>=20
> If the unit is the SPDY _stream_, then we can tunnel WebSocket with =
any extensions by simply transporting complete
> WebSocket frames inside the payload of SPDY data frames - with no =
flags whatsoever at SPDY header level.
>=20
> If the unit is the individual application _message_, then =
understanding of application message boundaries is required at SPDY =
level.
>=20
> Which unit did you think of?
>=20
> Messages, since I have to worry about other message-based protocols.
> =20
>=20
> >I'm not at all interested in needing to implement multiple separate =
protocols in order to implement the WS API.
>=20
> Sure, but it looks like a practical necessity. RFC6455 is out on the =
street and widely implemented.
>=20
> Well, I'm not interested in implementing it :)
> =20
>=20
> >I am uninterested in any extensions or extension frameworks that =
require any extension-specific knowledge in intermediaries.=20
>=20
> Not sure what this means ..
>=20
> WebSocket extensions are allowed to interleave messages explicitly in =
RFC6455. Intermediaries are not required
> to _understand_ the interleaving. They are required to _not touch_ the =
fragmentation in this case.
>=20
> A HTTP2.0 intermediary that does it's muxing/demuxing/loadbalancing at =
the SPDY stream level does not need
> to understand anything about tunneled WS.
>=20
> s/stream/message, and I agree.
> -=3DR

How do you mux, demux, or load balance an ordered message stream at the =
message level without taking into account application-layer semantics?=

--Apple-Mail=_99DAE372-B9CB-4983-B36F-59AC98D83979
Content-Transfer-Encoding: quoted-printable
Content-Type: text/html;
	charset=us-ascii

<html><head><meta http-equiv=3D"Content-Type" content=3D"text/html =
charset=3Dus-ascii"></head><body style=3D"word-wrap: break-word; =
-webkit-nbsp-mode: space; -webkit-line-break: =
after-white-space;"><br><div><div>On Feb 4, 2014, at 11:40 , Roberto =
Peon &lt;<a href=3D"mailto:fenix@google.com">fenix@google.com</a>&gt; =
wrote:</div><br class=3D"Apple-interchange-newline"><blockquote =
type=3D"cite"><div dir=3D"ltr" style=3D"font-family: Helvetica; =
font-size: 12px; font-style: normal; font-variant: normal; font-weight: =
normal; letter-spacing: normal; line-height: normal; orphans: auto; =
text-align: start; text-indent: 0px; text-transform: none; white-space: =
normal; widows: auto; word-spacing: 0px; -webkit-text-stroke-width: =
0px;"><div class=3D"gmail_extra">On Tue, Feb 4, 2014 at 7:02 AM, Tobias =
Oberstein&nbsp;<span dir=3D"ltr">&lt;<a =
href=3D"mailto:tobias.oberstein@tavendo.de" =
target=3D"_blank">tobias.oberstein@tavendo.de</a>&gt;</span>&nbsp;wrote:<b=
r><div class=3D"gmail_quote"><blockquote class=3D"gmail_quote" =
style=3D"margin: 0px 0px 0px 0.8ex; border-left-width: 1px; =
border-left-color: rgb(204, 204, 204); border-left-style: solid; =
padding-left: 1ex;"><div class=3D"im">&gt;I'm interested in correctly =
muxing, demuxing, and loadbalancing of streams-of-messages without =
needing to understand application-layer semantics.<br><br></div>What is =
the unit of muxing/demuxing/loadbalancing?<br><br>If the unit is the =
SPDY _stream_, then we can tunnel WebSocket with any extensions by =
simply transporting complete<br>WebSocket frames inside the payload of =
SPDY data frames - with no flags whatsoever at SPDY header =
level.<br><br>If the unit is the individual application _message_, then =
understanding of application message boundaries is required at SPDY =
level.<br><br>Which unit did you think =
of?<br></blockquote><div><br></div><div>Messages, since I have to worry =
about other message-based protocols.</div><div>&nbsp;</div><blockquote =
class=3D"gmail_quote" style=3D"margin: 0px 0px 0px 0.8ex; =
border-left-width: 1px; border-left-color: rgb(204, 204, 204); =
border-left-style: solid; padding-left: 1ex;"><div =
class=3D"im"><br>&gt;I'm not at all interested in needing to implement =
multiple separate protocols in order to implement the WS =
API.<br><br></div>Sure, but it looks like a practical necessity. RFC6455 =
is out on the street and widely =
implemented.<br></blockquote><div><br></div><div>Well, I'm not =
interested in implementing it :)</div><div>&nbsp;<br></div><blockquote =
class=3D"gmail_quote" style=3D"margin: 0px 0px 0px 0.8ex; =
border-left-width: 1px; border-left-color: rgb(204, 204, 204); =
border-left-style: solid; padding-left: 1ex;"><div class=3D"im"><br>&gt;I =
am uninterested in any extensions or extension frameworks that require =
any extension-specific knowledge in =
intermediaries.&nbsp;<br><br></div>Not sure what this means =
..<br><br>WebSocket extensions are allowed to interleave messages =
explicitly in RFC6455. Intermediaries are not required<br>to =
_understand_ the interleaving. They are required to _not touch_ the =
fragmentation in this case.<br><br>A HTTP2.0 intermediary that does it's =
muxing/demuxing/loadbalancing at the SPDY stream level does not =
need<br>to understand anything about tunneled =
WS.<br></blockquote><div><br></div><div>s/stream/message, and I =
agree.</div><div>-=3DR</div></div></div></div></blockquote><br></div><div>=
How do you mux, demux, or load balance an ordered message stream at the =
message level without taking into account application-layer =
semantics?</div></body></html>=

--Apple-Mail=_99DAE372-B9CB-4983-B36F-59AC98D83979--

From fenix@google.com  Tue Feb  4 10:44:15 2014
Return-Path: <fenix@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 354CA1A0035 for <hybi@ietfa.amsl.com>; Tue,  4 Feb 2014 10:44:15 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.913
X-Spam-Level: 
X-Spam-Status: No, score=-1.913 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.535, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id lPjItXmtP8ir for <hybi@ietfa.amsl.com>; Tue,  4 Feb 2014 10:44:12 -0800 (PST)
Received: from mail-qc0-x22d.google.com (mail-qc0-x22d.google.com [IPv6:2607:f8b0:400d:c01::22d]) by ietfa.amsl.com (Postfix) with ESMTP id DBBF41A00EC for <hybi@ietf.org>; Tue,  4 Feb 2014 10:44:11 -0800 (PST)
Received: by mail-qc0-f173.google.com with SMTP id i8so14339276qcq.32 for <hybi@ietf.org>; Tue, 04 Feb 2014 10:44:11 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=9fDFjV1ZLsPZ+k9uj/u0nG4RX7yy7z0dOe1d3jqa26s=; b=maa9/wYvse7OwO2W7/8lJ0JWrfGksym6GjmCNQ6izApxJfBoBoDJ8CkXmFa6KeFMIW zJKULAvTiv1R55BC85cPMUc1C0YAQioUjNOTU0nUL0mCeTUhnWe9estF4OG7DmxN/NAV A+wJzOuEHv3psHxhNpsIFZZNyydErKztUd7nNqsC9kazxhWZxL5TM6Qy+nvr6dNlkRlU J2fxn52PvuoMzEpBwdztQ/11opcpCESLjFcPB6hko0XKSMWwl6qhMCScFc0u1h0EsYxL 7x2ZeOApfyT4TBU+h9o99i72X9Pm8KVDFzp0gXk4MAWWYdk4iyWfIWoxkTgOd/F58mez eKfw==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=9fDFjV1ZLsPZ+k9uj/u0nG4RX7yy7z0dOe1d3jqa26s=; b=GydxOdCS+7eBzdM559r7ZeHK066HOEWi6O3sCzX62HCix0ix2izheXVUQdzxIIwAOW 4qdEqHV8GN3rDx+bj1fXIvLPiOlKPAUZP7zXsl/m1vSx9bpuIgNBPbmgRC7lyKgNQtzk XE5c2BjRnbmEpXiXo6w4+5zHHYrfGKR0lQ6A7Yfz7Mu39hRU0Z/QT8S0ukFE+6ZtP6Xi nZ5SenP2IttsYMkJiCoWM3xiU3V+PDhrSYfVxeSFQ763OrCXt1lCuFVAQjYWIjELRt5y xsIO0tTuOpWwbhByI/3TLjZse6FPVLXyiJUbKXuQjKxeZMgt98A3LwF5yaKR+Yva/UEf lcIA==
X-Gm-Message-State: ALoCoQnUHGJA95wWV0YMvsQKngyBtPpHCcK0xoOc/b68cojOrC/j2M611mNk6N8ZMFusgIavTWMXJOCbz2BPFd4h9UvkmHGtQ+DBizokjlPDN58QtOc96wVa49dSdyKDRn4InC4b99uadxP2cKkvRXQqkDsPzmI1VxrtQ0C3gg93R5rpwoRW9hSXow+q84mOJ3PyGIPmPJJY
MIME-Version: 1.0
X-Received: by 10.140.100.240 with SMTP id s103mr64315794qge.38.1391539451196;  Tue, 04 Feb 2014 10:44:11 -0800 (PST)
Received: by 10.229.13.7 with HTTP; Tue, 4 Feb 2014 10:44:10 -0800 (PST)
In-Reply-To: <0533B942-C1DA-4BFE-BE80-4722C0BD740D@zaphoyd.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <CAGzyod6H-To9EhaPqW5C0c9xyy5XnRg_851GuT3q=X+KoXe7Wg@mail.gmail.com> <0533B942-C1DA-4BFE-BE80-4722C0BD740D@zaphoyd.com>
Date: Tue, 4 Feb 2014 10:44:10 -0800
Message-ID: <CAGzyod7xtCZqUvcb0gX0PYKyikyd4kKc=kOzwzgfAmUFExCq3g@mail.gmail.com>
From: Roberto Peon <fenix@google.com>
To: Peter Thorson <webmaster@zaphoyd.com>
Content-Type: multipart/alternative; boundary=001a11c16cccae08ae04f199066d
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 04 Feb 2014 18:44:15 -0000

--001a11c16cccae08ae04f199066d
Content-Type: text/plain; charset=UTF-8

Note that the applications which use HTTP/WS define their own semantics on
top of (or sometimes replacing) that of the transport. I know of a number
of HTTP applications that enforce ordered messages, and I know of a few
usecases for streams-of-messages where the application would prefer
out-of-order delivery.

In any case, back to the thread:
Assuming that we're implementing a  proposal where END_MESSAGE delimits a
message boundary: to mux/demux/loadbalance any stream of messages all you
need to know is the application's policy for that stream of messages, since
END_MESSAGE allows one to differentiate between the different messages.
A loadbalancer can easily discover this policy at stream initialization
time by examining the path, scheme or any other interesting data from the
HEADERS fame which initialized the stream-of-messages, and it can do so
without any requirement to known anything more about the data being carried.

In other words, if a loadbalancer sees that the WS stream-of-messages is
targeted at an endpoint that can process out-of-order messages (by
examining the HEADERS frame initializing the stream-of-messages), then it
is free to demux the messages and forward without worrying about reordering
on the outbound path.

Note that property/mechanism would hold true for both HTTP and WS (and
other, e.g. RPC or API) traffic.

Tobias: We loadbalance myriad things, with RPC and API stuff being part of
the mix.
-=R


On Tue, Feb 4, 2014 at 10:11 AM, Peter Thorson <webmaster@zaphoyd.com>wrote:

>
> On Feb 4, 2014, at 11:40 , Roberto Peon <fenix@google.com> wrote:
>
> On Tue, Feb 4, 2014 at 7:02 AM, Tobias Oberstein <
> tobias.oberstein@tavendo.de> wrote:
>
>> >I'm interested in correctly muxing, demuxing, and loadbalancing of
>> streams-of-messages without needing to understand application-layer
>> semantics.
>>
>> What is the unit of muxing/demuxing/loadbalancing?
>>
>> If the unit is the SPDY _stream_, then we can tunnel WebSocket with any
>> extensions by simply transporting complete
>> WebSocket frames inside the payload of SPDY data frames - with no flags
>> whatsoever at SPDY header level.
>>
>> If the unit is the individual application _message_, then understanding
>> of application message boundaries is required at SPDY level.
>>
>> Which unit did you think of?
>>
>
> Messages, since I have to worry about other message-based protocols.
>
>
>>
>> >I'm not at all interested in needing to implement multiple separate
>> protocols in order to implement the WS API.
>>
>> Sure, but it looks like a practical necessity. RFC6455 is out on the
>> street and widely implemented.
>>
>
> Well, I'm not interested in implementing it :)
>
>
>>
>> >I am uninterested in any extensions or extension frameworks that require
>> any extension-specific knowledge in intermediaries.
>>
>> Not sure what this means ..
>>
>> WebSocket extensions are allowed to interleave messages explicitly in
>> RFC6455. Intermediaries are not required
>> to _understand_ the interleaving. They are required to _not touch_ the
>> fragmentation in this case.
>>
>> A HTTP2.0 intermediary that does it's muxing/demuxing/loadbalancing at
>> the SPDY stream level does not need
>> to understand anything about tunneled WS.
>>
>
> s/stream/message, and I agree.
> -=R
>
>
> How do you mux, demux, or load balance an ordered message stream at the
> message level without taking into account application-layer semantics?
>

--001a11c16cccae08ae04f199066d
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div>Note that the applications which use HTTP/WS define t=
heir own semantics on top of (or sometimes replacing) that of the transport=
. I know of a number of HTTP applications that enforce ordered messages, an=
d I know of a few usecases for streams-of-messages where the application wo=
uld prefer out-of-order delivery.</div>
<div><br></div><div>In any case, back to the thread:</div><div>Assuming tha=
t we&#39;re implementing a =C2=A0proposal where END_MESSAGE delimits a mess=
age boundary: to mux/demux/loadbalance any stream of messages all you need =
to know is the application&#39;s policy for that stream of messages, since =
END_MESSAGE allows one to differentiate between the different messages.</di=
v>
<div>A loadbalancer can easily discover this policy at stream initializatio=
n time by examining the path, scheme or any other interesting data from the=
 HEADERS fame which initialized the stream-of-messages, and it can do so wi=
thout any requirement to known anything more about the data being carried.<=
/div>
<div><br></div><div>In other words, if a loadbalancer sees that the WS stre=
am-of-messages is targeted at an endpoint that can process out-of-order mes=
sages (by examining the HEADERS frame initializing the stream-of-messages),=
 then it is free to demux the messages and forward without worrying about r=
eordering on the outbound path.</div>
<div><br></div><div>Note that property/mechanism would hold true for both H=
TTP and WS (and other, e.g. RPC or API) traffic.<br></div><div><br></div><d=
iv>Tobias: We loadbalance myriad things, with RPC and API stuff being part =
of the mix.</div>
<div>-=3DR</div><div class=3D"gmail_extra"><br><br><div class=3D"gmail_quot=
e">On Tue, Feb 4, 2014 at 10:11 AM, Peter Thorson <span dir=3D"ltr">&lt;<a =
href=3D"mailto:webmaster@zaphoyd.com" target=3D"_blank">webmaster@zaphoyd.c=
om</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div style=3D"word-wrap:break-word"><div cla=
ss=3D"im"><br><div><div>On Feb 4, 2014, at 11:40 , Roberto Peon &lt;<a href=
=3D"mailto:fenix@google.com" target=3D"_blank">fenix@google.com</a>&gt; wro=
te:</div>
<br><blockquote type=3D"cite"><div dir=3D"ltr" style=3D"font-family:Helveti=
ca;font-size:12px;font-style:normal;font-variant:normal;font-weight:normal;=
letter-spacing:normal;line-height:normal;text-align:start;text-indent:0px;t=
ext-transform:none;white-space:normal;word-spacing:0px">
<div class=3D"gmail_extra">On Tue, Feb 4, 2014 at 7:02 AM, Tobias Oberstein=
=C2=A0<span dir=3D"ltr">&lt;<a href=3D"mailto:tobias.oberstein@tavendo.de" =
target=3D"_blank">tobias.oberstein@tavendo.de</a>&gt;</span>=C2=A0wrote:<br=
><div class=3D"gmail_quote">
<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div>&gt;I&#39;m interested in correctly muxing, demuxing,=
 and loadbalancing of streams-of-messages without needing to understand app=
lication-layer semantics.<br>
<br></div>What is the unit of muxing/demuxing/loadbalancing?<br><br>If the =
unit is the SPDY _stream_, then we can tunnel WebSocket with any extensions=
 by simply transporting complete<br>WebSocket frames inside the payload of =
SPDY data frames - with no flags whatsoever at SPDY header level.<br>
<br>If the unit is the individual application _message_, then understanding=
 of application message boundaries is required at SPDY level.<br><br>Which =
unit did you think of?<br></blockquote><div><br></div><div>Messages, since =
I have to worry about other message-based protocols.</div>
<div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px =
0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-l=
eft-style:solid;padding-left:1ex"><div><br>&gt;I&#39;m not at all intereste=
d in needing to implement multiple separate protocols in order to implement=
 the WS API.<br>
<br></div>Sure, but it looks like a practical necessity. RFC6455 is out on =
the street and widely implemented.<br></blockquote><div><br></div><div>Well=
, I&#39;m not interested in implementing it :)</div><div>=C2=A0<br></div><b=
lockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-le=
ft-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;pad=
ding-left:1ex">
<div><br>&gt;I am uninterested in any extensions or extension frameworks th=
at require any extension-specific knowledge in intermediaries.=C2=A0<br><br=
></div>Not sure what this means ..<br><br>WebSocket extensions are allowed =
to interleave messages explicitly in RFC6455. Intermediaries are not requir=
ed<br>
to _understand_ the interleaving. They are required to _not touch_ the frag=
mentation in this case.<br><br>A HTTP2.0 intermediary that does it&#39;s mu=
xing/demuxing/loadbalancing at the SPDY stream level does not need<br>to un=
derstand anything about tunneled WS.<br>
</blockquote><div><br></div><div>s/stream/message, and I agree.</div><div>-=
=3DR</div></div></div></div></blockquote><br></div></div><div>How do you mu=
x, demux, or load balance an ordered message stream at the message level wi=
thout taking into account application-layer semantics?</div>
</div></blockquote></div><br></div></div>

--001a11c16cccae08ae04f199066d--

From tobias.oberstein@tavendo.de  Tue Feb  4 11:44:33 2014
Return-Path: <tobias.oberstein@tavendo.de>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 6172A1A01D2 for <hybi@ietfa.amsl.com>; Tue,  4 Feb 2014 11:44:33 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.9
X-Spam-Level: 
X-Spam-Status: No, score=-1.9 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, RCVD_IN_DNSWL_NONE=-0.0001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id ukQ07YIG0Edz for <hybi@ietfa.amsl.com>; Tue,  4 Feb 2014 11:44:31 -0800 (PST)
Received: from EXHUB020-3.exch020.serverdata.net (exhub020-3.exch020.serverdata.net [206.225.164.30]) by ietfa.amsl.com (Postfix) with ESMTP id 42A781A0127 for <hybi@ietf.org>; Tue,  4 Feb 2014 11:44:31 -0800 (PST)
Received: from EXVMBX020-12.exch020.serverdata.net ([169.254.3.11]) by EXHUB020-3.exch020.serverdata.net ([206.225.164.30]) with mapi; Tue, 4 Feb 2014 11:44:30 -0800
From: Tobias Oberstein <tobias.oberstein@tavendo.de>
To: Roberto Peon <fenix@google.com>, Peter Thorson <webmaster@zaphoyd.com>
Date: Tue, 4 Feb 2014 11:44:18 -0800
Thread-Topic: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
Thread-Index: Ac8h2RkLMvmbyBqwRNamxnBDm+BlZgABAnOg
Message-ID: <634914A010D0B943A035D226786325D4446C29584C@EXVMBX020-12.exch020.serverdata.net>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <CAGzyod6H-To9EhaPqW5C0c9xyy5XnRg_851GuT3q=X+KoXe7Wg@mail.gmail.com> <0533B942-C1DA-4BFE-BE80-4722C0BD740D@zaphoyd.com> <CAGzyod7xtCZqUvcb0gX0PYKyikyd4kKc=kOzwzgfAmUFExCq3g@mail.gmail.com>
In-Reply-To: <CAGzyod7xtCZqUvcb0gX0PYKyikyd4kKc=kOzwzgfAmUFExCq3g@mail.gmail.com>
Accept-Language: de-DE, en-US
Content-Language: de-DE
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
acceptlanguage: de-DE, en-US
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: base64
MIME-Version: 1.0
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 04 Feb 2014 19:44:33 -0000
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From yhirano@google.com  Wed Feb  5 00:27:01 2014
Return-Path: <yhirano@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id DD3D51A0063 for <hybi@ietfa.amsl.com>; Wed,  5 Feb 2014 00:27:01 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.913
X-Spam-Level: 
X-Spam-Status: No, score=-1.913 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.535, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id H0p0y1ipOWvy for <hybi@ietfa.amsl.com>; Wed,  5 Feb 2014 00:26:57 -0800 (PST)
Received: from mail-yh0-x22e.google.com (mail-yh0-x22e.google.com [IPv6:2607:f8b0:4002:c01::22e]) by ietfa.amsl.com (Postfix) with ESMTP id 1B8001A00A3 for <hybi@ietf.org>; Wed,  5 Feb 2014 00:26:57 -0800 (PST)
Received: by mail-yh0-f46.google.com with SMTP id v1so41535yhn.33 for <hybi@ietf.org>; Wed, 05 Feb 2014 00:26:56 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=M+HGAhnlbJfz8JDBN8gQ5Ddjzx7GryKEZDhl0iWc64E=; b=a6R4UBxxHnILQc2t26qFi7yD/7O8WDW2JyHXNhkCgzxjdlh5yOkXIOpYt2b99n+Q56 ggZ7Bdd49kKAyfUy3oFO4ylVMO6egcNkWJ6LWlDfcIoLCGIRVMrsxiF/GoDlteW+gaTQ fjX5g+8XzTPrRvx/XMnLiM+YYxUyE7kepoiSYRPvFd7av1AdulKd6CagkaEfb7YBJp3C JUyJ2cpb5DiXt+2NWvILm5Kk222rP4PyL1VotKCnfeHw8gk7XxXCdvZ93TzWtoxgFWbP gztd9Gb+TA6HF6nhIUCrF/g5lHuSw7+UbXcgvZ8US33G348TAAVE2dw5/mPKzQokO4e4 UrhA==
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=chromium.org; s=google; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=M+HGAhnlbJfz8JDBN8gQ5Ddjzx7GryKEZDhl0iWc64E=; b=L2RSvPw1B9a2/8+ecxQ6NC0HZQmJy6zvjVbZ0pTeKRJpZeYRWn2dlU9eDVCrU7N0CC QFhmJAT7I1KkGoNCatgUAzOOCB/uBgpPk1MVNAOWD/7cvNlQuepjnn4u9M465emuRCXs 5KOy4MZsOemKfXTAc0CYNa72IBEoLLyrJ92aY=
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:sender:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=M+HGAhnlbJfz8JDBN8gQ5Ddjzx7GryKEZDhl0iWc64E=; b=jNV5+NM7ptjSWwjIgzQYYs+YJxp/7YiouXhnTqSyPu8MTW0IaIpf8QVMF0cCrAclyx cl7ZO+NV0ncJM1M3cu42ZcQbucoxrU18bHIAd9mHWbUWDKC9kvUXJq71EnSOR2Ul5fAy S/FEnYY4roDTjw+NZGcQY4VTclKhyJdnLhipQ6CWkUHltHnAtCTZU/ZqN3ScI3hNozS/ 0GUEKiZi8X2SdhDHXYEBRLus1WxQbeojBARIYRTOIYbRBafHPgJvcK8pm+jEd2e/nl93 BACYZGr7dCcELi+ZzLoFEvntUFlCgDH50/+a/Hj35lo0MFrWAMLhPy6VhMTpTuZf8Bpg rDZQ==
X-Gm-Message-State: ALoCoQlTA+J4NWScDAymu8s02tX3xhM/nRtO6OSD4WouQTw3rPjyoASQyxRfKyCiHqFMY/FXymH1AyiFWj3tR6aS9nI/SJ99TFLF2/kOpiwTJ9Pzn6fV7y36tH7vXEJxi8eXa2YH3+EIhkjTHvFS1QawNogD39TOMsDSGnntFEXSIf3TvEc25hPbp+MSkTb8rs0YWNp2mS9D
MIME-Version: 1.0
X-Received: by 10.236.183.111 with SMTP id p75mr125251yhm.148.1391588816145; Wed, 05 Feb 2014 00:26:56 -0800 (PST)
Sender: yhirano@google.com
Received: by 10.170.135.22 with HTTP; Wed, 5 Feb 2014 00:26:56 -0800 (PST)
In-Reply-To: <634914A010D0B943A035D226786325D4446C2954EF@EXVMBX020-12.exch020.serverdata.net>
References: <CABihn6EtrUQvCjRXnjmEYHq_3z4KrPzX5e_JeRev0mAX-Od+fQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446C1E9261@EXVMBX020-12.exch020.serverdata.net> <CABihn6HAiNeVNoSFsHqMo_pccxwbWs26kFFXysOGxT+yOCcrxg@mail.gmail.com> <634914A010D0B943A035D226786325D4446C2954EF@EXVMBX020-12.exch020.serverdata.net>
Date: Wed, 5 Feb 2014 17:26:56 +0900
X-Google-Sender-Auth: 5R0OmDTRZ3ii1b1Mu8gRtnpCjrE
Message-ID: <CABihn6GtddoGOaOt5f0QWJCwe+ZsGit35QOCDjvPJ8m_LZX8rg@mail.gmail.com>
From: Yutaka Hirano <yhirano@chromium.org>
To: Tobias Oberstein <tobias.oberstein@tavendo.de>
Content-Type: multipart/alternative; boundary=bcaec52d55270f611b04f1a485bc
Cc: "hybi@ietf.org" <hybi@ietf.org>, "agl@chromium.org" <agl@chromium.org>
Subject: Re: [hybi] WebSocket over HTTP/2.0 protocol negotiation
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 05 Feb 2014 08:27:02 -0000

--bcaec52d55270f611b04f1a485bc
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

Adam pointed out that HTTP/2.0[ws] isn't useful since a client doesn't use
ALPN when it wants to use non-secure WebSocket.
He is right, I will fix the description.

1)
> What if the server does not support TLS ALPN extension?
>
> IOW: client sends ALPN with "HTTP/1.1, HTTP2.0[ws], HTTP2.0[wss]", and
> server answers
> with no TLS ALPN extension part (or no next protocol selected).
>
> Will the client then default to trying RFC6455 opening handshake
> (HTTP1.1/Upgrade)?
>
>
If there is no TLS ALPN extension part, yes.
If "no_application_protocol" alert is returned, I think the client should
give up.


> 2)
> "If the client has an existing HTTP/2.0 session and it knows that the
> server supports WebSocket over HTTP/2.0 =E2=80=A6"
>
> How does the client become aware of the server's support for WS-over-HTTP=
2?
>
> Is that via the HTTP/2.0 SETTINGS frame you mention below?
>
>
Yes. In addition to that, when there is an existing WebSocket over HTTP/2.0
connection to the server, even if without a SETTINGS frame the client knows
that the server supports the protocol.


>
> 3)
> If the client _already_ knows the server supports HTTP/WS, what ALPN will
> it send
> if no HTTP2 session is currently in place?
>
> Will it nevertheless send "HTTP/1.1, HTTP2.0[ws], HTTP2.0[wss]" (same
> order)?
>
> Or will it send "HTTP2.0[ws], HTTP2.0[wss]" only?
>
> Or will it send "HTTP2.0[ws], HTTP2.0[wss], HTTP/1.1" (reordered)?
>
Same order looks good.


> 4) ALPN name for native RFC6455: wouldn't "HTTP/1.1[wss]" better
> correspond to
> the actual protocol the client offers / desires to speak?
>
> The client offers to speak RFC6455, just upgrading from HTTP/1.1, and the
> server
> might want to select a dedicated implementation for that.
>
> In general, a server that supports TLS ALPN might just want to lookup a
> preferred
> implementation given a list of ALPN names offered by a client.
>
It sounds reasonable, but

ALPN "HTTP/1.1": client will _never_ HTTP1.1/Upgrade (to RFC6455 or
> anything other)
>
I think this is beyond our interest, isn't it?


> ALPN "HTTP/1.1[wss]": client will _immediately_ HTTP1.1/Upgrade to RFC645=
5
>
I think this is sufficient.

Thanks,




On Tue, Feb 4, 2014 at 7:32 PM, Tobias Oberstein <
tobias.oberstein@tavendo.de> wrote:

> Hi Yutaka,
>
>
>
> "No additional RTTs are introduced when the server supports the native
> WebSocket and HTTP/2.0 but does not support WebSocket over HTTP/2.0."
>
>
>
> Awesome! The negotiation scheme you outline below seems to provide this.
>
>
>
> Just some small detail questions:
>
>
>
> 1)
>
> What if the server does not support TLS ALPN extension?
>
>
>
> IOW: client sends ALPN with "HTTP/1.1, HTTP2.0[ws], HTTP2.0[wss]", and
> server answers
>
> with no TLS ALPN extension part (or no next protocol selected).
>
>
>
> Will the client then default to trying RFC6455 opening handshake
> (HTTP1.1/Upgrade)?
>
>
>
>
>
> 2)
>
> "If the client has an existing HTTP/2.0 session and it knows that the
> server supports WebSocket over HTTP/2.0 =E2=80=A6"
>
>
>
> How does the client become aware of the server's support for WS-over-HTTP=
2?
>
>
>
> Is that via the HTTP/2.0 SETTINGS frame you mention below?
>
>
>
>
>
> 3)
>
> If the client _already_ knows the server supports HTTP/WS, what ALPN will
> it send
>
> if no HTTP2 session is currently in place?
>
>
>
> Will it nevertheless send "HTTP/1.1, HTTP2.0[ws], HTTP2.0[wss]" (same
> order)?
>
>
>
> Or will it send "HTTP2.0[ws], HTTP2.0[wss]" only?
>
>
>
> Or will it send "HTTP2.0[ws], HTTP2.0[wss], HTTP/1.1" (reordered)?
>
>
>
>
>
> 4) ALPN name for native RFC6455: wouldn't "HTTP/1.1[wss]" better
> correspond to
>
> the actual protocol the client offers / desires to speak?
>
>
>
> The client offers to speak RFC6455, just upgrading from HTTP/1.1, and the
> server
>
> might want to select a dedicated implementation for that.
>
>
>
> In general, a server that supports TLS ALPN might just want to lookup a
> preferred
>
> implementation given a list of ALPN names offered by a client.
>
>
>
> Example:
>
> Take a server that implements HTTP/1.1 and HTTP/2.0 for serving static We=
b
> content,
>
> and RFC6455 for application messaging (all services on same ports), and
> that wants
>
> to realize the following behavior for TLS:
>
>
>
> no ALPN & ALPNs "HTTP/1.1":
>
>   =3D> Server choses HTTP/1.1 implementation, no later upgrade to RFC6455=
 is
> allowed
>
>
>
> ALPNs "HTTP/2.0":
>
>   =3D> Server choses HTTP/2.0 implementation (without announcing HTTP/WS
> support)
>
>
>
> ALPNs "HTTP/1.1[wss],HTTP/2.0[wss]":
>
>   =3D> server choses RFC6455 implementation requiring immediate upgrade
> (that is, no
>
> "plain" HTTP/GET requests before HTTP/Upgrade to WS are allowed)
>
>
>
> IOW: the server wants to adjust it's behavior with ALPNs "HTTP/1.1" versu=
s
> "HTTP/1.1[wss]":
>
>
>
> ALPN "HTTP/1.1": client will _never_ HTTP1.1/Upgrade (to RFC6455 or
> anything other)
>
> ALPN "HTTP/1.1[wss]": client will _immediately_ HTTP1.1/Upgrade to RFC645=
5
>
>
>
>
>
> Cheers,
>
> /Tobias
>
>
>
>
>
> *Von:* yhirano@google.com [mailto:yhirano@google.com] *Im Auftrag von *Yu=
taka
> Hirano
> *Gesendet:* Dienstag, 4. Februar 2014 08:17
> *An:* Tobias Oberstein
> *Cc:* hybi@ietf.org; agl@chromium.org
> *Betreff:* Re: [hybi] WebSocket over HTTP/2.0 protocol negotiation
>
>
>
> Hi Tobias,
>
>
>
> In other words: I think a server should actively announce support for any
> SPDY/WS, and only then
> should SPDY/WS be tried first when a JS opens a WebSocket connection to a=
n
> origin that has an
> existing SPDY connection and which already announced it's support and
> willingness to speak
> any SPDY/WS.
>
> I see. I propose a new negotiation method here and I think it addresses
> your concern.
>
>
>
> *Negotiation Decision Process:*
>
> 1: If the client has an existing HTTP/2.0 session and it knows that the
> server supports WebSocket over HTTP/2.0, create a WebSocket stream on the
> session. Both of WebSocket (ws) and secure WebSocket (wss) could share th=
e
> same HTTP/2.0 session.
>
> 2: If the client is going to establish a secure WebSocket connection,
> create a WebSocket connection over TLS.
>
> 3: If the client is going to establish a non-secure WebSocket connection,
> try the WebSocket protocol negotiation.
>
>
>
> *Negotiation Cases:*
>
>
>
> *Create WebSocket Stream on Existing HTTP/2.0 Session*
>
> The client SHOULD create a new HTTP/2.0 stream in the existing HTTP/2.0
> session.
>
> *Create WebSocket Connection over TLS*
>
> HTTP/2.0 uses TLS Application Level Protocol Negotiation (ALPN) for the
> protocol negotiation. The client sends the application protocols it wants
> to use, "HTTP/1.1" for the native WebSocket over TLS, "HTTP/2.0[ws]" for
> non secure WebSocket over HTTP/2.0, "HTTP/2.0[wss]" for secure WebSocket
> over HTTP/2.0.
>
> If the server selects "HTTP/1.1", the client SHOULD start the native
> WebSocket opening handshake on TLS.
>
> If the server selects "HTTP/2.0[ws]", the client SHOULD create an HTTP/2.=
0
> session and start the WebSocket opening handshake on it with "ws" scheme.
>
> If the server selects "HTTP/2.0[wss]", the client SHOULD create an
> HTTP/2.0 session and start the WebSocket opening handshake on it with "ws=
s"
> scheme.
>
>
>
> *The WebSocket Protocol Negotiation*
>
> (same as the document)
>
>
>
> *Supporting scheme negotiation using SETTINGS frame*
>
> By default, a client can send HEADERS frames on schemes that server
> supports based on the protocol server selected in ALPN. A server SHOULD
> send a SETTINGS frame as soon as possible whether it supports other schem=
es
> or not.
>
> Here is a proposed ID and value which should be defined in the HTTP/2.0
> specification. If the endpoint received a HEADERS on unsupporting schemes=
,
> the endpoint MUST reset the stream.
>
>
>
> Defined IDs: (same as the document)
>
>
> No additional RTTs are introduced when the server supports the native
> WebSocket and HTTP/2.0 but does not support WebSocket over HTTP/2.0.
>
> We should introduce the following in the HTTP/2.0 specification.
>
>  - Supporting scheme advertisement using SETTINGS frame
>
>  - HTTP/2.0[xx] in the ALPN stands for the HTTP/2.0 protocol with scheme
> xx.
>
>
> Thanks,
>
>
>
>
>
> On Thu, Jan 30, 2014 at 9:24 PM, Tobias Oberstein <
> tobias.oberstein@tavendo.de> wrote:
>
> Hi Yutaka,
>
>
>
> the point I was trying to raise is this:
>
>
>
> For a JavaScript in a browser that opens a "ws://=E2=80=A6" connection, t=
here
> should never be an additional roundtrip
>
> incurred by first probing for any SPDY/WS and then falling back to native
> WS, but the native WS should be
>
> tried first.
>
>
>
> In other words: the current, immediate connection establishment for nativ=
e
> WS should not be put
>
> at an disadvantage just because of introducing any SPDY/WS.
>
>
>
> A server should be able to provide both SPDY and native WebSocket on the
> _same_ origin without
>
> being required to implement SPDY/WS also, and without incurring an
> additional roundtrip for
>
> probing SPDY/WS while establishing a WS connection.
>
>
>
> In other words: I think a server should actively announce support for any
> SPDY/WS, and only then
>
> should SPDY/WS be tried first when a JS opens a WebSocket connection to a=
n
> origin that has an
>
> existing SPDY connection and which already announced it's support and
> willingness to speak
>
> any SPDY/WS.
>
>
>
> Probably I misunderstood the draft.
>
>
>
> Are browsers supposed to work like above when they implement any SPDY/WS?
>
>
>
> /Tobias
>
>
>
>
>
> *Von:* hybi [mailto:hybi-bounces@ietf.org] *Im Auftrag von *Yutaka Hirano
> *Gesendet:* Donnerstag, 30. Januar 2014 07:45
> *An:* hybi@ietf.org
> *Betreff:* [hybi] WebSocket over HTTP/2.0 protocol negotiation
>
>
>
> Hi,
>
>
>
> As I announced, I've updated the draft and requested for comments.
>
> Tobias gave a comment on protocol negotiation (thanks!) and I think it is
> good to discuss here, not at the draft comment fields.
>
>
>
> The draft says:
>
>    1. If the client has a HTTP/2.0 session established by previous
>    requests and it doesn=E2=80=99t know that the session does NOT support=
 WebSocket
>    layering over HTTP/2.0, create a WebSocket stream on the session. Both=
 of
>    WebSocket (ws) and secure WebSocket (wss) could share the same HTTP/2.=
0
>    session.
>    2. If the client is going to establish a secure WebSocket connection,
>    create a WebSocket connection over TLS.
>    3. If the client is going to establish a non-secure WebSocket
>    connection, try the WebSocket protocol negotiation.
>
> Tobias doesn't like the first step. He says:
>
> I think native WS should be tried first. Native WS has less overhead than
> any WS-over-SPDY layering, and hence should be preferred.
>
>
>
> I have some comments about it.
>
>  - WebSocket over HTTP/2.0 is multiplexed, so it is not so obvious which
> is preferable. I think it depends on the popularity of the services:
>
>    - WebSocket over HTTP/2.0
>
>    - Native WebSocket and HTTP/2.0
>
>    - Native WebSocket and WebSocket over HTTP/2.0
>
>    Note that servers which doesn't support HTTP/2.0 don't matter, because
> they don't meet the precondition of the first step.
>
>  -  There are two scheme negotiation hints described below.
>
>     If we choose Hint X, before receiving a SETTINGS frame from the
> server, the client doesn't know which service is preferable. It is rare,
> but trying connecting to the wrong service can occur.
>
>     If we choose Hint Y, whether the server supports WebSocket over
> HTTP/2.0 or not is detected by the client prior to HTTP/2.0 session
> establishment and hence the client always understands the server's
> preference.
>
>     So, the actual negotiation decision process depends on the hint we
> choose.
>
>
>
> I'm sorry that the draft is not clear at this point, but it will be when
> the hint is chosen.
>
>
>
> Thanks,
>
>
>
>
>

--bcaec52d55270f611b04f1a485bc
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Adam pointed out that HTTP/2.0[ws] isn&#39;t useful since =
a client doesn&#39;t use ALPN when it wants to use non-secure WebSocket.<di=
v>He is right, I will fix the description.</div><div><br></div><div><blockq=
uote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-wi=
dth:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-=
left:1ex">
1)<br>What if the server does not support TLS ALPN extension?<br>=C2=A0<br>=
IOW: client sends ALPN with &quot;HTTP/1.1, HTTP2.0[ws], HTTP2.0[wss]&quot;=
, and server answers<br>with no TLS ALPN extension part (or no next protoco=
l selected).<br>
=C2=A0<br>Will the client then default to trying RFC6455 opening handshake =
(HTTP1.1/Upgrade)?<br>=C2=A0<br></blockquote><div>If there is no TLS ALPN e=
xtension part, yes.</div><div>If=C2=A0<span style=3D"color:rgb(0,0,0);font-=
size:1em">&quot;no_application_protocol&quot; alert is returned</span>, I t=
hink the client should give up.=C2=A0</div>
<div><br></div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0p=
x 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-lef=
t-style:solid;padding-left:1ex">=C2=A0<br>2)<br>&quot;If the client has an =
existing HTTP/2.0 session and it knows that the server supports WebSocket o=
ver HTTP/2.0 =E2=80=A6&quot;<br>
=C2=A0<br>How does the client become aware of the server&#39;s support for =
WS-over-HTTP2?<br>=C2=A0<br>Is that via the HTTP/2.0 SETTINGS frame you men=
tion below?<br>=C2=A0<br></blockquote><div>Yes. In addition to that, when t=
here is an existing WebSocket over HTTP/2.0 connection to the server, even =
if without a SETTINGS frame the client knows that the server supports the p=
rotocol.</div>
<div>=C2=A0<br></div><blockquote class=3D"gmail_quote" style=3D"margin:0px =
0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);bord=
er-left-style:solid;padding-left:1ex">=C2=A0<br>3)<br>If the client _alread=
y_ knows the server supports HTTP/WS, what ALPN will it send<br>
if no HTTP2 session is currently in place?<br>=C2=A0<br>Will it nevertheles=
s send &quot;HTTP/1.1, HTTP2.0[ws], HTTP2.0[wss]&quot; (same order)?<br>=C2=
=A0<br>Or will it send &quot;HTTP2.0[ws], HTTP2.0[wss]&quot; only?<br>=C2=
=A0<br>Or will it send &quot;HTTP2.0[ws], HTTP2.0[wss], HTTP/1.1&quot; (reo=
rdered)?<br>
</blockquote><div>Same order looks good.</div><div>=C2=A0<br></div><blockqu=
ote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-wid=
th:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-l=
eft:1ex">
4) ALPN name for native RFC6455: wouldn&#39;t &quot;HTTP/1.1[wss]&quot; bet=
ter correspond to<br>the actual protocol the client offers / desires to spe=
ak?<br>=C2=A0<br>The client offers to speak RFC6455, just upgrading from HT=
TP/1.1, and the server<br>
might want to select a dedicated implementation for that.<br>=C2=A0<br>In g=
eneral, a server that supports TLS ALPN might just want to lookup a preferr=
ed<br>implementation given a list of ALPN names offered by a client.=C2=A0<=
br></blockquote>
<div>It sounds reasonable, but</div><div><br></div><blockquote class=3D"gma=
il_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-le=
ft-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">ALPN &q=
uot;HTTP/1.1&quot;: client will _never_ HTTP1.1/Upgrade (to RFC6455 or anyt=
hing other)<br>
</blockquote><div>I think this is beyond our interest, isn&#39;t it?</div><=
div>=C2=A0<br></div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0=
px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);borde=
r-left-style:solid;padding-left:1ex">
ALPN &quot;HTTP/1.1[wss]&quot;: client will _immediately_ HTTP1.1/Upgrade t=
o RFC6455<br></blockquote><div>I think this is sufficient.=C2=A0</div><div>=
<br></div><div>Thanks,</div><div>=C2=A0</div><p class=3D"MsoNormal" style=
=3D"font-family:arial,sans-serif;font-size:13px">
<u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,sans-serif;fon=
t-size:13px"><u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,s=
ans-serif;font-size:13px"><u></u><u></u></p><p class=3D"MsoNormal" style=3D=
"font-family:arial,sans-serif;font-size:13px">
<u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,sans-serif;fon=
t-size:13px"><u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,s=
ans-serif;font-size:13px"><u></u><u></u></p><p class=3D"MsoNormal" style=3D=
"font-family:arial,sans-serif;font-size:13px">
<u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,sans-serif;fon=
t-size:13px"><u></u><u></u></p><p class=3D"MsoNormal" style=3D"font-family:=
arial,sans-serif;font-size:13px"><u></u><u></u></p><p class=3D"MsoNormal" s=
tyle=3D"font-family:arial,sans-serif;font-size:13px">
<u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,sans-serif;fon=
t-size:13px"><u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,s=
ans-serif;font-size:13px"><u></u><u></u></p><p class=3D"MsoNormal" style=3D=
"font-family:arial,sans-serif;font-size:13px">
<u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,sans-serif;fon=
t-size:13px"><u></u><u></u></p><p class=3D"MsoNormal" style=3D"font-family:=
arial,sans-serif;font-size:13px"><u></u></p><p class=3D"MsoNormal" style=3D=
"font-family:arial,sans-serif;font-size:13px">
<u></u><u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,sans-se=
rif;font-size:13px"><u></u><u></u></p><p class=3D"MsoNormal" style=3D"font-=
family:arial,sans-serif;font-size:13px"><u></u></p><p class=3D"MsoNormal" s=
tyle=3D"font-family:arial,sans-serif;font-size:13px">
<u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,sans-serif;fon=
t-size:13px"><u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,s=
ans-serif;font-size:13px"><u></u><u></u></p><p class=3D"MsoNormal" style=3D=
"font-family:arial,sans-serif;font-size:13px">
<u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,sans-serif;fon=
t-size:13px"><u></u><u></u></p><p class=3D"MsoNormal" style=3D"font-family:=
arial,sans-serif;font-size:13px"><u></u></p><p class=3D"MsoNormal" style=3D=
"font-family:arial,sans-serif;font-size:13px">
<u></u><u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,sans-se=
rif;font-size:13px"><u></u></p><p class=3D"MsoNormal" style=3D"font-family:=
arial,sans-serif;font-size:13px"><u></u><u></u></p><p class=3D"MsoNormal" s=
tyle=3D"font-family:arial,sans-serif;font-size:13px">
<u></u><u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,sans-se=
rif;font-size:13px"><u></u></p><p class=3D"MsoNormal" style=3D"font-family:=
arial,sans-serif;font-size:13px"><u></u></p><p class=3D"MsoNormal" style=3D=
"font-family:arial,sans-serif;font-size:13px">
<u></u><u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,sans-se=
rif;font-size:13px"><u></u></p><p class=3D"MsoNormal" style=3D"font-family:=
arial,sans-serif;font-size:13px"><u></u></p><p class=3D"MsoNormal" style=3D=
"font-family:arial,sans-serif;font-size:13px">
<u></u><u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,sans-se=
rif;font-size:13px"><u></u></p><p class=3D"MsoNormal" style=3D"font-family:=
arial,sans-serif;font-size:13px"><u></u></p><p class=3D"MsoNormal" style=3D=
"font-family:arial,sans-serif;font-size:13px">
<u></u><u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,sans-se=
rif;font-size:13px"><u></u></p><p class=3D"MsoNormal" style=3D"font-family:=
arial,sans-serif;font-size:13px"><u></u></p><p class=3D"MsoNormal" style=3D=
"font-family:arial,sans-serif;font-size:13px">
<u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,sans-serif;fon=
t-size:13px"><u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,s=
ans-serif;font-size:13px"><u></u><u></u></p><p class=3D"MsoNormal" style=3D=
"font-family:arial,sans-serif;font-size:13px">
<u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,sans-serif;fon=
t-size:13px"><u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,s=
ans-serif;font-size:13px"><u></u><u></u></p><p class=3D"MsoNormal" style=3D=
"font-family:arial,sans-serif;font-size:13px">
<u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,sans-serif;fon=
t-size:13px"><u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,s=
ans-serif;font-size:13px"><u></u><u></u></p><p class=3D"MsoNormal" style=3D=
"font-family:arial,sans-serif;font-size:13px">
<u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,sans-serif;fon=
t-size:13px"><u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,s=
ans-serif;font-size:13px"><u></u></p><p class=3D"MsoNormal" style=3D"font-f=
amily:arial,sans-serif;font-size:13px">
<u></u><u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,sans-se=
rif;font-size:13px"><u></u></p><p class=3D"MsoNormal" style=3D"font-family:=
arial,sans-serif;font-size:13px"><u></u><u></u></p><p class=3D"MsoNormal" s=
tyle=3D"font-family:arial,sans-serif;font-size:13px">
<u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,sans-serif;fon=
t-size:13px"><u></u></p><p class=3D"MsoNormal" style=3D"font-family:arial,s=
ans-serif;font-size:13px"><u></u></p></div></div><div class=3D"gmail_extra"=
><br><br>
<div class=3D"gmail_quote">On Tue, Feb 4, 2014 at 7:32 PM, Tobias Oberstein=
 <span dir=3D"ltr">&lt;<a href=3D"mailto:tobias.oberstein@tavendo.de" targe=
t=3D"_blank">tobias.oberstein@tavendo.de</a>&gt;</span> wrote:<br><blockquo=
te class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc so=
lid;padding-left:1ex">
<div lang=3D"DE" link=3D"blue" vlink=3D"purple"><div><p class=3D"MsoNormal"=
><span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans=
-serif&quot;;color:#1f497d">Hi Yutaka,<u></u><u></u></span></p><div class=
=3D"im"><p class=3D"MsoNormal">
<span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-=
serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></span></p><p class=3D"MsoNo=
rmal"><span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot=
;sans-serif&quot;;color:#1f497d">&quot;</span>No additional RTTs are introd=
uced when the server supports the native WebSocket and HTTP/2.0 but does no=
t support WebSocket over HTTP/2.0.&quot;<u></u><u></u></p>
<p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div><p class=3D"MsoNormal"=
>Awesome! The negotiation scheme you outline below seems to provide this.<u=
></u><u></u></p><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p><p class=3D"=
MsoNormal">Just some small detail questions:<u></u><u></u></p>
<p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p><p class=3D"MsoNormal">1)<u>=
</u><u></u></p><p class=3D"MsoNormal">What if the server does not support T=
LS ALPN extension?<u></u><u></u></p><p class=3D"MsoNormal"><u></u>=C2=A0<u>=
</u></p><p class=3D"MsoNormal">
IOW: client sends ALPN with &quot;HTTP/1.1, HTTP2.0[ws], HTTP2.0[wss]&quot;=
, and server answers<u></u><u></u></p><p class=3D"MsoNormal">with no TLS AL=
PN extension part (or no next protocol selected).<u></u><u></u></p><p class=
=3D"MsoNormal">
<u></u>=C2=A0<u></u></p><p class=3D"MsoNormal">Will the client then default=
 to trying RFC6455 opening handshake (HTTP1.1/Upgrade)?<u></u><u></u></p><p=
 class=3D"MsoNormal"><u></u>=C2=A0<u></u></p><p class=3D"MsoNormal"><u></u>=
=C2=A0<u></u></p>
<p class=3D"MsoNormal">2)<u></u><u></u></p><p class=3D"MsoNormal">&quot;If =
the client has an existing HTTP/2.0 session and it knows that the server su=
pports WebSocket over HTTP/2.0 =E2=80=A6&quot;<u></u><u></u></p><p class=3D=
"MsoNormal">
<u></u>=C2=A0<u></u></p><p class=3D"MsoNormal">How does the client become a=
ware of the server&#39;s support for WS-over-HTTP2?<u></u><u></u></p><p cla=
ss=3D"MsoNormal"><u></u>=C2=A0<u></u></p><p class=3D"MsoNormal">Is that via=
 the HTTP/2.0 SETTINGS frame you mention below?<u></u><u></u></p>
<p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p><p class=3D"MsoNormal"><u></=
u>=C2=A0<u></u></p><p class=3D"MsoNormal">3)<u></u><u></u></p><p class=3D"M=
soNormal">If the client _already_ knows the server supports HTTP/WS, what A=
LPN will it send<u></u><u></u></p>
<p class=3D"MsoNormal">if no HTTP2 session is currently in place?<u></u><u>=
</u></p><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p><p class=3D"MsoNorma=
l">Will it nevertheless send &quot;HTTP/1.1, HTTP2.0[ws], HTTP2.0[wss]&quot=
; (same order)?<u></u><u></u></p>
<p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p><p class=3D"MsoNormal">Or wi=
ll it send &quot;HTTP2.0[ws], HTTP2.0[wss]&quot; only?<u></u><u></u></p><p =
class=3D"MsoNormal"><u></u>=C2=A0<u></u></p><p class=3D"MsoNormal">Or will =
it send &quot;HTTP2.0[ws], HTTP2.0[wss], HTTP/1.1&quot; (reordered)?<u></u>=
<u></u></p>
<p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p><p class=3D"MsoNormal"><u></=
u>=C2=A0<u></u></p><p class=3D"MsoNormal">4) ALPN name for native RFC6455: =
wouldn&#39;t &quot;HTTP/1.1[wss]&quot; better correspond to<u></u><u></u></=
p><p class=3D"MsoNormal">
the actual protocol the client offers / desires to speak?<u></u><u></u></p>=
<p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p><p class=3D"MsoNormal">The c=
lient offers to speak RFC6455, just upgrading from HTTP/1.1, and the server=
<u></u><u></u></p>
<p class=3D"MsoNormal">might want to select a dedicated implementation for =
that.<u></u><u></u></p><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p><p cl=
ass=3D"MsoNormal">In general, a server that supports TLS ALPN might just wa=
nt to lookup a preferred<u></u><u></u></p>
<p class=3D"MsoNormal">implementation given a list of ALPN names offered by=
 a client.<u></u><u></u></p><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p>=
<p class=3D"MsoNormal">Example:<u></u><u></u></p><p class=3D"MsoNormal">Tak=
e a server that implements HTTP/1.1 and HTTP/2.0 for serving static Web con=
tent,<u></u><u></u></p>
<p class=3D"MsoNormal">and RFC6455 for application messaging (all services =
on same ports), and that wants<u></u><u></u></p><p class=3D"MsoNormal">to r=
ealize the following behavior for TLS:<u></u><u></u></p><p class=3D"MsoNorm=
al">
<u></u>=C2=A0<u></u></p><p class=3D"MsoNormal">no ALPN &amp; ALPNs &quot;HT=
TP/1.1&quot;:<u></u><u></u></p><p class=3D"MsoNormal">=C2=A0 =3D&gt; Server=
 choses HTTP/1.1 implementation, no later upgrade to RFC6455 is allowed<u><=
/u><u></u></p>
<p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p><p class=3D"MsoNormal">ALPNs=
 &quot;HTTP/2.0&quot;:<u></u><u></u></p><p class=3D"MsoNormal">=C2=A0 =3D&g=
t; Server choses HTTP/2.0 implementation (without announcing HTTP/WS suppor=
t)<u></u><u></u></p>
<p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p><p class=3D"MsoNormal">ALPNs=
 &quot;HTTP/1.1[wss],HTTP/2.0[wss]&quot;:<u></u><u></u></p><p class=3D"MsoN=
ormal">=C2=A0 =3D&gt; server choses RFC6455 implementation requiring immedi=
ate upgrade (that is, no<u></u><u></u></p>
<p class=3D"MsoNormal">&quot;plain&quot; HTTP/GET requests before HTTP/Upgr=
ade to WS are allowed)<u></u><u></u></p><p class=3D"MsoNormal"><u></u>=C2=
=A0<u></u></p><p class=3D"MsoNormal">IOW: the server wants to adjust it&#39=
;s behavior with ALPNs &quot;HTTP/1.1&quot; versus &quot;HTTP/1.1[wss]&quot=
;:<u></u><u></u></p>
<p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p><p class=3D"MsoNormal">ALPN =
&quot;HTTP/1.1&quot;: client will _never_ HTTP1.1/Upgrade (to RFC6455 or an=
ything other)<u></u><u></u></p><p class=3D"MsoNormal">ALPN &quot;HTTP/1.1[w=
ss]&quot;: client will _immediately_ HTTP1.1/Upgrade to RFC6455<u></u><u></=
u></p>
<p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p><p class=3D"MsoNormal"><u></=
u>=C2=A0<u></u></p><p class=3D"MsoNormal">Cheers,<u></u><u></u></p><p class=
=3D"MsoNormal">/Tobias<u></u><u></u></p><p class=3D"MsoNormal"><u></u>=C2=
=A0<u></u></p><p class=3D"MsoNormal">
<span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-=
serif&quot;;color:#1f497d"><u></u>=C2=A0<u></u></span></p><div style=3D"bor=
der:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt"><div><div =
style=3D"border:none;border-top:solid #e1e1e1 1.0pt;padding:3.0pt 0cm 0cm 0=
cm">
<p class=3D"MsoNormal"><b><span style=3D"font-size:11.0pt;font-family:&quot=
;Calibri&quot;,&quot;sans-serif&quot;">Von:</span></b><span style=3D"font-s=
ize:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;"> <a href=
=3D"mailto:yhirano@google.com" target=3D"_blank">yhirano@google.com</a> [ma=
ilto:<a href=3D"mailto:yhirano@google.com" target=3D"_blank">yhirano@google=
.com</a>] <b>Im Auftrag von </b>Yutaka Hirano<br>
<b>Gesendet:</b> Dienstag, 4. Februar 2014 08:17<br><b>An:</b> Tobias Obers=
tein<br><b>Cc:</b> <a href=3D"mailto:hybi@ietf.org" target=3D"_blank">hybi@=
ietf.org</a>; <a href=3D"mailto:agl@chromium.org" target=3D"_blank">agl@chr=
omium.org</a><br>
<b>Betreff:</b> Re: [hybi] WebSocket over HTTP/2.0 protocol negotiation<u><=
/u><u></u></span></p></div></div><div><div class=3D"h5"><p class=3D"MsoNorm=
al"><u></u>=C2=A0<u></u></p><div><p class=3D"MsoNormal">Hi Tobias,<u></u><u=
></u></p>
<div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div><blockquote style=
=3D"border:none;border-left:solid #cccccc 1.0pt;padding:0cm 0cm 0cm 6.0pt;m=
argin-left:4.8pt;margin-right:0cm"><p class=3D"MsoNormal">In other words: I=
 think a server should actively announce support for any SPDY/WS, and only =
then<br>
should SPDY/WS be tried first when a JS opens a WebSocket connection to an =
origin that has an<br>existing SPDY connection and which already announced =
it&#39;s support and willingness to speak<br>any SPDY/WS.<u></u><u></u></p>
</blockquote><div><p class=3D"MsoNormal">I see. I propose a new negotiation=
 method here and I think it addresses your concern.<u></u><u></u></p></div>=
<div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div><blockquote style=
=3D"margin-left:30.0pt;margin-right:0cm">
<div><p class=3D"MsoNormal"><b>Negotiation Decision Process:</b><u></u><u><=
/u></p></div><div><div><p class=3D"MsoNormal">1: If the client has an exist=
ing HTTP/2.0 session and it knows that the server supports WebSocket over H=
TTP/2.0, create a WebSocket stream on the session. Both of WebSocket (ws) a=
nd secure WebSocket (wss) could share the same HTTP/2.0 session.<u></u><u><=
/u></p>
</div></div><div><p class=3D"MsoNormal">2: If the client is going to establ=
ish a secure WebSocket connection, create a WebSocket connection over TLS.<=
u></u><u></u></p></div><div><p class=3D"MsoNormal">3: If the client is goin=
g to establish a non-secure WebSocket connection, try the WebSocket protoco=
l negotiation.<u></u><u></u></p>
</div><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div><div><p cla=
ss=3D"MsoNormal"><b>Negotiation Cases:</b><u></u><u></u></p></div><div><p c=
lass=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div><div><p class=3D"MsoNormal=
"><b>Create WebSocket Stream on Existing HTTP/2.0 Session</b><u></u><u></u>=
</p>
</div><p class=3D"MsoNormal" style=3D"margin-bottom:12.0pt">The client SHOU=
LD create a new HTTP/2.0 stream in the existing HTTP/2.0 session.<u></u><u>=
</u></p></blockquote><blockquote style=3D"margin-left:30.0pt;margin-right:0=
cm">
<p class=3D"MsoNormal"><b>Create WebSocket Connection over TLS</b><u></u><u=
></u></p><div><p class=3D"MsoNormal">HTTP/2.0 uses TLS Application Level Pr=
otocol Negotiation (ALPN) for the protocol negotiation. The client sends th=
e application protocols it wants to use, &quot;HTTP/1.1&quot; for the nativ=
e WebSocket over TLS, &quot;HTTP/2.0[ws]&quot; for non secure WebSocket ove=
r HTTP/2.0, &quot;HTTP/2.0[wss]&quot; for secure WebSocket over HTTP/2.0.<u=
></u><u></u></p>
</div><div><p class=3D"MsoNormal">If the server selects &quot;HTTP/1.1&quot=
;, the client SHOULD start the native WebSocket opening handshake on TLS.<u=
></u><u></u></p></div><div><div><p class=3D"MsoNormal">If the server select=
s &quot;HTTP/2.0[ws]&quot;, the client SHOULD create an HTTP/2.0 session an=
d start the WebSocket opening handshake on it with &quot;ws&quot; scheme.<u=
></u><u></u></p>
</div></div><div><div><p class=3D"MsoNormal">If the server selects &quot;HT=
TP/2.0[wss]&quot;, the client SHOULD create an HTTP/2.0 session and start t=
he WebSocket opening handshake on it with &quot;wss&quot; scheme.<u></u><u>=
</u></p>
</div></div><div><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div>=
</div><div><div><p class=3D"MsoNormal"><b>The WebSocket Protocol Negotiatio=
n</b><u></u><u></u></p></div></div><div><div><p class=3D"MsoNormal">(same a=
s the document)<u></u><u></u></p>
</div></div><div><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div>=
</div><div><p class=3D"MsoNormal"><b>Supporting scheme negotiation using SE=
TTINGS frame</b><u></u><u></u></p></div><div><p class=3D"MsoNormal">By defa=
ult, a client can send HEADERS frames on schemes that server supports based=
 on the protocol server selected in ALPN. A server SHOULD send a SETTINGS f=
rame as soon as possible whether it supports other schemes or not.<u></u><u=
></u></p>
</div><div><p class=3D"MsoNormal">Here is a proposed ID and value which sho=
uld be defined in the HTTP/2.0 specification. If the endpoint received a HE=
ADERS on unsupporting schemes, the endpoint MUST reset the stream.<u></u><u=
></u></p>
</div><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div><div><p cla=
ss=3D"MsoNormal">Defined IDs: (same as the document)<u></u><u></u></p></div=
></blockquote><div><p class=3D"MsoNormal"><br>No additional RTTs are introd=
uced when the server supports the native WebSocket and HTTP/2.0 but does no=
t support WebSocket over HTTP/2.0.<u></u><u></u></p>
</div><div><p class=3D"MsoNormal">We should introduce the following in the =
HTTP/2.0 specification.<u></u><u></u></p></div><div><p class=3D"MsoNormal">=
=C2=A0- Supporting scheme advertisement using SETTINGS frame<u></u><u></u><=
/p></div>
<div><p class=3D"MsoNormal">=C2=A0- HTTP/2.0[xx] in the ALPN stands for the=
 HTTP/2.0 protocol with scheme xx.<u></u><u></u></p></div><div><p class=3D"=
MsoNormal"><br>Thanks,<u></u><u></u></p><div><p class=3D"MsoNormal"><u></u>=
=C2=A0<u></u></p>
</div></div></div><div><p class=3D"MsoNormal" style=3D"margin-bottom:12.0pt=
"><u></u>=C2=A0<u></u></p><div><p class=3D"MsoNormal">On Thu, Jan 30, 2014 =
at 9:24 PM, Tobias Oberstein &lt;<a href=3D"mailto:tobias.oberstein@tavendo=
.de" target=3D"_blank">tobias.oberstein@tavendo.de</a>&gt; wrote:<u></u><u>=
</u></p>
<blockquote style=3D"border:none;border-left:solid #cccccc 1.0pt;padding:0c=
m 0cm 0cm 6.0pt;margin-left:4.8pt;margin-right:0cm"><div><div><p class=3D"M=
soNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&=
quot;sans-serif&quot;;color:#1f497d">Hi Yutaka,</span><u></u><u></u></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">=C2=A0</span><u></u><u></=
u></p><p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&q=
uot;Calibri&quot;,&quot;sans-serif&quot;;color:#1f497d">the point I was try=
ing to raise is this:</span><u></u><u></u></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">=C2=A0</span><u></u><u></=
u></p><p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&q=
uot;Calibri&quot;,&quot;sans-serif&quot;;color:#1f497d">For a JavaScript in=
 a browser that opens a &quot;ws://=E2=80=A6&quot; connection, there should=
 never be an additional roundtrip</span><u></u><u></u></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">incurred by first probing=
 for any SPDY/WS and then falling back to native WS, but the native WS shou=
ld be</span><u></u><u></u></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">tried first.</span><u></u=
><u></u></p><p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-fam=
ily:&quot;Calibri&quot;,&quot;sans-serif&quot;;color:#1f497d">=C2=A0</span>=
<u></u><u></u></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">In other words: the curre=
nt, immediate connection establishment for native WS should not be put</spa=
n><u></u><u></u></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">at an disadvantage just b=
ecause of introducing any SPDY/WS.</span><u></u><u></u></p><p class=3D"MsoN=
ormal">
<span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-=
serif&quot;;color:#1f497d">=C2=A0</span><u></u><u></u></p><p class=3D"MsoNo=
rmal"><span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot=
;sans-serif&quot;;color:#1f497d">A server should be able to provide both SP=
DY and native WebSocket on the _same_ origin without</span><u></u><u></u></=
p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">being required to impleme=
nt SPDY/WS also, and without incurring an additional roundtrip for</span><u=
></u><u></u></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">probing SPDY/WS while est=
ablishing a WS connection.</span><u></u><u></u></p><p class=3D"MsoNormal"><=
span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-s=
erif&quot;;color:#1f497d">=C2=A0</span><u></u><u></u></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">In other words: I think a=
 server should actively announce support for any SPDY/WS, and only then</sp=
an><u></u><u></u></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">should SPDY/WS be tried f=
irst when a JS opens a WebSocket connection to an origin that has an</span>=
<u></u><u></u></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">existing SPDY connection =
and which already announced it&#39;s support and willingness to speak</span=
><u></u><u></u></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">any SPDY/WS.</span><u></u=
><u></u></p><p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-fam=
ily:&quot;Calibri&quot;,&quot;sans-serif&quot;;color:#1f497d">=C2=A0</span>=
<u></u><u></u></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">Probably I misunderstood =
the draft.</span><u></u><u></u></p><p class=3D"MsoNormal"><span style=3D"fo=
nt-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;;color=
:#1f497d">=C2=A0</span><u></u><u></u></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">Are browsers supposed to =
work like above when they implement any SPDY/WS?</span><u></u><u></u></p><p=
 class=3D"MsoNormal">
<span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-=
serif&quot;;color:#1f497d">=C2=A0</span><u></u><u></u></p><p class=3D"MsoNo=
rmal"><span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot=
;sans-serif&quot;;color:#1f497d">/Tobias</span><u></u><u></u></p>
<p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&quot;Ca=
libri&quot;,&quot;sans-serif&quot;;color:#1f497d">=C2=A0</span><u></u><u></=
u></p><p class=3D"MsoNormal"><span style=3D"font-size:11.0pt;font-family:&q=
uot;Calibri&quot;,&quot;sans-serif&quot;;color:#1f497d">=C2=A0</span><u></u=
><u></u></p>
<div style=3D"border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm =
4.0pt"><div><div style=3D"border:none;border-top:solid #e1e1e1 1.0pt;paddin=
g:3.0pt 0cm 0cm 0cm"><p class=3D"MsoNormal"><b><span style=3D"font-size:11.=
0pt;font-family:&quot;Calibri&quot;,&quot;sans-serif&quot;">Von:</span></b>=
<span style=3D"font-size:11.0pt;font-family:&quot;Calibri&quot;,&quot;sans-=
serif&quot;"> hybi [mailto:<a href=3D"mailto:hybi-bounces@ietf.org" target=
=3D"_blank">hybi-bounces@ietf.org</a>] <b>Im Auftrag von </b>Yutaka Hirano<=
br>
<b>Gesendet:</b> Donnerstag, 30. Januar 2014 07:45<br><b>An:</b> <a href=3D=
"mailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><br><b>Betreff:</=
b> [hybi] WebSocket over HTTP/2.0 protocol negotiation</span><u></u><u></u>=
</p>
</div></div><div><div><p class=3D"MsoNormal">=C2=A0<u></u><u></u></p><div><=
p class=3D"MsoNormal">Hi,<u></u><u></u></p><div><p class=3D"MsoNormal">=C2=
=A0<u></u><u></u></p></div><div><p class=3D"MsoNormal">As I announced, I&#3=
9;ve updated the draft and requested for comments.<u></u><u></u></p>
</div><div><p class=3D"MsoNormal">Tobias gave a comment on protocol negotia=
tion (thanks!) and I think it is good to discuss here, not at the draft com=
ment fields.<u></u><u></u></p></div><div><p class=3D"MsoNormal">=C2=A0<u></=
u><u></u></p>
</div><div><p class=3D"MsoNormal">The draft says:<u></u><u></u></p></div><o=
l start=3D"1" type=3D"1"><li class=3D"MsoNormal">If the client has a HTTP/2=
.0 session established by previous requests and it doesn=E2=80=99t know tha=
t the session does NOT support WebSocket layering over HTTP/2.0, create a W=
ebSocket stream on the session. Both of WebSocket (ws) and secure WebSocket=
 (wss) could share the same HTTP/2.0 session.<u></u><u></u></li>
<li class=3D"MsoNormal">If the client is going to establish a secure WebSoc=
ket connection, create a WebSocket connection over TLS.<u></u><u></u></li><=
li class=3D"MsoNormal">If the client is going to establish a non-secure Web=
Socket connection, try the WebSocket protocol negotiation.<u></u><u></u></l=
i>
</ol><p class=3D"MsoNormal">Tobias doesn&#39;t like the first step. He says=
:<u></u><u></u></p><blockquote style=3D"border:none;border-left:solid #cccc=
cc 1.0pt;padding:0cm 0cm 0cm 6.0pt;margin-left:4.8pt;margin-top:5.0pt;margi=
n-right:0cm;margin-bottom:5.0pt">
<p class=3D"MsoNormal">I think native WS should be tried first. Native WS h=
as less overhead than any WS-over-SPDY layering, and hence should be prefer=
red.<u></u><u></u></p></blockquote><div><p class=3D"MsoNormal">=C2=A0<u></u=
><u></u></p>
</div><div><p class=3D"MsoNormal">I have some comments about it.<u></u><u><=
/u></p></div><div><p class=3D"MsoNormal">=C2=A0- WebSocket over HTTP/2.0 is=
 multiplexed, so it is not so obvious which is preferable. I think it depen=
ds on the popularity of the services:<u></u><u></u></p>
</div><div><p class=3D"MsoNormal">=C2=A0 =C2=A0- WebSocket over HTTP/2.0<u>=
</u><u></u></p></div><div><p class=3D"MsoNormal">=C2=A0 =C2=A0- Native WebS=
ocket and HTTP/2.0<u></u><u></u></p></div><div><p class=3D"MsoNormal">=C2=
=A0 =C2=A0- Native WebSocket and WebSocket over HTTP/2.0<u></u><u></u></p>
</div><div><p class=3D"MsoNormal">=C2=A0 =C2=A0Note that servers which does=
n&#39;t support HTTP/2.0 don&#39;t matter, because they don&#39;t meet the =
precondition of the first step.<u></u><u></u></p></div><div><p class=3D"Mso=
Normal">=C2=A0- =C2=A0There are two scheme negotiation hints described belo=
w.<u></u><u></u></p>
</div><div><p class=3D"MsoNormal">=C2=A0 =C2=A0 If we choose Hint X, before=
 receiving a SETTINGS frame from the server, the client doesn&#39;t know wh=
ich service is preferable. It is rare, but trying connecting to the wrong s=
ervice can occur.<u></u><u></u></p>
</div><div><p class=3D"MsoNormal">=C2=A0 =C2=A0 If we choose Hint Y, whethe=
r the server supports WebSocket over HTTP/2.0 or not is detected by the cli=
ent prior to HTTP/2.0 session establishment and hence the client always und=
erstands the server&#39;s preference.<u></u><u></u></p>
</div><div><p class=3D"MsoNormal">=C2=A0 =C2=A0 So, the actual negotiation =
decision process depends on the hint we choose.<u></u><u></u></p></div><div=
><p class=3D"MsoNormal">=C2=A0<u></u><u></u></p></div><div><p class=3D"MsoN=
ormal">I&#39;m sorry that the draft is not clear at this point, but it will=
 be when the hint is chosen.<u></u><u></u></p>
</div><div><p class=3D"MsoNormal">=C2=A0<u></u><u></u></p></div><div><p cla=
ss=3D"MsoNormal">Thanks,<u></u><u></u></p></div><div><p class=3D"MsoNormal"=
>=C2=A0<u></u><u></u></p></div></div></div></div></div></div></div></blockq=
uote></div><p class=3D"MsoNormal">
<u></u>=C2=A0<u></u></p></div></div></div></div></div></div></blockquote></=
div><br></div>

--bcaec52d55270f611b04f1a485bc--


From yhirano@google.com  Wed Feb  5 02:56:09 2014
Return-Path: <yhirano@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 7DAF71A00DC for <hybi@ietfa.amsl.com>; Wed,  5 Feb 2014 02:56:09 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.913
X-Spam-Level: 
X-Spam-Status: No, score=-1.913 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.535, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id L4ZNEFEQramS for <hybi@ietfa.amsl.com>; Wed,  5 Feb 2014 02:56:07 -0800 (PST)
Received: from mail-yh0-x22b.google.com (mail-yh0-x22b.google.com [IPv6:2607:f8b0:4002:c01::22b]) by ietfa.amsl.com (Postfix) with ESMTP id 445391A00D4 for <hybi@ietf.org>; Wed,  5 Feb 2014 02:56:07 -0800 (PST)
Received: by mail-yh0-f43.google.com with SMTP id z6so151624yhz.16 for <hybi@ietf.org>; Wed, 05 Feb 2014 02:56:06 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=sQtPA0TtKZXhK3NNL08iIR17XTxWycAc7aB7ZftkXxo=; b=cnhc40qo43565Q0GgPTAbUvmXFGTQwvZVMPDTjdTIPN77ZR/6LzUecQLF5pUvr9XPZ iJZx0EQ01/BqJGb5w8YqNjf94Y7G48t/3tjCH9YCvXLdtc75TmDicqBDOYD74KJ3d8lB evoMC/fwlP2U2Xtth7G/qnI/bwxhQ2iktRy579g1ShoU6xpNU6uC0X0B8E4H8v7Z6LRQ 3cIo2QNvntG9UOyH/FwE0kANLhrYquoTIF4trRCmgm/kKWjKh6TuWVL9WgicM7pIvN3m N9Uf9FOHkR3d2P9ZUBbmRCr4evqbKgQYaX3MOADWtZpBbqC5mHrypUXvNTPjeDLZq/2o BCiw==
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=chromium.org; s=google; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=sQtPA0TtKZXhK3NNL08iIR17XTxWycAc7aB7ZftkXxo=; b=ibvBX2iobNJ6M2Arp7gNMP97F+OTOif7XfwLcwslheOhikOibQ+QeLM6eZeVz7idEd 9LU5Xh2o4OKsJ3XXmixUqREUhkuUGu9ViojyyGQxqVp6DUb0w8+bnT56w6DsyLAQbc1h db6A7TM62jt8iar/GUTRPse0dpEOguKZ/rgm4=
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:sender:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=sQtPA0TtKZXhK3NNL08iIR17XTxWycAc7aB7ZftkXxo=; b=m0qe92cGLWhSoHOI7LjtgBap8ca+fYr1/AbFYQ30RfQxx6cpWQBs09wSaA9dRcM1Bs b4ZZEZWfv3oHroS5XITFNUSWRMOyWkPiN+Xb7LeyuxW3L6mmYPk9qP2/d0Taa2Zj8dMC tgH4p52RVWugC0A760shtH+MnZujzozpRqk0EgEgGt+jFBIgIP1VOjMvrtOBFclMFMMr xkvEy//ft5tWcRAG3jspgSgGKj47jV3oCBqrnRzgkNw3Sqdn3gKb4cAGMXOOiWcwn1kI Xtxoac8qdDlN54Ti3xpgs/NNA5LSoaEF++cFhTCjfwTELvElann/8eeDBOYc+isMwsdq MpMA==
X-Gm-Message-State: ALoCoQlrrND4yZaS3jMylyJkAuxasNB12za+I0NT0RN+Ctf2rp1NcOmzyAwxtYwlYmS2PdvKzi48YfRwD4pE2PdNiUkdH5g5VgZX8jR3xOLISvgjY02U32NXyed+CLprxI0sx5eUCxwwHhbnrySQol+ASkRzCme7J73ZiCM+6NAciNCLKUpGrv8eClETYC/EFlzWMBClwhHo
MIME-Version: 1.0
X-Received: by 10.236.230.3 with SMTP id i3mr601699yhq.13.1391597766312; Wed, 05 Feb 2014 02:56:06 -0800 (PST)
Sender: yhirano@google.com
Received: by 10.170.135.22 with HTTP; Wed, 5 Feb 2014 02:56:06 -0800 (PST)
In-Reply-To: <3274C7C2-41D4-49AD-971E-DA3D85898BC2@zaphoyd.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <CABihn6Gc34oSR70kkYUxpxQOMb2ZepR9STkLoe-SrZPTP8voYg@mail.gmail.com> <3274C7C2-41D4-49AD-971E-DA3D85898BC2@zaphoyd.com>
Date: Wed, 5 Feb 2014 19:56:06 +0900
X-Google-Sender-Auth: VXeLlS2XP0ArNXqWQHj1IgDW314
Message-ID: <CABihn6GGezUU2aSboK0wsxYX3TBWTz5Av6-ZFJTPy6PQ50LrTQ@mail.gmail.com>
From: Yutaka Hirano <yhirano@chromium.org>
To: Peter Thorson <webmaster@zaphoyd.com>
Content-Type: multipart/alternative; boundary=001a11c1d40088329a04f1a69a37
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 05 Feb 2014 10:56:09 -0000

--001a11c1d40088329a04f1a69a37
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

>
> Can we clarify again what plan A, B, and C are?
> I don=E2=80=99t see any issue with message boundary detection using only =
MSG_DONE
> in either a native HTTP/2.0 message plan or a purely tunneled plan.


Let me talk about PlanA, WS-TUNNEL with MSG_DONE.
Roberto says he is interested in loadbalancing ***without needing to
understand application-layer semantics***. (1)
I say we can't detect message boundary ***without understanding WebSocket
level semantics***. (2)
You say we can detect message boundary ***with the RFC6455 parser***. (3)

Year, I think (2) and (3) are both true.

I thought that Roberto didn't want to implement an RFC6455 paser for the
loadbalancer.
If it is wrong: Yay! PlanA (with MSG_DONE) seems good.

If it is correct:
I still think MSG_DONE is insufficient for the loadbalancing.
Imagine a WebSocket connection consisting of three WebSocket frames, f1
[without FIN], f2 [with FIN] and f3 [with FIN].
Assume that there is no extensions.
Assume that one HTTP/2.0 frame is mapped to one WebSocket frame for
simplicity.
Let the HTTP/2.0 frames be hf1 [without MSG_DONE], hf2 [with MSG_DONE] and
hf3 [with MSG_DONE].
If the loadbalancer does NOT understand RFC6455, there are two
possibilities,
 - A message consists of (hf1, hf2) and another message consists of (hf3).
 - A data message consists of (hf1, hf3) and a control message consists of
(hf2).
Hence it fails to recognize messages.

Note: For enabling intermediaries to transfer WS over HTTP/2.0 data without
understanding RFC6455, MSG_DONE is sufficient.



On Wed, Feb 5, 2014 at 2:08 AM, Peter Thorson <webmaster@zaphoyd.com> wrote=
:

>
> On Feb 4, 2014, at 08:24 , Yutaka Hirano <yhirano@chromium.org> wrote:
>
> Hi Roberto,
>
>> I'm interested in correctly muxing, demuxing, and loadbalancing of
>> streams-of-messages without needing to understand application-layer
>> semantics.
>
> IIUC None of Plan A, B and C provide the ability to detect exact message
> boundary without understanding WebSocket level semantics even if we give =
up
> some extensions, because of control message injection.
> Introducing START_MESSAGE flag in addition to MSG_DONE enables it with
> Plan A or Plan C. Is it acceptable for you?
>
>
> Can we clarify again what plan A, B, and C are?
>
> I don=E2=80=99t see any issue with message boundary detection using only =
MSG_DONE
> in either a native HTTP/2.0 message plan or a purely tunneled plan.
>
> Under tunneled plan the only information that is necessary for HTTP/2.0
> intermediaries to understand in order to mux/demux/load balance/etc is th=
e
> stream end bit (unless you mean to balance messages within a stream to
> different backends). All information about message boundaries and control
> messages is contained in the stream data itself and the RFC6455 parser
> eventually reading the stream will be able to pick all that up without an=
y
> additional help.
>
> Under the native HTTP/2.0 message plan there is no need for control
> messages at all except implicitly a =E2=80=9Cclose=E2=80=9D frame (i.e. a=
 frame with stream
> end set) which as the last frame in the stream should not conflict with
> message boundary detection.
>

--001a11c1d40088329a04f1a69a37
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"im" style=3D"font-family:arial,sans-serif;fo=
nt-size:13px"><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px=
 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left=
-style:solid;padding-left:1ex">
Can we clarify again what plan A, B, and C are?<br>I don=E2=80=99t see any =
issue with message boundary detection using only MSG_DONE in either a nativ=
e HTTP/2.0 message plan or a purely tunneled plan.</blockquote><div>=C2=A0<=
/div><div>
<span style=3D"color:rgb(34,34,34);font-family:arial;font-size:small">Let m=
e talk about PlanA, WS-TUNNEL with MSG_DONE.</span><br></div></div><div>Rob=
erto says he is interested in loadbalancing ***without needing to understan=
d application-layer semantics***. (1)</div>
<div>I say we can&#39;t detect message boundary ***without understanding We=
bSocket level semantics***. (2)</div><div>You say we can detect message bou=
ndary ***with the RFC6455 parser***. (3)</div><div><br></div><div>Year, I t=
hink (2) and (3) are both true.</div>
<div><br></div><div>I thought that Roberto didn&#39;t want to implement an =
RFC6455 paser for the loadbalancer.</div><div>If it is wrong: Yay! PlanA (w=
ith MSG_DONE) seems good.<br></div><div><div><br></div><div>If it is correc=
t:</div>
<div>I still think MSG_DONE is insufficient for the loadbalancing.</div></d=
iv><div>Imagine a WebSocket connection consisting of three WebSocket frames=
, f1 [without FIN], f2 [with FIN] and f3 [with FIN].<br></div><div>Assume t=
hat there is no extensions.</div>
<div>Assume that one HTTP/2.0 frame is mapped to one WebSocket frame for si=
mplicity.</div><div>Let the HTTP/2.0 frames be hf1 [without MSG_DONE], hf2 =
[with MSG_DONE] and hf3 [with MSG_DONE].</div><div>If the loadbalancer does=
 NOT understand RFC6455, there are two possibilities,</div>
<div><div>=C2=A0- A message consists of (hf1, hf2) and another message cons=
ists of (hf3).</div></div><div><div>=C2=A0- A data message consists of (hf1=
, hf3) and a control message consists of (hf2).</div></div><div>Hence it fa=
ils to recognize messages.</div>
<div><br></div><div>Note: For enabling intermediaries to transfer WS over H=
TTP/2.0 data without understanding RFC6455, MSG_DONE is sufficient.=C2=A0</=
div><div><br></div></div><div class=3D"gmail_extra"><br><br><div class=3D"g=
mail_quote">
On Wed, Feb 5, 2014 at 2:08 AM, Peter Thorson <span dir=3D"ltr">&lt;<a href=
=3D"mailto:webmaster@zaphoyd.com" target=3D"_blank">webmaster@zaphoyd.com</=
a>&gt;</span> wrote:<br><blockquote class=3D"gmail_quote" style=3D"margin:0=
 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div style=3D"word-wrap:break-word"><div class=3D"im"><br><div><div>On Feb =
4, 2014, at 08:24 , Yutaka Hirano &lt;<a href=3D"mailto:yhirano@chromium.or=
g" target=3D"_blank">yhirano@chromium.org</a>&gt; wrote:</div><br><blockquo=
te type=3D"cite">
<div dir=3D"ltr"><div>Hi Roberto,=C2=A0</div><blockquote class=3D"gmail_quo=
te" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-col=
or:rgb(204,204,204);border-left-style:solid;padding-left:1ex"><span style=
=3D"font-family:arial,sans-serif;font-size:13px">I&#39;m interested in corr=
ectly muxing, demuxing, and loadbalancing of streams-of-messages without ne=
eding to understand application-layer semantics.</span></blockquote>


<div>IIUC None of Plan A, B and C provide the ability to detect exact messa=
ge boundary without understanding WebSocket level semantics even if we give=
 up some extensions, because of control message injection.</div><div>Introd=
ucing START_MESSAGE flag in addition to MSG_DONE enables it with Plan A or =
Plan C. Is it acceptable for you?</div>
</div></blockquote><br></div></div><div>Can we clarify again what plan A, B=
, and C are?</div><div><br></div><div>I don=E2=80=99t see any issue with me=
ssage boundary detection using only MSG_DONE in either a native HTTP/2.0 me=
ssage plan or a purely tunneled plan.</div>
<div><br></div><div>Under tunneled plan the only information that is necess=
ary for HTTP/2.0 intermediaries to understand in order to mux/demux/load ba=
lance/etc is the stream end bit (unless you mean to balance messages within=
 a stream to different backends). All information about message boundaries =
and control messages is contained in the stream data itself and the RFC6455=
 parser eventually reading the stream will be able to pick all that up with=
out any additional help.</div>
<div><br></div><div>Under the native HTTP/2.0 message plan there is no need=
 for control messages at all except implicitly a =E2=80=9Cclose=E2=80=9D fr=
ame (i.e. a frame with stream end set) which as the last frame in the strea=
m should not conflict with message boundary detection.</div>
</div></blockquote></div><br></div>

--001a11c1d40088329a04f1a69a37--

From webmaster@zaphoyd.com  Wed Feb  5 07:42:23 2014
Return-Path: <webmaster@zaphoyd.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 7397F1A0193 for <hybi@ietfa.amsl.com>; Wed,  5 Feb 2014 07:42:23 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.899
X-Spam-Level: 
X-Spam-Status: No, score=-1.899 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_NONE=-0.0001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id qBNROd-IfqmW for <hybi@ietfa.amsl.com>; Wed,  5 Feb 2014 07:42:20 -0800 (PST)
Received: from authsmtp01.uchicago.edu (authsmtp01.uchicago.edu [128.135.12.121]) by ietfa.amsl.com (Postfix) with ESMTP id 903031A00FD for <hybi@ietf.org>; Wed,  5 Feb 2014 07:42:20 -0800 (PST)
Received: from [10.0.1.89] (c-68-51-76-178.hsd1.il.comcast.net [68.51.76.178]) (Authenticated sender: zaphoyd) by authsmtp01.uchicago.edu (Postfix) with ESMTP id 3B8FF49809E; Wed,  5 Feb 2014 09:42:19 -0600 (CST)
Content-Type: multipart/alternative; boundary="Apple-Mail=_88B90F19-12CD-41C3-8DE1-48F7ABE1E62F"
Mime-Version: 1.0 (Mac OS X Mail 7.0 \(1822\))
From: Peter Thorson <webmaster@zaphoyd.com>
In-Reply-To: <CABihn6GGezUU2aSboK0wsxYX3TBWTz5Av6-ZFJTPy6PQ50LrTQ@mail.gmail.com>
Date: Wed, 5 Feb 2014 09:46:24 -0600
Message-Id: <740FF040-F1E9-43E6-ACDE-0014FE4D012F@zaphoyd.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <CABih n6Gc34oSR70kkYUxpxQOMb2ZepR9STkLoe-SrZPTP8voYg@mail.gmail.com> <3274C7C2-41D4-49AD-971E-DA3D85898BC2@zaphoyd.com> <CABihn6GGezUU2aSboK0wsxYX3TBWTz5Av6-ZFJTPy6PQ50LrTQ@mail.gmail.com>
To: Yutaka Hirano <yhirano@chromium.org>
X-Mailer: Apple Mail (2.1822)
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 05 Feb 2014 15:42:23 -0000

--Apple-Mail=_88B90F19-12CD-41C3-8DE1-48F7ABE1E62F
Content-Transfer-Encoding: quoted-printable
Content-Type: text/plain;
	charset=us-ascii


On Feb 5, 2014, at 4:56, Yutaka Hirano <yhirano@chromium.org> wrote:

> Let me talk about PlanA, WS-TUNNEL with MSG_DONE.
> Roberto says he is interested in loadbalancing ***without needing to =
understand application-layer semantics***. (1)
> I say we can't detect message boundary ***without understanding =
WebSocket level semantics***. (2)
> You say we can detect message boundary ***with the RFC6455 parser***. =
(3)
>=20
> Year, I think (2) and (3) are both true.
>=20
> I thought that Roberto didn't want to implement an RFC6455 paser for =
the loadbalancer.
> If it is wrong: Yay! PlanA (with MSG_DONE) seems good.
>=20
> If it is correct:
> I still think MSG_DONE is insufficient for the loadbalancing.
> Imagine a WebSocket connection consisting of three WebSocket frames, =
f1 [without FIN], f2 [with FIN] and f3 [with FIN].
> Assume that there is no extensions.
> Assume that one HTTP/2.0 frame is mapped to one WebSocket frame for =
simplicity.
> Let the HTTP/2.0 frames be hf1 [without MSG_DONE], hf2 [with MSG_DONE] =
and hf3 [with MSG_DONE].
> If the loadbalancer does NOT understand RFC6455, there are two =
possibilities,
>  - A message consists of (hf1, hf2) and another message consists of =
(hf3).
>  - A data message consists of (hf1, hf3) and a control message =
consists of (hf2).
> Hence it fails to recognize messages.
>=20
> Note: For enabling intermediaries to transfer WS over HTTP/2.0 data =
without understanding RFC6455, MSG_DONE is sufficient.=20

Let me define "Plan D" more precisely:
HTTP/2.0 DATA frames are used as a dumb transport for an RFC6455 byte =
stream. There is no 1:1 frame mapping between WS frames and HTTP/2.0 =
frames. A browser could send two WebSocket frames back to back in one =
HTTP/2.0 frame. An intermediary could split this frame evenly into four =
HTTP/2.0 frames each containing half of an RFC6455 frame. The remote =
endpoint would accept an HTTP/2.0 frame, feed its payload bytes straight =
to an RFC6455 parser which would handle the rest. A MSG_DONE flag is not =
required at all even for full ignorant intermediary compatibility.

Benefits:
- 100% RFC6455 compatibility present and future
- Simple implementation:
-- Nothing needs to be added to HTTP/2.0 or RFC6455.=20
-- Browsers leverage existing RFC6455 parsers and just add logic to =
choose whether to use an HTTP/2.0 stream or a raw TCP stream. Data =
centers demux the streams into RFC6455 message processors.
-- Everyone gets trivial backwards compatibility with existing RFC6455 =
software and hardware, both on the client and server side.
- Intermediaries that don't care can safely ignore or re-fragment =
however they like.

Drawbacks
- Requires intermediaries and endpoints that want to understand messages =
to read RFC6455. (I don't think this is an issue, but it seems important =
to Roberto)
- Retains unnecessary/redundant features (masking, control messages, =
frame length fields)

I think this is the best option if we value RFC6455 compatibility or =
desire ease of integration into existing software and environments. =
Otherwise, I would prefer that we go the "Plan B extreme" route that =
ditches RFC6455 entirely and builds a completely native HTTP/2.0 message =
stream. MSG_DONE/type bits, no masking, no opcode, no rsv, no control =
frames.

Benefits:
- HTTP/2.0 software doesn't need to know about RFC6455 unless it wants =
backwards compatibility with non-HTTP/2.0 software
- Doesn't re-create lots of irrelevant/duplicated features from RFC6455
- Optimal wire efficiency.

Drawbacks:
- Intermediaries need to implement "HTTP/2.0 messaging". (it is simple =
enough now though that that shouldn't be a great hardship.)
- Limited RFC6455 extension compatibility.

I don't have a strong preference for either of these options. I do have =
a strong preference that we not do something in between. RFC6455 =
compatibility is useful for me because it allows sharing =
code/infrastructure and backwards compatibility. Not because the =
specific features of RFC6455 are preserved.

That is the reason I think adding MSG_DONE to WS-TUNNEL doesn't make =
sense. It puts us in a situation where *all* HTTP/2.0 implementations =
must implement a bunch of message related functionality that is already =
in RFC6455. As Yukata points out MSG_DONE alone is not sufficient to =
detect messages in a tunneled RFC6455 stream. You'd need to throw in an =
opcode as well. At this point we end up re-implementing half of RFC6455 =
in HTTP/2.0 in order to avoid using RFC6455 in a product that already =
has an RFC6455 parser?? The only gain I see is certain types of limited =
and specialized intermediaries can drop RFC6455. That doesn't seem like =
a good tradeoff given that we could just choose Plan B and get rid of =
RFC6455 for everybody.

Regarding message based load balancing for WebSocket:
I agree with Tobias that this doesn't make sense. An application that =
wants to send independent messages can send HTTP requests. Such requests =
can be easily muxed, demuxed, or load balanced. Each message has headers =
to store application specific information necessary to correctly load =
balance. This can all be done efficiently with the functionality =
HTTP/2.0 provides already.


--Apple-Mail=_88B90F19-12CD-41C3-8DE1-48F7ABE1E62F
Content-Transfer-Encoding: quoted-printable
Content-Type: text/html;
	charset=us-ascii

<html><head><meta http-equiv=3D"Content-Type" content=3D"text/html =
charset=3Dus-ascii"></head><body style=3D"word-wrap: break-word; =
-webkit-nbsp-mode: space; -webkit-line-break: =
after-white-space;"><br><div><div>On Feb 5, 2014, at 4:56, Yutaka Hirano =
&lt;<a href=3D"mailto:yhirano@chromium.org">yhirano@chromium.org</a>&gt; =
wrote:</div><br class=3D"Apple-interchange-newline"><blockquote =
type=3D"cite"><div dir=3D"ltr"><div class=3D"im" =
style=3D"font-family:arial,sans-serif;font-size:13px"><blockquote =
class=3D"gmail_quote" style=3D"margin:0px 0px 0px =
0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left=
-style:solid;padding-left:1ex"><span style=3D"color: rgb(34, 34, 34); =
font-family: arial; font-size: small;">Let me talk about PlanA, =
WS-TUNNEL with MSG_DONE.</span></blockquote></div><div>Roberto says he =
is interested in loadbalancing ***without needing to understand =
application-layer semantics***. (1)</div>
<div>I say we can't detect message boundary ***without understanding =
WebSocket level semantics***. (2)</div><div>You say we can detect =
message boundary ***with the RFC6455 parser***. =
(3)</div><div><br></div><div>Year, I think (2) and (3) are both =
true.</div>
<div><br></div><div>I thought that Roberto didn't want to implement an =
RFC6455 paser for the loadbalancer.</div><div>If it is wrong: Yay! PlanA =
(with MSG_DONE) seems good.<br></div><div><div><br></div><div>If it is =
correct:</div>
<div>I still think MSG_DONE is insufficient for the =
loadbalancing.</div></div><div>Imagine a WebSocket connection consisting =
of three WebSocket frames, f1 [without FIN], f2 [with FIN] and f3 [with =
FIN].<br></div><div>Assume that there is no extensions.</div>
<div>Assume that one HTTP/2.0 frame is mapped to one WebSocket frame for =
simplicity.</div><div>Let the HTTP/2.0 frames be hf1 [without MSG_DONE], =
hf2 [with MSG_DONE] and hf3 [with MSG_DONE].</div><div>If the =
loadbalancer does NOT understand RFC6455, there are two =
possibilities,</div>
<div>&nbsp;- A message consists of (hf1, hf2) and another message =
consists of (hf3).</div><div>&nbsp;- A data message consists of (hf1, =
hf3) and a control message consists of (hf2).</div><div>Hence it fails =
to recognize messages.</div>
<div><br></div><div>Note: For enabling intermediaries to transfer WS =
over HTTP/2.0 data without understanding RFC6455, MSG_DONE is =
sufficient.&nbsp;</div></div></blockquote><br></div><div><div>Let me =
define "Plan D" more precisely:</div><div>HTTP/2.0 DATA frames are used =
as a dumb transport for an RFC6455 byte stream. There is no 1:1 frame =
mapping between WS frames and HTTP/2.0 frames. A browser could send two =
WebSocket frames back to back in one HTTP/2.0 frame. An intermediary =
could split this frame evenly into four HTTP/2.0 frames each containing =
half of an RFC6455 frame. The remote endpoint would accept an HTTP/2.0 =
frame, feed its payload bytes straight to an RFC6455 parser which would =
handle the rest. A MSG_DONE flag is not required at all even for full =
ignorant intermediary =
compatibility.</div><div><br></div><div>Benefits:</div><div>- 100% =
RFC6455 compatibility present and future</div><div>- Simple =
implementation:</div><div>-- Nothing needs to be added to HTTP/2.0 or =
RFC6455.&nbsp;</div><div>-- Browsers leverage existing RFC6455 parsers =
and just add logic to choose whether to use an HTTP/2.0 stream or a raw =
TCP stream. Data centers demux the streams into RFC6455 message =
processors.</div><div>-- Everyone gets trivial backwards compatibility =
with existing RFC6455 software and hardware, both on the client and =
server side.</div><div>- Intermediaries that don't care can safely =
ignore or re-fragment however they =
like.</div><div><br></div><div>Drawbacks</div><div>- Requires =
intermediaries and endpoints that want to understand messages to read =
RFC6455. (I don't think this is an issue, but it seems important to =
Roberto)</div><div>- Retains unnecessary/redundant features (masking, =
control messages, frame length fields)</div><div><br></div><div>I think =
this is the best option if we value RFC6455 compatibility or desire ease =
of integration into existing software and environments. Otherwise, I =
would prefer that we go the "Plan B extreme" route that ditches RFC6455 =
entirely and builds a completely native HTTP/2.0 message stream. =
MSG_DONE/type bits, no masking, no opcode, no rsv, no control =
frames.</div><div><br></div><div>Benefits:</div><div>- HTTP/2.0 software =
doesn't need to know about RFC6455 unless it wants backwards =
compatibility with non-HTTP/2.0 software</div><div>- Doesn't re-create =
lots of irrelevant/duplicated features from RFC6455</div><div>- Optimal =
wire efficiency.</div><div><br></div><div>Drawbacks:</div><div>- =
Intermediaries need to implement "HTTP/2.0 messaging". (it is simple =
enough now though that that shouldn't be a great hardship.)</div><div>- =
Limited RFC6455 extension compatibility.</div><div><br></div><div>I =
don't have a strong preference for either of these options. I do have a =
strong preference that we not do something in between. RFC6455 =
compatibility is useful for me because it allows sharing =
code/infrastructure and backwards compatibility. Not because the =
specific features of RFC6455 are =
preserved.</div><div><br></div><div>That is the reason I think adding =
MSG_DONE to WS-TUNNEL doesn't make sense. It puts us in a situation =
where *all* HTTP/2.0 implementations must implement a bunch of message =
related functionality that is already in RFC6455. As Yukata points out =
MSG_DONE alone is not sufficient to detect messages in a tunneled =
RFC6455 stream. You'd need to throw in an opcode as well. At this point =
we end up re-implementing half of RFC6455 in HTTP/2.0 in order to avoid =
using RFC6455 in a product that already has an RFC6455 parser?? The only =
gain I see is certain types of limited and specialized intermediaries =
can drop RFC6455. That doesn't seem like a good tradeoff given that we =
could just choose Plan B and get rid of RFC6455 for =
everybody.</div><div><br></div><div>Regarding message based load =
balancing for WebSocket:</div><div>I agree with Tobias that this doesn't =
make sense. An application that wants to send independent messages can =
send HTTP requests. Such requests can be easily muxed, demuxed, or load =
balanced. Each message has headers to store application specific =
information necessary to correctly load balance. This can all be done =
efficiently with the functionality HTTP/2.0 provides =
already.</div></div><br></body></html>=

--Apple-Mail=_88B90F19-12CD-41C3-8DE1-48F7ABE1E62F--

From fenix@google.com  Wed Feb  5 10:19:00 2014
Return-Path: <fenix@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id A7EBD1A00EB for <hybi@ietfa.amsl.com>; Wed,  5 Feb 2014 10:19:00 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.913
X-Spam-Level: 
X-Spam-Status: No, score=-1.913 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.535, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id oqIQq_fzUZ7Q for <hybi@ietfa.amsl.com>; Wed,  5 Feb 2014 10:18:58 -0800 (PST)
Received: from mail-qc0-x22f.google.com (mail-qc0-x22f.google.com [IPv6:2607:f8b0:400d:c01::22f]) by ietfa.amsl.com (Postfix) with ESMTP id D7AF81A00FA for <hybi@ietf.org>; Wed,  5 Feb 2014 10:18:57 -0800 (PST)
Received: by mail-qc0-f175.google.com with SMTP id x13so1251445qcv.20 for <hybi@ietf.org>; Wed, 05 Feb 2014 10:18:57 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=C7MzkfEUJIRpqScbLkL/YlctACPzI/7C3IKnjpWklyI=; b=Pr9TmdNh2QO2YJEGMcIfqDsyBr/rOLaQgiH3MuuZpYQFCBETQpLfIZLsnhjNwBLCqV S/epw1bPnjitSrj/Fd6OnSXn96jyPaW3/EncEvDzzMLNvsq9VYSsiKACwFnk/SHdfljy qXSOfz4OD291IqQlKWDl48yHWX9HFp0a5elXix58rMjBFfL7J8V2RPUp5ms6QNYjTCdr A4/CLXEMlMYmCEiMgWFT4+yXa5c0Uxayflk8ZI4HWW38vo9Ykt3a0DawhTSOlVTxuIKY GgUrcm38OvXyeoWASv90+OhuOyLcUzqgLlD5s6v9JEnCzy2gJ3opnyaYmAUiIGrrNHcn ReWw==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=C7MzkfEUJIRpqScbLkL/YlctACPzI/7C3IKnjpWklyI=; b=ViBoC2yP+9xp2BzAz5j4yuiUaCq0WG+wtWj8eslAwp4PVH3MeL9seAIdpZMdNWJLb8 rrzQEQ8t+txKzGTxh3NMjLwZS3yizXSvDATAbewiiqM9ddQpL232Op3B/vLGjLs6HMhr /wOK1w1K/dKWPtt3whNlq7iSRUbSzG2FAZ1dom2dFRWRouRVve6VRxQWEozGDQyf05v7 Z0t/xzb1inys9TUGP5YHXtFTc5itU81oRMZuTjVJu+/mwNwivrFgENNfu+ENWRJN43M5 j8LCgM2oJ7XU7OeniHm5+x2BKDkWKAJX9xNsS+QfvDgUD30QJ/htsTGZFNUO5qPJK5gm qdAA==
X-Gm-Message-State: ALoCoQlldl3Ffu7DwHpdTN5nQHwdu5S+D6s3IUrsV+SNNFu/4MACNvKy18Exkbm9xnY0AkO5fe/aL3n7xxZPMPwWN3KG6nGCWGOR4hMRxZRH2weRu4GnDz0ekE9Na6FnEySe7z8D/20a4HnSIpWodDR5hIm2NJhr9kkAY3F4FkwjnAIomZZGQoR8CK35MnmX3lV26zQua07O
MIME-Version: 1.0
X-Received: by 10.224.96.135 with SMTP id h7mr4980758qan.53.1391624336498; Wed, 05 Feb 2014 10:18:56 -0800 (PST)
Received: by 10.229.13.7 with HTTP; Wed, 5 Feb 2014 10:18:56 -0800 (PST)
In-Reply-To: <634914A010D0B943A035D226786325D4446C29584C@EXVMBX020-12.exch020.serverdata.net>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <CAGzyod6H-To9EhaPqW5C0c9xyy5XnRg_851GuT3q=X+KoXe7Wg@mail.gmail.com> <0533B942-C1DA-4BFE-BE80-4722C0BD740D@zaphoyd.com> <CAGzyod7xtCZqUvcb0gX0PYKyikyd4kKc=kOzwzgfAmUFExCq3g@mail.gmail.com> <634914A010D0B943A035D226786325D4446C29584C@EXVMBX020-12.exch020.serverdata.net>
Date: Wed, 5 Feb 2014 10:18:56 -0800
Message-ID: <CAGzyod6LZ=wXj3jXYQ4iCo+jtFEQHfm+NEnei9N6AoaLrYyH1g@mail.gmail.com>
From: Roberto Peon <fenix@google.com>
To: Tobias Oberstein <tobias.oberstein@tavendo.de>
Content-Type: multipart/alternative; boundary=001a11c3ec923cff8104f1acca3e
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 05 Feb 2014 18:19:00 -0000

--001a11c3ec923cff8104f1acca3e
Content-Type: text/plain; charset=UTF-8

On Tue, Feb 4, 2014 at 11:44 AM, Tobias Oberstein <
tobias.oberstein@tavendo.de> wrote:

> >In other words, if a loadbalancer sees that the WS stream-of-messages is
> targeted at an endpoint that can process out-of-order messages (by
> examining the HEADERS frame initializing the stream-of-messages), >then it
> is free to demux the messages and forward without worrying about reordering
> on the outbound path.
>
> I can't follow the whole argumentation.
>
> Why would an application that does not need an ordered message channel,
> but is better served by e.g. independent
> request/response message exchanges choose an ordered transport protocol in
> the first place?
>
> And then need some ad-hoc, non-standard / application-specific way of
> signaling to intermediaries
> that in fact it doesn't rely on the ordering the transport otherwise would
> guarantee?
>

Neither application semantics nor the format of the data for an application
are defined in a standard-- they're defined by the application. The
loadbalancer should be able to help deal with application semantics without
having to worry about the format of the data.

HTTP/2 allows for the generic demarcation of message-boundaries, which
enables this kind of application-specific loadbalancing with generic rules
that don't need to vary based on the protocol spoken inside the HTTP/2
stream.



>
> Such an app should just do HTTP request/response style coms.
>

Why? Using the WS API would be more efficient. The loadbalancer can execute
policy such as "last received wins": Why bother to send back responses for
messages that you know are now uninteresting?


>
> An app that needs ordering will request an ordered transport, and _not_
> later give up on that again.
>
>
That has not been my experience.


> And it's no longer a generic intermediary: it must be configured to "know"
> that
> certain targets are in fact allowed to be reordered, despite the transport
> being ordered
> originally.
>

I disagree: It is completely generic as it allows for any protocol,
including application-layer protocol


>
> >Tobias: We loadbalance myriad things, with RPC and API stuff being part
> of the mix.
>
> I guess you are talking about protocols _above_ HTTP or WS?
>
> Curious: which of those protocols choose an ordered message transport, and
> in fact
> do _not_ rely on the actual ordering guarantees?
>

You're making strong assertions about why people use a protocol, but in
fact, people just use the hammer that is available, and pick the shape
closest to what they need.

Both RPC and API things often don't care about careful ordering of requests
to responses.
They're seeking semantics of both WS and HTTP: from WS-like semantics they
want message-based and guaranteed-single-session and from HTTP-like
semantics they want out-of-order.
They're solving an application-layer problem which the transport doesn't
solve on its own.
Such a "protocol" has not been defined because we do a poor job of
designing APIs.
The intent here is to remedy this by thinking about how the protocol
interacts with the entire ecosystem, in particular with proxies and
loadbalancers.
Why do we have to change the LB code for every protocol? If we do this the
smart way, then we don't have to!



>
> Since that is the niche we are now talking about.
>

We're talking about being able to loadbalance *any* bytestream protocol or
*any* stream-of-message protocol. That pretty much covers the gamut.


>
> In any case: my impression/opinion is, that this is the core of the crux:
>
> The simplest approach of W3C API plus per-stream
> muxing/demuxing/loadbalancing WS over SPDY with 100% WS compatibility would
> be full tunneling.
>

It is simple only if you are willing to commit to adding more code ad
infinitum to servers, one for each protocol.
That seems like a poor tradeoff, especially given that it doesn't actually
offer any more WS compatibility.



>
> The use case not covered by the latter is "_per message_
> muxing/demuxing/loadbalancing".
>
> And that "use case" I can't follow, since it gives up message ordering,
> and would be better served by plain HTTP anyway ..
>
> There is no "unordered WebSocket" .. that is HTTP;)
>

WS here is a protocol, and possibly an API. Neither of these specify
in-order responses. Certainly both HTTP and WS specify in-order delivery.
HTTP actually mandates in-order response, which is counter to what many
applications want and is a HORRIBLE thing for demuxing loadbalancers as it
requires potentially infinite buffering. This is a real practical problem :)

Applications generally want to choose from:
unordered/ordered | reliable/unreliable | message/bytestream |
single/multi-session

We don't offer the full range of choice for these, and applications make-do
with the closest they have and then add enhancements to make more possible.

-=R


>
> /Tobias
>
>
>
>
>
>
> On Tue, Feb 4, 2014 at 10:11 AM, Peter Thorson <webmaster@zaphoyd.com>
> wrote:
>
> On Feb 4, 2014, at 11:40 , Roberto Peon <fenix@google.com> wrote:
>
>
> On Tue, Feb 4, 2014 at 7:02 AM, Tobias Oberstein <
> tobias.oberstein@tavendo.de> wrote:
> >I'm interested in correctly muxing, demuxing, and loadbalancing of
> streams-of-messages without needing to understand application-layer
> semantics.
> What is the unit of muxing/demuxing/loadbalancing?
>
> If the unit is the SPDY _stream_, then we can tunnel WebSocket with any
> extensions by simply transporting complete
> WebSocket frames inside the payload of SPDY data frames - with no flags
> whatsoever at SPDY header level.
>
> If the unit is the individual application _message_, then understanding of
> application message boundaries is required at SPDY level.
>
> Which unit did you think of?
>
> Messages, since I have to worry about other message-based protocols.
>
>
> >I'm not at all interested in needing to implement multiple separate
> protocols in order to implement the WS API.
> Sure, but it looks like a practical necessity. RFC6455 is out on the
> street and widely implemented.
>
> Well, I'm not interested in implementing it :)
>
>
> >I am uninterested in any extensions or extension frameworks that require
> any extension-specific knowledge in intermediaries.
> Not sure what this means ..
>
> WebSocket extensions are allowed to interleave messages explicitly in
> RFC6455. Intermediaries are not required
> to _understand_ the interleaving. They are required to _not touch_ the
> fragmentation in this case.
>
> A HTTP2.0 intermediary that does it's muxing/demuxing/loadbalancing at the
> SPDY stream level does not need
> to understand anything about tunneled WS.
>
> s/stream/message, and I agree.
> -=R
>
> How do you mux, demux, or load balance an ordered message stream at the
> message level without taking into account application-layer semantics?
>
>

--001a11c3ec923cff8104f1acca3e
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><br><div class=3D"gmail_extra"><br><br><div class=3D"gmail=
_quote">On Tue, Feb 4, 2014 at 11:44 AM, Tobias Oberstein <span dir=3D"ltr"=
>&lt;<a href=3D"mailto:tobias.oberstein@tavendo.de" target=3D"_blank">tobia=
s.oberstein@tavendo.de</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div class=3D"im">&gt;In other words, if a loadbalancer se=
es that the WS stream-of-messages is targeted at an endpoint that can proce=
ss out-of-order messages (by examining the HEADERS frame initializing the s=
tream-of-messages), &gt;then it is free to demux the messages and forward w=
ithout worrying about reordering on the outbound path.<br>

<br>
</div>I can&#39;t follow the whole argumentation.<br>
<br>
Why would an application that does not need an ordered message channel, but=
 is better served by e.g. independent<br>
request/response message exchanges choose an ordered transport protocol in =
the first place?<br>
<br>
And then need some ad-hoc, non-standard / application-specific way of signa=
ling to intermediaries<br>
that in fact it doesn&#39;t rely on the ordering the transport otherwise wo=
uld guarantee?<br></blockquote><div><br></div><div>Neither application sema=
ntics nor the format of the data for an application are defined in a standa=
rd-- they&#39;re defined by the application. The loadbalancer should be abl=
e to help deal with application semantics without having to worry about the=
 format of the data.</div>
<div><br></div><div>HTTP/2 allows for the generic demarcation of message-bo=
undaries, which enables this kind of application-specific loadbalancing wit=
h generic rules that don&#39;t need to vary based on the protocol spoken in=
side the HTTP/2 stream.</div>
<div><br></div><div>=C2=A0<br></div><blockquote class=3D"gmail_quote" style=
=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(20=
4,204,204);border-left-style:solid;padding-left:1ex">
<br>
Such an app should just do HTTP request/response style coms.<br></blockquot=
e><div><br></div><div>Why? Using the WS API would be more efficient. The lo=
adbalancer can execute policy such as &quot;last received wins&quot;: Why b=
other to send back responses for messages that you know are now uninteresti=
ng?</div>
<div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px =
0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-l=
eft-style:solid;padding-left:1ex">
<br>
An app that needs ordering will request an ordered transport, and _not_ lat=
er give up on that again.<br>
<br></blockquote><div><br></div><div>That has not been my experience.</div>=
<div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px =
0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-l=
eft-style:solid;padding-left:1ex">

And it&#39;s no longer a generic intermediary: it must be configured to &qu=
ot;know&quot; that<br>
certain targets are in fact allowed to be reordered, despite the transport =
being ordered<br>
originally.<br></blockquote><div><br></div><div>I disagree: It is completel=
y generic as it allows for any protocol, including application-layer protoc=
ol</div><div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:=
0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);=
border-left-style:solid;padding-left:1ex">

<div class=3D"im"><br>
&gt;Tobias: We loadbalance myriad things, with RPC and API stuff being part=
 of the mix.<br>
<br>
</div>I guess you are talking about protocols _above_ HTTP or WS?<br>
<br>
Curious: which of those protocols choose an ordered message transport, and =
in fact<br>
do _not_ rely on the actual ordering guarantees?<br></blockquote><div><br><=
/div><div><div>You&#39;re making strong assertions about why people use a p=
rotocol, but in fact, people just use the hammer that is available, and pic=
k the shape closest to what they need.</div>
</div><div><br></div><div>Both RPC and API things often don&#39;t care abou=
t careful ordering of requests to responses.</div><div>They&#39;re seeking =
semantics of both WS and HTTP: from WS-like semantics they want message-bas=
ed and guaranteed-single-session and from HTTP-like semantics they want out=
-of-order.</div>
<div>They&#39;re solving an application-layer problem which the transport d=
oesn&#39;t solve on its own.</div><div>Such a &quot;protocol&quot; has not =
been defined because we do a poor job of designing APIs.<br></div><div>
The intent here is to remedy this by thinking about how the protocol intera=
cts with the entire ecosystem, in particular with proxies and loadbalancers=
.</div><div>Why do we have to change the LB code for every protocol? If we =
do this the smart way, then we don&#39;t have to!</div>
<div><br></div><div>=C2=A0<br></div><blockquote class=3D"gmail_quote" style=
=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(20=
4,204,204);border-left-style:solid;padding-left:1ex">
<br>
Since that is the niche we are now talking about.<br></blockquote><div><br>=
</div><div>We&#39;re talking about being able to loadbalance *any* bytestre=
am protocol or *any* stream-of-message protocol. That pretty much covers th=
e gamut.</div>
<div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px =
0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-l=
eft-style:solid;padding-left:1ex">
<br>
In any case: my impression/opinion is, that this is the core of the crux:<b=
r>
<br>
The simplest approach of W3C API plus per-stream muxing/demuxing/loadbalanc=
ing WS over SPDY with 100% WS compatibility would be full tunneling.<br></b=
lockquote><div><br></div><div>It is simple only if you are willing to commi=
t to adding more code ad infinitum to servers, one for each protocol.</div>
<div>That seems like a poor tradeoff, especially given that it doesn&#39;t =
actually offer any more WS compatibility.=C2=A0</div><div><br></div><div>=
=C2=A0<br></div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0=
px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-le=
ft-style:solid;padding-left:1ex">

<br>
The use case not covered by the latter is &quot;_per message_ muxing/demuxi=
ng/loadbalancing&quot;.<br>
<br>
And that &quot;use case&quot; I can&#39;t follow, since it gives up message=
 ordering, and would be better served by plain HTTP anyway ..<br>
<br>
There is no &quot;unordered WebSocket&quot; .. that is HTTP;)<br></blockquo=
te><div><br></div><div>WS here is a protocol, and possibly an API. Neither =
of these specify in-order responses. Certainly both HTTP and WS specify in-=
order delivery. HTTP actually mandates in-order response, which is counter =
to what many applications want and is a HORRIBLE thing for demuxing loadbal=
ancers as it requires potentially infinite buffering. This is a real practi=
cal problem :)</div>
<div><br></div><div>Applications generally want to choose from:</div><div>u=
nordered/ordered | reliable/unreliable | message/bytestream | single/multi-=
session</div><div><br></div><div>We don&#39;t offer the full range of choic=
e for these, and applications make-do with the closest they have and then a=
dd enhancements to make more possible.</div>
<div><br></div><div>-=3DR</div><div>=C2=A0</div><blockquote class=3D"gmail_=
quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-=
color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">
<span class=3D""><font color=3D"#888888"><br>
/Tobias<br>
</font></span><div class=3D""><div class=3D"h5"><br>
<br>
<br>
<br>
<br>
<br>
On Tue, Feb 4, 2014 at 10:11 AM, Peter Thorson &lt;<a href=3D"mailto:webmas=
ter@zaphoyd.com">webmaster@zaphoyd.com</a>&gt; wrote:<br>
<br>
On Feb 4, 2014, at 11:40 , Roberto Peon &lt;<a href=3D"mailto:fenix@google.=
com">fenix@google.com</a>&gt; wrote:<br>
<br>
<br>
On Tue, Feb 4, 2014 at 7:02 AM, Tobias Oberstein=C2=A0&lt;<a href=3D"mailto=
:tobias.oberstein@tavendo.de">tobias.oberstein@tavendo.de</a>&gt;=C2=A0wrot=
e:<br>
&gt;I&#39;m interested in correctly muxing, demuxing, and loadbalancing of =
streams-of-messages without needing to understand application-layer semanti=
cs.<br>
What is the unit of muxing/demuxing/loadbalancing?<br>
<br>
If the unit is the SPDY _stream_, then we can tunnel WebSocket with any ext=
ensions by simply transporting complete<br>
WebSocket frames inside the payload of SPDY data frames - with no flags wha=
tsoever at SPDY header level.<br>
<br>
If the unit is the individual application _message_, then understanding of =
application message boundaries is required at SPDY level.<br>
<br>
Which unit did you think of?<br>
<br>
Messages, since I have to worry about other message-based protocols.<br>
=C2=A0<br>
<br>
&gt;I&#39;m not at all interested in needing to implement multiple separate=
 protocols in order to implement the WS API.<br>
Sure, but it looks like a practical necessity. RFC6455 is out on the street=
 and widely implemented.<br>
<br>
Well, I&#39;m not interested in implementing it :)<br>
=C2=A0<br>
<br>
&gt;I am uninterested in any extensions or extension frameworks that requir=
e any extension-specific knowledge in intermediaries.=C2=A0<br>
Not sure what this means ..<br>
<br>
WebSocket extensions are allowed to interleave messages explicitly in RFC64=
55. Intermediaries are not required<br>
to _understand_ the interleaving. They are required to _not touch_ the frag=
mentation in this case.<br>
<br>
A HTTP2.0 intermediary that does it&#39;s muxing/demuxing/loadbalancing at =
the SPDY stream level does not need<br>
to understand anything about tunneled WS.<br>
<br>
s/stream/message, and I agree.<br>
-=3DR<br>
<br>
How do you mux, demux, or load balance an ordered message stream at the mes=
sage level without taking into account application-layer semantics?<br>
<br>
</div></div></blockquote></div><br></div></div>

--001a11c3ec923cff8104f1acca3e--

From wenboz@google.com  Wed Feb  5 12:04:33 2014
Return-Path: <wenboz@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 25BFD1A01E6 for <hybi@ietfa.amsl.com>; Wed,  5 Feb 2014 12:04:33 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.913
X-Spam-Level: 
X-Spam-Status: No, score=-1.913 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.535, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id ygaOsUgcE68t for <hybi@ietfa.amsl.com>; Wed,  5 Feb 2014 12:04:30 -0800 (PST)
Received: from mail-qc0-x22c.google.com (mail-qc0-x22c.google.com [IPv6:2607:f8b0:400d:c01::22c]) by ietfa.amsl.com (Postfix) with ESMTP id E24671A0197 for <hybi@ietf.org>; Wed,  5 Feb 2014 12:04:29 -0800 (PST)
Received: by mail-qc0-f172.google.com with SMTP id c9so1523362qcz.3 for <hybi@ietf.org>; Wed, 05 Feb 2014 12:04:29 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=xQAbOIR2uwlsVCPjN1qdjRmH5TzdRIhLWaUkPogeXw0=; b=br6eWHmQ+OGKQirk2GkGXwixo9skZx3ckgdG4UGgdc+gMKdwzdf/vBvtZykiMaeFaE l2CZ24ag2x/aY+oHXJZBUzutixdXgZh88/iaJ/IsMNbRn9T63vdlu294IJFdwB0EgBVW 2lRMaimuQQ0ikzzeioShI1SvWVSnB5/+Q+c5IEYIT/fSSPie7Y5GPhBmZUnMSbNUPKtC uKKZs064bn3+4HZuyylEcZgd798/8UUAJQ7EHokXImI/rciH5/fohUX7/nqT2MuO7vyJ eYM4WGueO6dlnXSROkGUXhZke8Kmq7iIZ0/6qs4tIYFwU09ubiEH2pxchULghXQX4bDm ZDuA==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=xQAbOIR2uwlsVCPjN1qdjRmH5TzdRIhLWaUkPogeXw0=; b=Dv3tWA5YghJX6KcE4Yi0mcm8nCpOlj5jRnPEJdd1y5sMK+M6VJ1R+ojecVDl3nDO6F HkKkyCsd6zhfcu1hSGPtl0ta1QHshw2revX2puJ7tCMnszaCtkU0v7DZ9Jgyb+mJQafX G7h9MXLBw5kG+fb7SmOjcnJIa4KHRz4vkh4SHqf4qKu96hzJDOz3mEAGIM2467MVwJ+/ ZUT0b9736P5FOYdthxIyMyCJ53+RHqCEVK94ThFgD9onwyg66V+9L4V/OqF/lrdgrwD2 Ayn56Dol8xVUxFiMbFIdCneeOK2ACJopM8n7lCbUbMBLGxGOUrEFxqG2XuJ45YCb/d+R 5Zeg==
X-Gm-Message-State: ALoCoQkuniv2eX1l2kDNJHywUU9VZ9n4Jv2dZTHCUc3YCh/buzDpsNkuxZEK4AK49/2GbDICQhFLQa5ieutvDQXE4KPZUqYHsLRDW86lWpmRIB4G8GsObP2o7N/toxQ921FGJAK6d3BPabFuwPk5e+oXLatoE/HuZweEOJYbeXGKWuYZyWb+uo7wYQ9+2hGMzLUOem12imgv
MIME-Version: 1.0
X-Received: by 10.224.40.130 with SMTP id k2mr5777619qae.91.1391630668996; Wed, 05 Feb 2014 12:04:28 -0800 (PST)
Received: by 10.229.147.142 with HTTP; Wed, 5 Feb 2014 12:04:28 -0800 (PST)
In-Reply-To: <634914A010D0B943A035D226786325D4446C29584C@EXVMBX020-12.exch020.serverdata.net>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <CAGzyod6H-To9EhaPqW5C0c9xyy5XnRg_851GuT3q=X+KoXe7Wg@mail.gmail.com> <0533B942-C1DA-4BFE-BE80-4722C0BD740D@zaphoyd.com> <CAGzyod7xtCZqUvcb0gX0PYKyikyd4kKc=kOzwzgfAmUFExCq3g@mail.gmail.com> <634914A010D0B943A035D226786325D4446C29584C@EXVMBX020-12.exch020.serverdata.net>
Date: Wed, 5 Feb 2014 12:04:28 -0800
Message-ID: <CAD3-0rMouRJ_GEo557COf827zDpKVEtPw68BKfwVt2kgr7gRzA@mail.gmail.com>
From: Wenbo Zhu <wenboz@google.com>
To: Tobias Oberstein <tobias.oberstein@tavendo.de>
Content-Type: multipart/alternative; boundary=047d7bf1651caf411f04f1ae4302
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 05 Feb 2014 20:04:33 -0000

--047d7bf1651caf411f04f1ae4302
Content-Type: text/plain; charset=UTF-8

On Tue, Feb 4, 2014 at 11:44 AM, Tobias Oberstein <
tobias.oberstein@tavendo.de> wrote:

> >In other words, if a loadbalancer sees that the WS stream-of-messages is
> targeted at an endpoint that can process out-of-order messages (by
> examining the HEADERS frame initializing the stream-of-messages), >then it
> is free to demux the messages and forward without worrying about reordering
> on the outbound path.
>
> I can't follow the whole argumentation.
>
> Why would an application that does not need an ordered message channel,
> but is better served by e.g. independent
> request/response message exchanges choose an ordered transport protocol in
> the first place?
>
> And then need some ad-hoc, non-standard / application-specific way of
> signaling to intermediaries
> that in fact it doesn't rely on the ordering the transport otherwise would
> guarantee?
>
> Such an app should just do HTTP request/response style coms.
>
> An app that needs ordering will request an ordered transport, and _not_
> later give up on that again.
>
> And it's no longer a generic intermediary: it must be configured to "know"
> that
> certain targets are in fact allowed to be reordered, despite the transport
> being ordered
> originally.
>
> >Tobias: We loadbalance myriad things, with RPC and API stuff being part
> of the mix.
>
> I guess you are talking about protocols _above_ HTTP or WS?
>
> Curious: which of those protocols choose an ordered message transport, and
> in fact
> do _not_ rely on the actual ordering guarantees?
>
> Since that is the niche we are now talking about.
>
> In any case: my impression/opinion is, that this is the core of the crux:
>
> The simplest approach of W3C API plus per-stream
> muxing/demuxing/loadbalancing WS over SPDY with 100% WS compatibility would
> be full tunneling.
>
So the question is whether we tunnel the W3C API or the existing wire
protocol, which happens to exist only to support the API (i.e., not the
case for HTTP).

The load-balancing requirement only concerns the API (in the form of URLs,
headers), or more accurately the message delivery. As for HTTP,
load-balancing often has to address session stickiness, which may or may
not imply ordering, independently of the transport being RPC.

- Wenbo


> The use case not covered by the latter is "_per message_
> muxing/demuxing/loadbalancing".
>
> And that "use case" I can't follow, since it gives up message ordering,
> and would be better served by plain HTTP anyway ..
>
> There is no "unordered WebSocket" .. that is HTTP;)
>
> /Tobias
>
>
>
>
>
>
> On Tue, Feb 4, 2014 at 10:11 AM, Peter Thorson <webmaster@zaphoyd.com>
> wrote:
>
> On Feb 4, 2014, at 11:40 , Roberto Peon <fenix@google.com> wrote:
>
>
> On Tue, Feb 4, 2014 at 7:02 AM, Tobias Oberstein <
> tobias.oberstein@tavendo.de> wrote:
> >I'm interested in correctly muxing, demuxing, and loadbalancing of
> streams-of-messages without needing to understand application-layer
> semantics.
> What is the unit of muxing/demuxing/loadbalancing?
>
> If the unit is the SPDY _stream_, then we can tunnel WebSocket with any
> extensions by simply transporting complete
> WebSocket frames inside the payload of SPDY data frames - with no flags
> whatsoever at SPDY header level.
>
> If the unit is the individual application _message_, then understanding of
> application message boundaries is required at SPDY level.
>
> Which unit did you think of?
>
> Messages, since I have to worry about other message-based protocols.
>
>
> >I'm not at all interested in needing to implement multiple separate
> protocols in order to implement the WS API.
> Sure, but it looks like a practical necessity. RFC6455 is out on the
> street and widely implemented.
>
> Well, I'm not interested in implementing it :)
>
>
> >I am uninterested in any extensions or extension frameworks that require
> any extension-specific knowledge in intermediaries.
> Not sure what this means ..
>
> WebSocket extensions are allowed to interleave messages explicitly in
> RFC6455. Intermediaries are not required
> to _understand_ the interleaving. They are required to _not touch_ the
> fragmentation in this case.
>
> A HTTP2.0 intermediary that does it's muxing/demuxing/loadbalancing at the
> SPDY stream level does not need
> to understand anything about tunneled WS.
>
> s/stream/message, and I agree.
> -=R
>
> How do you mux, demux, or load balance an ordered message stream at the
> message level without taking into account application-layer semantics?
>
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi
>

--047d7bf1651caf411f04f1ae4302
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><br><div class=3D"gmail_extra"><br><br><div class=3D"gmail=
_quote">On Tue, Feb 4, 2014 at 11:44 AM, Tobias Oberstein <span dir=3D"ltr"=
>&lt;<a href=3D"mailto:tobias.oberstein@tavendo.de" target=3D"_blank">tobia=
s.oberstein@tavendo.de</a>&gt;</span> wrote:<br>


<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div>&gt;In other words, if a loadbalancer s=
ees that the WS stream-of-messages is targeted at an endpoint that can proc=
ess out-of-order messages (by examining the HEADERS frame initializing the =
stream-of-messages), &gt;then it is free to demux the messages and forward =
without worrying about reordering on the outbound path.<br>



<br>
</div>I can&#39;t follow the whole argumentation.<br>
<br>
Why would an application that does not need an ordered message channel, but=
 is better served by e.g. independent<br>
request/response message exchanges choose an ordered transport protocol in =
the first place?<br>
<br>
And then need some ad-hoc, non-standard / application-specific way of signa=
ling to intermediaries<br>
that in fact it doesn&#39;t rely on the ordering the transport otherwise wo=
uld guarantee?<br>
<br>
Such an app should just do HTTP request/response style coms.<br>
<br>
An app that needs ordering will request an ordered transport, and _not_ lat=
er give up on that again.<br>
<br>
And it&#39;s no longer a generic intermediary: it must be configured to &qu=
ot;know&quot; that<br>
certain targets are in fact allowed to be reordered, despite the transport =
being ordered<br>
originally.<br>
<div><br>
&gt;Tobias: We loadbalance myriad things, with RPC and API stuff being part=
 of the mix.<br>
<br>
</div>I guess you are talking about protocols _above_ HTTP or WS?<br>
<br>
Curious: which of those protocols choose an ordered message transport, and =
in fact<br>
do _not_ rely on the actual ordering guarantees?<br>
<br>
Since that is the niche we are now talking about.<br>
<br>
In any case: my impression/opinion is, that this is the core of the crux:<b=
r>
<br>
The simplest approach of W3C API plus per-stream muxing/demuxing/loadbalanc=
ing WS over SPDY with 100% WS compatibility would be full tunneling.<br></b=
lockquote><div>So the question is whether we tunnel the W3C API or the exis=
ting wire protocol, which happens to exist only to support the API (i.e., n=
ot the case for HTTP).</div>

<div><br></div><div>The load-balancing requirement only concerns the API (i=
n the form of URLs, headers), or more accurately the message delivery. As f=
or HTTP, load-balancing often has to address session stickiness, which may =
or may not imply ordering, independently of the transport being RPC.</div>
<div><br></div><div>- Wenbo</div><div><br>
</div><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-l=
eft:1px #ccc solid;padding-left:1ex">

<br>
The use case not covered by the latter is &quot;_per message_ muxing/demuxi=
ng/loadbalancing&quot;.<br>
<br>
And that &quot;use case&quot; I can&#39;t follow, since it gives up message=
 ordering, and would be better served by plain HTTP anyway ..<br>
<br>
There is no &quot;unordered WebSocket&quot; .. that is HTTP;)<br>
<span><font color=3D"#888888"><br>
/Tobias<br>
</font></span><div><br>
<br>
<br>
<br>
<br>
<br>
On Tue, Feb 4, 2014 at 10:11 AM, Peter Thorson &lt;<a href=3D"mailto:webmas=
ter@zaphoyd.com" target=3D"_blank">webmaster@zaphoyd.com</a>&gt; wrote:<br>
<br>
On Feb 4, 2014, at 11:40 , Roberto Peon &lt;<a href=3D"mailto:fenix@google.=
com" target=3D"_blank">fenix@google.com</a>&gt; wrote:<br>
<br>
<br>
On Tue, Feb 4, 2014 at 7:02 AM, Tobias Oberstein=C2=A0&lt;<a href=3D"mailto=
:tobias.oberstein@tavendo.de" target=3D"_blank">tobias.oberstein@tavendo.de=
</a>&gt;=C2=A0wrote:<br>
&gt;I&#39;m interested in correctly muxing, demuxing, and loadbalancing of =
streams-of-messages without needing to understand application-layer semanti=
cs.<br>
What is the unit of muxing/demuxing/loadbalancing?<br>
<br>
If the unit is the SPDY _stream_, then we can tunnel WebSocket with any ext=
ensions by simply transporting complete<br>
WebSocket frames inside the payload of SPDY data frames - with no flags wha=
tsoever at SPDY header level.<br>
<br>
If the unit is the individual application _message_, then understanding of =
application message boundaries is required at SPDY level.<br>
<br>
Which unit did you think of?<br>
<br>
Messages, since I have to worry about other message-based protocols.<br>
=C2=A0<br>
<br>
&gt;I&#39;m not at all interested in needing to implement multiple separate=
 protocols in order to implement the WS API.<br>
Sure, but it looks like a practical necessity. RFC6455 is out on the street=
 and widely implemented.<br>
<br>
Well, I&#39;m not interested in implementing it :)<br>
=C2=A0<br>
<br>
&gt;I am uninterested in any extensions or extension frameworks that requir=
e any extension-specific knowledge in intermediaries.=C2=A0<br>
Not sure what this means ..<br>
<br>
WebSocket extensions are allowed to interleave messages explicitly in RFC64=
55. Intermediaries are not required<br>
to _understand_ the interleaving. They are required to _not touch_ the frag=
mentation in this case.<br>
<br>
A HTTP2.0 intermediary that does it&#39;s muxing/demuxing/loadbalancing at =
the SPDY stream level does not need<br>
to understand anything about tunneled WS.<br>
<br>
s/stream/message, and I agree.<br>
-=3DR<br>
<br>
How do you mux, demux, or load balance an ordered message stream at the mes=
sage level without taking into account application-layer semantics?<br>
<br>
</div><div><div>_______________________________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
</div></div></blockquote></div><br></div></div>

--047d7bf1651caf411f04f1ae4302--

From yhirano@google.com  Wed Feb  5 22:04:10 2014
Return-Path: <yhirano@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id CA24F1A02B2 for <hybi@ietfa.amsl.com>; Wed,  5 Feb 2014 22:04:10 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.913
X-Spam-Level: 
X-Spam-Status: No, score=-1.913 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.535, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 2I_e0NPu0JYF for <hybi@ietfa.amsl.com>; Wed,  5 Feb 2014 22:04:07 -0800 (PST)
Received: from mail-yh0-x22d.google.com (mail-yh0-x22d.google.com [IPv6:2607:f8b0:4002:c01::22d]) by ietfa.amsl.com (Postfix) with ESMTP id 9DF801A0030 for <hybi@ietf.org>; Wed,  5 Feb 2014 22:04:07 -0800 (PST)
Received: by mail-yh0-f45.google.com with SMTP id i57so1529850yha.4 for <hybi@ietf.org>; Wed, 05 Feb 2014 22:04:06 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=59r5L5XhWeKKZYoCEY2Ag+4BbV/7drNvwO78uJLMSec=; b=p4ATHQ7mknd+77fR1/QAxIckM5f7pUGRJ1q/QSy5zlFL4nBjYQHNHwsNcmwI3/StYB yDw8Dc8YDJECNRg94EUZiPTWYVXpE+AfyX6JW6cEzf3itwwfDKuJJ8OWgEjMUGmpqVuJ M+0DeT8wp7fN2wKWI7QrgNDg2GkgvVhWhzdp5EMoUwnd8YbSQR+sSORDx48wiFGKCX3a JLjxdrdwGiasrlT69gLTmtj6CuMuRPkWn4NAGr5u8r21tOzb6KtcxbSNtEEsPVH9pDcz MvW4CYjGZzSdgVdhdLuLfiS5CFtj9RZhd/ITt3fcH8l4kviqB+6HI9FT1uSiBIgkX/hh 5H2w==
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=chromium.org; s=google; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=59r5L5XhWeKKZYoCEY2Ag+4BbV/7drNvwO78uJLMSec=; b=g2a00WTnkrJcrWp2mFzFBAA/iYSSCuHvJt1Ev9O1bYDzgeE3u991gWGyEL0nK3sfqy E2r3FLGqsm1A+VNByNc66ZQF4sV7NP+v/sx2AlV2C31AecoQwydH+a+JH+9zsJs7BRvR auMCiQncoTJndwo42oi6i0ASr7fq+ERuTLe1g=
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:sender:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=59r5L5XhWeKKZYoCEY2Ag+4BbV/7drNvwO78uJLMSec=; b=hQicnG4Fd9ixHO8VkByCqam6itDQ+VeZYxCF5RAcbn7pcgvnDoAiNvGxUeur69l16c Bxd+B2PdDBk0ADFh+tKVrW9TDXMmUevbqkNxwX0AZJjv4m/xJZe6SDAccnlZsFot9yog xvXulyK2lApLVZFYjl+s8vPTeSUE3TUnRW3xdK3oTcvtUIrTY6Fxd8UcBBkTonE8GFhL 9x3z1v8LG9B+Pi7P0nroaQJ73HGJ+2tf5/h2ImxRbwIibcVPnaKkvcXXJvIf5L1m1sA/ x9txCFhv3MZuqfjD+caqMA1IMr2YIEbzh6jjrs1m2pQsWPp651GBFjrzigwHE0lP+FJe rzpw==
X-Gm-Message-State: ALoCoQm6V+7ki2eJr5ugkojcIlIvqTMs3AdlGRjYC4ALT5yNCKL8PpYwbzlTTzVT0O6IQNXVg6aq+qOwa5aWHIcuEmGAdVNft7DRgMBO4DOC51wpf8hrz4ILpvMVp7gFfgjqrJw75Uz6BeWpTRMrqz2WLIzuFYRQVfv8KtR0eikCUZ0GqE+yWNewm7Az7PdZ9ndSWAlv7von
MIME-Version: 1.0
X-Received: by 10.236.180.33 with SMTP id i21mr4388242yhm.101.1391666646518; Wed, 05 Feb 2014 22:04:06 -0800 (PST)
Sender: yhirano@google.com
Received: by 10.170.135.22 with HTTP; Wed, 5 Feb 2014 22:04:06 -0800 (PST)
In-Reply-To: <740FF040-F1E9-43E6-ACDE-0014FE4D012F@zaphoyd.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <3274C7C2-41D4-49AD-971E-DA3D85898BC2@zaphoyd.com> <CABihn6GGezUU2aSboK0wsxYX3TBWTz5Av6-ZFJTPy6PQ50LrTQ@mail.gmail.com> <740FF040-F1E9-43E6-ACDE-0014FE4D012F@zaphoyd.com>
Date: Thu, 6 Feb 2014 15:04:06 +0900
X-Google-Sender-Auth: YCd3WkcaAs3V_2GFAeKJK4iPCsI
Message-ID: <CABihn6HM_yXSB9RapNEQEG0_9OYfpHG3uH+=wnDwRJAFbh9EJA@mail.gmail.com>
From: Yutaka Hirano <yhirano@chromium.org>
To: Peter Thorson <webmaster@zaphoyd.com>
Content-Type: multipart/alternative; boundary=20cf305e23991ca60b04f1b6a416
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 06 Feb 2014 06:04:11 -0000

--20cf305e23991ca60b04f1b6a416
Content-Type: text/plain; charset=UTF-8

Peter,

I don't have a strong preference for either of these options. I do have a
> strong preference that we not do something in between. RFC6455
> compatibility is useful for me because it allows sharing
> code/infrastructure and backwards compatibility. Not because the specific
> features of RFC6455 are preserved.
>


> That is the reason I think adding MSG_DONE to WS-TUNNEL doesn't make
> sense. It puts us in a situation where *all* HTTP/2.0 implementations must
> implement a bunch of message related functionality that is already in
> RFC6455.

I still disagree.
As I said before, enabling intermediaries to transfer WS over HTTP/2.0
*without an RFC6455 parser* is important, I think.

No matter what protocol we choose, we must observe the HTTP/2.0 framing
protocol, which means:
  - The size of a frame must not exceed the fixed limit.
  - We must set and respect the END_STREAM flag appropriately.
  - ...
Adding one rule (MSG_DONE) is harmless, and I think it has a merit.



On Thu, Feb 6, 2014 at 12:46 AM, Peter Thorson <webmaster@zaphoyd.com>wrote:

>
> On Feb 5, 2014, at 4:56, Yutaka Hirano <yhirano@chromium.org> wrote:
>
> Let me talk about PlanA, WS-TUNNEL with MSG_DONE.
>
> Roberto says he is interested in loadbalancing ***without needing to
> understand application-layer semantics***. (1)
> I say we can't detect message boundary ***without understanding WebSocket
> level semantics***. (2)
> You say we can detect message boundary ***with the RFC6455 parser***. (3)
>
> Year, I think (2) and (3) are both true.
>
> I thought that Roberto didn't want to implement an RFC6455 paser for the
> loadbalancer.
> If it is wrong: Yay! PlanA (with MSG_DONE) seems good.
>
> If it is correct:
> I still think MSG_DONE is insufficient for the loadbalancing.
> Imagine a WebSocket connection consisting of three WebSocket frames, f1
> [without FIN], f2 [with FIN] and f3 [with FIN].
> Assume that there is no extensions.
> Assume that one HTTP/2.0 frame is mapped to one WebSocket frame for
> simplicity.
> Let the HTTP/2.0 frames be hf1 [without MSG_DONE], hf2 [with MSG_DONE] and
> hf3 [with MSG_DONE].
> If the loadbalancer does NOT understand RFC6455, there are two
> possibilities,
>  - A message consists of (hf1, hf2) and another message consists of (hf3).
>  - A data message consists of (hf1, hf3) and a control message consists of
> (hf2).
> Hence it fails to recognize messages.
>
> Note: For enabling intermediaries to transfer WS over HTTP/2.0 data
> without understanding RFC6455, MSG_DONE is sufficient.
>
>
> Let me define "Plan D" more precisely:
> HTTP/2.0 DATA frames are used as a dumb transport for an RFC6455 byte
> stream. There is no 1:1 frame mapping between WS frames and HTTP/2.0
> frames. A browser could send two WebSocket frames back to back in one
> HTTP/2.0 frame. An intermediary could split this frame evenly into four
> HTTP/2.0 frames each containing half of an RFC6455 frame. The remote
> endpoint would accept an HTTP/2.0 frame, feed its payload bytes straight to
> an RFC6455 parser which would handle the rest. A MSG_DONE flag is not
> required at all even for full ignorant intermediary compatibility.
>
> Benefits:
> - 100% RFC6455 compatibility present and future
> - Simple implementation:
> -- Nothing needs to be added to HTTP/2.0 or RFC6455.
> -- Browsers leverage existing RFC6455 parsers and just add logic to choose
> whether to use an HTTP/2.0 stream or a raw TCP stream. Data centers demux
> the streams into RFC6455 message processors.
> -- Everyone gets trivial backwards compatibility with existing RFC6455
> software and hardware, both on the client and server side.
> - Intermediaries that don't care can safely ignore or re-fragment however
> they like.
>
> Drawbacks
> - Requires intermediaries and endpoints that want to understand messages
> to read RFC6455. (I don't think this is an issue, but it seems important to
> Roberto)
> - Retains unnecessary/redundant features (masking, control messages, frame
> length fields)
>
> I think this is the best option if we value RFC6455 compatibility or
> desire ease of integration into existing software and environments.
> Otherwise, I would prefer that we go the "Plan B extreme" route that
> ditches RFC6455 entirely and builds a completely native HTTP/2.0 message
> stream. MSG_DONE/type bits, no masking, no opcode, no rsv, no control
> frames.
>
> Benefits:
> - HTTP/2.0 software doesn't need to know about RFC6455 unless it wants
> backwards compatibility with non-HTTP/2.0 software
> - Doesn't re-create lots of irrelevant/duplicated features from RFC6455
> - Optimal wire efficiency.
>
> Drawbacks:
> - Intermediaries need to implement "HTTP/2.0 messaging". (it is simple
> enough now though that that shouldn't be a great hardship.)
> - Limited RFC6455 extension compatibility.
>
> I don't have a strong preference for either of these options. I do have a
> strong preference that we not do something in between. RFC6455
> compatibility is useful for me because it allows sharing
> code/infrastructure and backwards compatibility. Not because the specific
> features of RFC6455 are preserved.
>
> That is the reason I think adding MSG_DONE to WS-TUNNEL doesn't make
> sense. It puts us in a situation where *all* HTTP/2.0 implementations must
> implement a bunch of message related functionality that is already in
> RFC6455. As Yukata points out MSG_DONE alone is not sufficient to detect
> messages in a tunneled RFC6455 stream. You'd need to throw in an opcode as
> well. At this point we end up re-implementing half of RFC6455 in HTTP/2.0
> in order to avoid using RFC6455 in a product that already has an RFC6455
> parser?? The only gain I see is certain types of limited and specialized
> intermediaries can drop RFC6455. That doesn't seem like a good tradeoff
> given that we could just choose Plan B and get rid of RFC6455 for everybody.
>
> Regarding message based load balancing for WebSocket:
> I agree with Tobias that this doesn't make sense. An application that
> wants to send independent messages can send HTTP requests. Such requests
> can be easily muxed, demuxed, or load balanced. Each message has headers to
> store application specific information necessary to correctly load balance.
> This can all be done efficiently with the functionality HTTP/2.0 provides
> already.
>
>

--20cf305e23991ca60b04f1b6a416
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Peter,<div><br></div><div><blockquote class=3D"gmail_quote=
" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color=
:rgb(204,204,204);border-left-style:solid;padding-left:1ex">I don&#39;t hav=
e a strong preference for either of these options. I do have a strong prefe=
rence that we not do something in between. RFC6455 compatibility is useful =
for me because it allows sharing code/infrastructure and backwards compatib=
ility. Not because the specific features of RFC6455 are preserved.<br>
</blockquote><div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"ma=
rgin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,=
204);border-left-style:solid;padding-left:1ex">That is the reason I think a=
dding MSG_DONE to WS-TUNNEL doesn&#39;t make sense. It puts us in a situati=
on where *all* HTTP/2.0 implementations must implement a bunch of message r=
elated functionality that is already in RFC6455.</blockquote>
</div><div>I still disagree.<br></div><div>As I said before, enabling inter=
mediaries to transfer WS over HTTP/2.0 *without an RFC6455 parser* is impor=
tant, I think.</div><div><br></div><div>No matter what protocol we choose, =
we must observe the HTTP/2.0 framing protocol, which means:</div>
<div>=C2=A0 - The size of a frame must not exceed the fixed limit.</div><di=
v>=C2=A0 - We must set and respect the END_STREAM flag appropriately.</div>=
<div>=C2=A0 - ...</div><div>Adding one rule (MSG_DONE) is harmless, and I t=
hink it has a merit.</div>
<div><br></div></div><div class=3D"gmail_extra"><br><br><div class=3D"gmail=
_quote">On Thu, Feb 6, 2014 at 12:46 AM, Peter Thorson <span dir=3D"ltr">&l=
t;<a href=3D"mailto:webmaster@zaphoyd.com" target=3D"_blank">webmaster@zaph=
oyd.com</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div style=3D"word-wrap:break-word"><div cla=
ss=3D"im"><br><div><div>On Feb 5, 2014, at 4:56, Yutaka Hirano &lt;<a href=
=3D"mailto:yhirano@chromium.org" target=3D"_blank">yhirano@chromium.org</a>=
&gt; wrote:</div>
<br><blockquote type=3D"cite"><div dir=3D"ltr"><div style=3D"font-family:ar=
ial,sans-serif;font-size:13px"><blockquote class=3D"gmail_quote" style=3D"m=
argin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204=
,204);border-left-style:solid;padding-left:1ex">
<span style=3D"color:rgb(34,34,34);font-family:arial;font-size:small">Let m=
e talk about PlanA, WS-TUNNEL with MSG_DONE.</span></blockquote></div><div>=
Roberto says he is interested in loadbalancing ***without needing to unders=
tand application-layer semantics***. (1)</div>

<div>I say we can&#39;t detect message boundary ***without understanding We=
bSocket level semantics***. (2)</div><div>You say we can detect message bou=
ndary ***with the RFC6455 parser***. (3)</div><div><br></div><div>Year, I t=
hink (2) and (3) are both true.</div>

<div><br></div><div>I thought that Roberto didn&#39;t want to implement an =
RFC6455 paser for the loadbalancer.</div><div>If it is wrong: Yay! PlanA (w=
ith MSG_DONE) seems good.<br></div><div><div><br></div><div>If it is correc=
t:</div>

<div>I still think MSG_DONE is insufficient for the loadbalancing.</div></d=
iv><div>Imagine a WebSocket connection consisting of three WebSocket frames=
, f1 [without FIN], f2 [with FIN] and f3 [with FIN].<br></div><div>Assume t=
hat there is no extensions.</div>

<div>Assume that one HTTP/2.0 frame is mapped to one WebSocket frame for si=
mplicity.</div><div>Let the HTTP/2.0 frames be hf1 [without MSG_DONE], hf2 =
[with MSG_DONE] and hf3 [with MSG_DONE].</div><div>If the loadbalancer does=
 NOT understand RFC6455, there are two possibilities,</div>

<div>=C2=A0- A message consists of (hf1, hf2) and another message consists =
of (hf3).</div><div>=C2=A0- A data message consists of (hf1, hf3) and a con=
trol message consists of (hf2).</div><div>Hence it fails to recognize messa=
ges.</div>

<div><br></div><div>Note: For enabling intermediaries to transfer WS over H=
TTP/2.0 data without understanding RFC6455, MSG_DONE is sufficient.=C2=A0</=
div></div></blockquote><br></div></div><div><div>Let me define &quot;Plan D=
&quot; more precisely:</div>
<div>HTTP/2.0 DATA frames are used as a dumb transport for an RFC6455 byte =
stream. There is no 1:1 frame mapping between WS frames and HTTP/2.0 frames=
. A browser could send two WebSocket frames back to back in one HTTP/2.0 fr=
ame. An intermediary could split this frame evenly into four HTTP/2.0 frame=
s each containing half of an RFC6455 frame. The remote endpoint would accep=
t an HTTP/2.0 frame, feed its payload bytes straight to an RFC6455 parser w=
hich would handle the rest. A MSG_DONE flag is not required at all even for=
 full ignorant intermediary compatibility.</div>
<div><br></div><div>Benefits:</div><div>- 100% RFC6455 compatibility presen=
t and future</div><div>- Simple implementation:</div><div>-- Nothing needs =
to be added to HTTP/2.0 or RFC6455.=C2=A0</div><div>-- Browsers leverage ex=
isting RFC6455 parsers and just add logic to choose whether to use an HTTP/=
2.0 stream or a raw TCP stream. Data centers demux the streams into RFC6455=
 message processors.</div>
<div>-- Everyone gets trivial backwards compatibility with existing RFC6455=
 software and hardware, both on the client and server side.</div><div>- Int=
ermediaries that don&#39;t care can safely ignore or re-fragment however th=
ey like.</div>
<div><br></div><div>Drawbacks</div><div>- Requires intermediaries and endpo=
ints that want to understand messages to read RFC6455. (I don&#39;t think t=
his is an issue, but it seems important to Roberto)</div><div>- Retains unn=
ecessary/redundant features (masking, control messages, frame length fields=
)</div>
<div><br></div><div>I think this is the best option if we value RFC6455 com=
patibility or desire ease of integration into existing software and environ=
ments. Otherwise, I would prefer that we go the &quot;Plan B extreme&quot; =
route that ditches RFC6455 entirely and builds a completely native HTTP/2.0=
 message stream. MSG_DONE/type bits, no masking, no opcode, no rsv, no cont=
rol frames.</div>
<div><br></div><div>Benefits:</div><div>- HTTP/2.0 software doesn&#39;t nee=
d to know about RFC6455 unless it wants backwards compatibility with non-HT=
TP/2.0 software</div><div>- Doesn&#39;t re-create lots of irrelevant/duplic=
ated features from RFC6455</div>
<div>- Optimal wire efficiency.</div><div><br></div><div>Drawbacks:</div><d=
iv>- Intermediaries need to implement &quot;HTTP/2.0 messaging&quot;. (it i=
s simple enough now though that that shouldn&#39;t be a great hardship.)</d=
iv>
<div>- Limited RFC6455 extension compatibility.</div><div><br></div><div>I =
don&#39;t have a strong preference for either of these options. I do have a=
 strong preference that we not do something in between. RFC6455 compatibili=
ty is useful for me because it allows sharing code/infrastructure and backw=
ards compatibility. Not because the specific features of RFC6455 are preser=
ved.</div>
<div><br></div><div>That is the reason I think adding MSG_DONE to WS-TUNNEL=
 doesn&#39;t make sense. It puts us in a situation where *all* HTTP/2.0 imp=
lementations must implement a bunch of message related functionality that i=
s already in RFC6455. As Yukata points out MSG_DONE alone is not sufficient=
 to detect messages in a tunneled RFC6455 stream. You&#39;d need to throw i=
n an opcode as well. At this point we end up re-implementing half of RFC645=
5 in HTTP/2.0 in order to avoid using RFC6455 in a product that already has=
 an RFC6455 parser?? The only gain I see is certain types of limited and sp=
ecialized intermediaries can drop RFC6455. That doesn&#39;t seem like a goo=
d tradeoff given that we could just choose Plan B and get rid of RFC6455 fo=
r everybody.</div>
<div><br></div><div>Regarding message based load balancing for WebSocket:</=
div><div>I agree with Tobias that this doesn&#39;t make sense. An applicati=
on that wants to send independent messages can send HTTP requests. Such req=
uests can be easily muxed, demuxed, or load balanced. Each message has head=
ers to store application specific information necessary to correctly load b=
alance. This can all be done efficiently with the functionality HTTP/2.0 pr=
ovides already.</div>
</div><br></div></blockquote></div><br></div>

--20cf305e23991ca60b04f1b6a416--

From tobias.oberstein@tavendo.de  Thu Feb  6 04:10:04 2014
Return-Path: <tobias.oberstein@tavendo.de>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id EDB591A03A3 for <hybi@ietfa.amsl.com>; Thu,  6 Feb 2014 04:10:03 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.9
X-Spam-Level: 
X-Spam-Status: No, score=-1.9 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, RCVD_IN_DNSWL_NONE=-0.0001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id RM1ky-luM0Z5 for <hybi@ietfa.amsl.com>; Thu,  6 Feb 2014 04:10:01 -0800 (PST)
Received: from EXHUB020-4.exch020.serverdata.net (exhub020-4.exch020.serverdata.net [206.225.164.31]) by ietfa.amsl.com (Postfix) with ESMTP id A60591A0393 for <hybi@ietf.org>; Thu,  6 Feb 2014 04:10:00 -0800 (PST)
Received: from EXVMBX020-12.exch020.serverdata.net ([169.254.3.11]) by EXHUB020-4.exch020.serverdata.net ([206.225.164.31]) with mapi; Thu, 6 Feb 2014 04:09:59 -0800
From: Tobias Oberstein <tobias.oberstein@tavendo.de>
To: Yutaka Hirano <yhirano@chromium.org>, Peter Thorson <webmaster@zaphoyd.com>
Date: Thu, 6 Feb 2014 04:09:57 -0800
Thread-Topic: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
Thread-Index: Ac8jAUPJpW/6UFa+Q4CrS1oxZiPORAALFC7w
Message-ID: <634914A010D0B943A035D226786325D4446C295D7E@EXVMBX020-12.exch020.serverdata.net>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <3274C7C2-41D4-49AD-971E-DA3D85898BC2@zaphoyd.com> <CABihn6GGezUU2aSboK0wsxYX3TBWTz5Av6-ZFJTPy6PQ50LrTQ@mail.gmail.com> <740FF040-F1E9-43E6-ACDE-0014FE4D012F@zaphoyd.com> <CABihn6HM_yXSB9RapNEQEG0_9OYfpHG3uH+=wnDwRJAFbh9EJA@mail.gmail.com>
In-Reply-To: <CABihn6HM_yXSB9RapNEQEG0_9OYfpHG3uH+=wnDwRJAFbh9EJA@mail.gmail.com>
Accept-Language: de-DE, en-US
Content-Language: de-DE
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
acceptlanguage: de-DE, en-US
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: base64
MIME-Version: 1.0
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of	WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 06 Feb 2014 12:10:04 -0000
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==

From yhirano@google.com  Thu Feb  6 04:32:44 2014
Return-Path: <yhirano@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 900AA1A03A3 for <hybi@ietfa.amsl.com>; Thu,  6 Feb 2014 04:32:44 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.913
X-Spam-Level: 
X-Spam-Status: No, score=-1.913 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.535, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id YYoTy6oVV3uS for <hybi@ietfa.amsl.com>; Thu,  6 Feb 2014 04:32:40 -0800 (PST)
Received: from mail-qc0-x234.google.com (mail-qc0-x234.google.com [IPv6:2607:f8b0:400d:c01::234]) by ietfa.amsl.com (Postfix) with ESMTP id BB4E61A0385 for <hybi@ietf.org>; Thu,  6 Feb 2014 04:32:39 -0800 (PST)
Received: by mail-qc0-f180.google.com with SMTP id i17so2911528qcy.25 for <hybi@ietf.org>; Thu, 06 Feb 2014 04:32:38 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=MBxs4jipkcELkAmZXJjxVINhUbQ24ZB5sa6lH83fF0A=; b=Hqts5xwaJtTIU3KXP/gsgIuvGydlniotaYRg/QtXZ+cNM7RXaq8iTHA7BL2DmfQCrs LgK36xQaMGVHOg2KteRu8iNm1jgMwrh5CwJ6e8KK0CJldpkXYRVOlnOg3MXuCylD3U60 Wxj54NNAcNocYBq+PL3o7CLRsCM8GGrnoO078ezjSDL/FsDhzITyDhpequkdW7ZoCyAx i0997ZQXUQZGTSOFU2GhWvZWSF+Xlmem9vCAa3N9dTuB6k8b7bYfz+WdCuPT1IWrYMyo 4JMk8u1KYI8ryl3QctX4uKvOUdd9a8nSoNaYI4e313ghfzPCsfGpLu1Uepjrs9snKSfO CZ7g==
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=chromium.org; s=google; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=MBxs4jipkcELkAmZXJjxVINhUbQ24ZB5sa6lH83fF0A=; b=LEjhGtFlSBTpNqCdJ2iMEx3XpJexyKPVLwNt8G2harmPEwuiOwqA2QkHNz6HSwKEsK /yPzK4syonvhM0lQ0au2IvFypDY0hpR2Dhl2/V9OxYRvIsRKhVq1DXpbKgFp26mWeHt+ Rpt5mSlnTiZIF82LQNpe3F3CE2TdcGtT7JFsA=
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:sender:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=MBxs4jipkcELkAmZXJjxVINhUbQ24ZB5sa6lH83fF0A=; b=hF7s3hnWhU97mvuPuTn0rZ9hiUvaRWr8IxDRqiH6gZvg/m5SGbnDenhYI+CMkD+u/Y yq9O1lGUiemYoEtnktBPDNPgN/0IE4Foxl0HiEZA2PZ+5C5MVVkRc5nI50cknA7qX7v3 Mco51vVXK4rs2Z9Z3CJbuAZCvG7bRXcfyYcC0rDDWyROJXTXYIiPYdwBC60DQ+h0WqCP FWlML0PGVFNR/hzehrJfjDroF5xAIJYSzFIvf8Aef7QYfub3ljo4gKGjy3i8X0MYnfX9 DGV4csHi82GDlrwahcXYZIlqwRK6kFZHgluLcBjC5UGtMdOA6/w23f4fudszSi5dsub4 VIVw==
X-Gm-Message-State: ALoCoQnjTboupQTlZGv5SHD83+ML1Vh6VzC6+NSv/g48o93qbPeTqdrdImB8abcLIHCIvUqRwEgeNycpF6KS1XxAtEvku2FAgmPsvmNzBgHQPUDzVsQ1AYDOaBsPPLdv3KEhIkcO1+K/ojn+uKeymFrmqiXDTk4bNiNa7/PvdT6qnyFXPQz6UyWBy2x+GLR6pGIHRZeQUYRl
MIME-Version: 1.0
X-Received: by 10.140.100.181 with SMTP id s50mr11078286qge.113.1391689958485;  Thu, 06 Feb 2014 04:32:38 -0800 (PST)
Sender: yhirano@google.com
Received: by 10.140.109.202 with HTTP; Thu, 6 Feb 2014 04:32:38 -0800 (PST)
In-Reply-To: <634914A010D0B943A035D226786325D4446C295D7E@EXVMBX020-12.exch020.serverdata.net>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <3274C7C2-41D4-49AD-971E-DA3D85898BC2@zaphoyd.com> <CABihn6GGezUU2aSboK0wsxYX3TBWTz5Av6-ZFJTPy6PQ50LrTQ@mail.gmail.com> <740FF040-F1E9-43E6-ACDE-0014FE4D012F@zaphoyd.com> <CABihn6HM_yXSB9RapNEQEG0_9OYfpHG3uH+=wnDwRJAFbh9EJA@mail.gmail.com> <634914A010D0B943A035D226786325D4446C295D7E@EXVMBX020-12.exch020.serverdata.net>
Date: Thu, 6 Feb 2014 04:32:38 -0800
X-Google-Sender-Auth: -uxRLeIGZVWLqMXImZx9YXvbSN0
Message-ID: <CABihn6HaRGNqUqug-FTdzY=ya1CfUrEixSPuYBAz4VZTEHcxfg@mail.gmail.com>
From: Yutaka Hirano <yhirano@chromium.org>
To: Tobias Oberstein <tobias.oberstein@tavendo.de>
Content-Type: multipart/alternative; boundary=001a1134f2b89d0eb904f1bc111e
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 06 Feb 2014 12:32:44 -0000

--001a1134f2b89d0eb904f1bc111e
Content-Type: text/plain; charset=UTF-8

Hi Tobias,

>As I said before, enabling intermediaries to transfer WS over HTTP/2.0
> *without an RFC6455 parser* is important, I think.
> Agree. And Plan D allows this.

I'm sorry, I don't understand how. Can you explain?

FYI, what I'm worrying about is the following.

> Some intermediaries can buffer DATA frames, but they should be flushed at
> each message end. Otherwise, deadlock can occur.


 MSG_DONE is a problem.
> Assume we would
> a) set MSG_DONE == FIN Bit of WS Frame and
> b) assign semantics to MSG_DONE: everything in the payload octet stream up
> to MSG_DONE is 1 message
> This leads to incompatiblity with RFC6455, which allows WS extensions to
> _interleave_ messages. Without knowledge of
> how the interleaving is done, MSG_DONE is void of meaning. (It also leads
> to problem with WS control frames.)

MSG_DONE is used just as a flush signal (at HTTP/2.0 frame level) in Plan A.
I don't think it breaks permessage-priority.



On Thu, Feb 6, 2014 at 4:09 AM, Tobias Oberstein <
tobias.oberstein@tavendo.de> wrote:

> >>I don't have a strong preference for either of these options. I do have
> a strong preference that we not do something in between. RFC6455
> compatibility is useful for me because it allows sharing
> code/infrastructure and backwards compatibility. Not because the specific
> features of RFC6455 are preserved.
>
> >>That is the reason I think adding MSG_DONE to WS-TUNNEL doesn't make
> sense. It puts us in a situation where *all* HTTP/2.0 implementations must
> implement a bunch of message related functionality that is already in
> RFC6455.
> >I still disagree.
> >As I said before, enabling intermediaries to transfer WS over HTTP/2.0
> *without an RFC6455 parser* is important, I think.
>
> Agree. And Plan D allows this.
>
> >No matter what protocol we choose, we must observe the HTTP/2.0 framing
> protocol, which means:
> >  - The size of a frame must not exceed the fixed limit.
> >  - We must set and respect the END_STREAM flag appropriately.
> >  - ...
> >Adding one rule (MSG_DONE) is harmless, and I think it has a merit.
>
> MSG_DONE is a problem.
>
> Assume we would
>
> a) set MSG_DONE == FIN Bit of WS Frame and
> b) assign semantics to MSG_DONE: everything in the payload octet stream up
> to MSG_DONE is 1 message
>
> This leads to incompatiblity with RFC6455, which allows WS extensions to
> _interleave_ messages. Without knowledge of
> how the interleaving is done, MSG_DONE is void of meaning. (It also leads
> to problem with WS control frames.)
>
> If we allow MSG_DONE, the result might be:
>
> SPDY/WS puts practically limits on the design space of WebSocket
> extensibility.
>
> And this exactly is my problem: why should SPDY, and it's wish to tunnel
> WS (practically) _limit_ the design space of an unrelated protocol (WS)?
>
> In fact, I find this idea quite disturbing:
>
> Protocol A that wants to tunnel protocol B puts up limits on protocol A to
> fulfill it's perceived requirements.
>
> No, Protocol A should just tunnel B "as is". That is tunneling.
>
> Note that Plan D does not have this problems. It also allows to
> mux/demux/load-balance tunneled WS on
> a per-stream basis.
>
> Plan D does not allow to mux/demux/load-balance on a _per-message_ basis.
>
> And as far as I understand Robert, this is his itch. And I can't follow
> that. Since:
>
> muxing/demuxing/load-balancing an _ordered_ message stream on a
> _per-message_ basis
> breaks the _ordering_ guarantee made in the first place.
>
> And on the other hand: if you want to mux/demux/lb on a _per-message_
> basis, great, but not
> on top of a strictly ordered, message-based protocol (WS).
>
> So currently I think what Robert wants is a _new_ message-based protocol
> over SPDY - not WS.
>
> Lets call that Plan E  ("new message-based proto on SPDY - unrelated to WS
> (other than probably W3C API)").
>
> FWIW, I would support Plan D or Plan E. I think Plan A, B and C are less
> desirable.
>
> Plan D is straight-forward - in particular taking into account your
> (Yutaka's) work already done on initial handshake and negotiation.
>
> Plan E: this might be interesting, since
>
> 1) HTTP: provides apps with independent request/response message exchange
> pattern
> 2) WS: provides apps with strictly ordered message stream
>
> obviously does not cover all message exchange patterns apps might want.
>
> I agree with Robert on that (if I understood his other mail correctly).
>
> But this is a bigger undertaking than "tunnel WS over SDPY".
>
> It involves not only protocol, but also API for app to express their
> desired message exchange pattern. etc etc ..
>
> Also, there is work being done on WebRTC Data Channels, which provide apps
> with all combinations of
>
> - ordered/unordered
> - reliable/unreliable
>
> message delivery.
>
> In fact, WebRTC has a complete, independent protocol stack on top of UDP.
>
> So as far as I can see, there are at least 3 efforts now on the Web data
> have business in "message-based" protocols:
>
> - WebSocket
> - WebRTC
> - SPDY
>
> /Tobias
>
>
>
>
> On Thu, Feb 6, 2014 at 12:46 AM, Peter Thorson <webmaster@zaphoyd.com>
> wrote:
>
> On Feb 5, 2014, at 4:56, Yutaka Hirano <yhirano@chromium.org> wrote:
>
>
> Let me talk about PlanA, WS-TUNNEL with MSG_DONE.
> Roberto says he is interested in loadbalancing ***without needing to
> understand application-layer semantics***. (1)
> I say we can't detect message boundary ***without understanding WebSocket
> level semantics***. (2)
> You say we can detect message boundary ***with the RFC6455 parser***. (3)
>
> Year, I think (2) and (3) are both true.
>
> I thought that Roberto didn't want to implement an RFC6455 paser for the
> loadbalancer.
> If it is wrong: Yay! PlanA (with MSG_DONE) seems good.
>
> If it is correct:
> I still think MSG_DONE is insufficient for the loadbalancing.
> Imagine a WebSocket connection consisting of three WebSocket frames, f1
> [without FIN], f2 [with FIN] and f3 [with FIN].
> Assume that there is no extensions.
> Assume that one HTTP/2.0 frame is mapped to one WebSocket frame for
> simplicity.
> Let the HTTP/2.0 frames be hf1 [without MSG_DONE], hf2 [with MSG_DONE] and
> hf3 [with MSG_DONE].
> If the loadbalancer does NOT understand RFC6455, there are two
> possibilities,
>  - A message consists of (hf1, hf2) and another message consists of (hf3).
>  - A data message consists of (hf1, hf3) and a control message consists of
> (hf2).
> Hence it fails to recognize messages.
>
> Note: For enabling intermediaries to transfer WS over HTTP/2.0 data
> without understanding RFC6455, MSG_DONE is sufficient.
>
> Let me define "Plan D" more precisely:
> HTTP/2.0 DATA frames are used as a dumb transport for an RFC6455 byte
> stream. There is no 1:1 frame mapping between WS frames and HTTP/2.0
> frames. A browser could send two WebSocket frames back to back in one
> HTTP/2.0 frame. An intermediary could split this frame evenly into four
> HTTP/2.0 frames each containing half of an RFC6455 frame. The remote
> endpoint would accept an HTTP/2.0 frame, feed its payload bytes straight to
> an RFC6455 parser which would handle the rest. A MSG_DONE flag is not
> required at all even for full ignorant intermediary compatibility.
>
> Benefits:
> - 100% RFC6455 compatibility present and future
> - Simple implementation:
> -- Nothing needs to be added to HTTP/2.0 or RFC6455.
> -- Browsers leverage existing RFC6455 parsers and just add logic to choose
> whether to use an HTTP/2.0 stream or a raw TCP stream. Data centers demux
> the streams into RFC6455 message processors.
> -- Everyone gets trivial backwards compatibility with existing RFC6455
> software and hardware, both on the client and server side.
> - Intermediaries that don't care can safely ignore or re-fragment however
> they like.
>
> Drawbacks
> - Requires intermediaries and endpoints that want to understand messages
> to read RFC6455. (I don't think this is an issue, but it seems important to
> Roberto)
> - Retains unnecessary/redundant features (masking, control messages, frame
> length fields)
>
> I think this is the best option if we value RFC6455 compatibility or
> desire ease of integration into existing software and environments.
> Otherwise, I would prefer that we go the "Plan B extreme" route that
> ditches RFC6455 entirely and builds a completely native HTTP/2.0 message
> stream. MSG_DONE/type bits, no masking, no opcode, no rsv, no control
> frames.
>
> Benefits:
> - HTTP/2.0 software doesn't need to know about RFC6455 unless it wants
> backwards compatibility with non-HTTP/2.0 software
> - Doesn't re-create lots of irrelevant/duplicated features from RFC6455
> - Optimal wire efficiency.
>
> Drawbacks:
> - Intermediaries need to implement "HTTP/2.0 messaging". (it is simple
> enough now though that that shouldn't be a great hardship.)
> - Limited RFC6455 extension compatibility.
>
> I don't have a strong preference for either of these options. I do have a
> strong preference that we not do something in between. RFC6455
> compatibility is useful for me because it allows sharing
> code/infrastructure and backwards compatibility. Not because the specific
> features of RFC6455 are preserved.
>
> That is the reason I think adding MSG_DONE to WS-TUNNEL doesn't make
> sense. It puts us in a situation where *all* HTTP/2.0 implementations must
> implement a bunch of message related functionality that is already in
> RFC6455. As Yukata points out MSG_DONE alone is not sufficient to detect
> messages in a tunneled RFC6455 stream. You'd need to throw in an opcode as
> well. At this point we end up re-implementing half of RFC6455 in HTTP/2.0
> in order to avoid using RFC6455 in a product that already has an RFC6455
> parser?? The only gain I see is certain types of limited and specialized
> intermediaries can drop RFC6455. That doesn't seem like a good tradeoff
> given that we could just choose Plan B and get rid of RFC6455 for everybody.
>
> Regarding message based load balancing for WebSocket:
> I agree with Tobias that this doesn't make sense. An application that
> wants to send independent messages can send HTTP requests. Such requests
> can be easily muxed, demuxed, or load balanced. Each message has headers to
> store application specific information necessary to correctly load balance.
> This can all be done efficiently with the functionality HTTP/2.0 provides
> already.
>
>
>

--001a1134f2b89d0eb904f1bc111e
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"im" style=3D"font-family:arial,sans-serif;fo=
nt-size:13px"><font color=3D"#000000">Hi Tobias,</font></div><div class=3D"=
im" style=3D"font-family:arial,sans-serif;font-size:13px"><br></div><blockq=
uote style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-co=
lor:rgb(204,204,204);border-left-style:solid;padding-left:1ex" class=3D"gma=
il_quote">
&gt;As I said before, enabling intermediaries to transfer WS over HTTP/2.0 =
*without an RFC6455 parser* is important, I think.<br><span style=3D"font-f=
amily:arial,sans-serif;font-size:13px">Agree. And Plan D allows this.</span=
></blockquote>
<div>I&#39;m sorry, I don&#39;t understand how. Can you explain?</div><div>=
<br></div><div>FYI, what I&#39;m worrying about is the following.</div><blo=
ckquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left=
-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;paddi=
ng-left:1ex">
Some intermediaries can buffer DATA frames, but they should be flushed at e=
ach message end. Otherwise, deadlock can occur.</blockquote><div><br></div>=
<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex">
=C2=A0<span style=3D"font-family:arial,sans-serif;font-size:13px">MSG_DONE =
is a problem.</span><br style=3D"font-family:arial,sans-serif;font-size:13p=
x"><span style=3D"font-family:arial,sans-serif;font-size:13px">Assume we wo=
uld</span><br style=3D"font-family:arial,sans-serif;font-size:13px">
<span style=3D"font-family:arial,sans-serif;font-size:13px">a) set MSG_DONE=
 =3D=3D FIN Bit of WS Frame and<br></span><span style=3D"font-family:arial,=
sans-serif;font-size:13px">b) assign semantics to MSG_DONE: everything in t=
he payload octet stream up to MSG_DONE is 1 message</span><br style=3D"font=
-family:arial,sans-serif;font-size:13px">
<span style=3D"font-family:arial,sans-serif;font-size:13px">This leads to i=
ncompatiblity with RFC6455, which allows WS extensions to _interleave_ mess=
ages. Without knowledge of<br></span><span style=3D"font-family:arial,sans-=
serif;font-size:13px">how the interleaving is done, MSG_DONE is void of mea=
ning. (It also leads to problem with WS control frames.)</span></blockquote=
>
<div>MSG_DONE is used just as a flush signal (at HTTP/2.0 frame level) in P=
lan A.</div><div>I don&#39;t think it breaks permessage-priority.</div><div=
><br></div></div><div class=3D"gmail_extra"><br><br><div class=3D"gmail_quo=
te">
On Thu, Feb 6, 2014 at 4:09 AM, Tobias Oberstein <span dir=3D"ltr">&lt;<a h=
ref=3D"mailto:tobias.oberstein@tavendo.de" target=3D"_blank">tobias.oberste=
in@tavendo.de</a>&gt;</span> wrote:<br><blockquote class=3D"gmail_quote" st=
yle=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div class=3D"im">&gt;&gt;I don&#39;t have a strong preference for either o=
f these options. I do have a strong preference that we not do something in =
between. RFC6455 compatibility is useful for me because it allows sharing c=
ode/infrastructure and backwards compatibility. Not because the specific fe=
atures of RFC6455 are preserved.<br>

=C2=A0<br>
&gt;&gt;That is the reason I think adding MSG_DONE to WS-TUNNEL doesn&#39;t=
 make sense. It puts us in a situation where *all* HTTP/2.0 implementations=
 must implement a bunch of message related functionality that is already in=
 RFC6455.<br>

&gt;I still disagree.<br>
&gt;As I said before, enabling intermediaries to transfer WS over HTTP/2.0 =
*without an RFC6455 parser* is important, I think.<br>
<br>
</div>Agree. And Plan D allows this.<br>
<div class=3D"im"><br>
&gt;No matter what protocol we choose, we must observe the HTTP/2.0 framing=
 protocol, which means:<br>
&gt;=C2=A0 - The size of a frame must not exceed the fixed limit.<br>
&gt;=C2=A0 - We must set and respect the END_STREAM flag appropriately.<br>
&gt;=C2=A0 - ...<br>
&gt;Adding one rule (MSG_DONE) is harmless, and I think it has a merit.<br>
<br>
</div>MSG_DONE is a problem.<br>
<br>
Assume we would<br>
<br>
a) set MSG_DONE =3D=3D FIN Bit of WS Frame and<br>
b) assign semantics to MSG_DONE: everything in the payload octet stream up =
to MSG_DONE is 1 message<br>
<br>
This leads to incompatiblity with RFC6455, which allows WS extensions to _i=
nterleave_ messages. Without knowledge of<br>
how the interleaving is done, MSG_DONE is void of meaning. (It also leads t=
o problem with WS control frames.)<br>
<br>
If we allow MSG_DONE, the result might be:<br>
<br>
SPDY/WS puts practically limits on the design space of WebSocket extensibil=
ity.<br>
<br>
And this exactly is my problem: why should SPDY, and it&#39;s wish to tunne=
l WS (practically) _limit_ the design space of an unrelated protocol (WS)?<=
br>
<br>
In fact, I find this idea quite disturbing:<br>
<br>
Protocol A that wants to tunnel protocol B puts up limits on protocol A to =
fulfill it&#39;s perceived requirements.<br>
<br>
No, Protocol A should just tunnel B &quot;as is&quot;. That is tunneling.<b=
r>
<br>
Note that Plan D does not have this problems. It also allows to mux/demux/l=
oad-balance tunneled WS on<br>
a per-stream basis.<br>
<br>
Plan D does not allow to mux/demux/load-balance on a _per-message_ basis.<b=
r>
<br>
And as far as I understand Robert, this is his itch. And I can&#39;t follow=
 that. Since:<br>
<br>
muxing/demuxing/load-balancing an _ordered_ message stream on a _per-messag=
e_ basis<br>
breaks the _ordering_ guarantee made in the first place.<br>
<br>
And on the other hand: if you want to mux/demux/lb on a _per-message_ basis=
, great, but not<br>
on top of a strictly ordered, message-based protocol (WS).<br>
<br>
So currently I think what Robert wants is a _new_ message-based protocol ov=
er SPDY - not WS.<br>
<br>
Lets call that Plan E =C2=A0(&quot;new message-based proto on SPDY - unrela=
ted to WS (other than probably W3C API)&quot;).<br>
<br>
FWIW, I would support Plan D or Plan E. I think Plan A, B and C are less de=
sirable.<br>
<br>
Plan D is straight-forward - in particular taking into account your (Yutaka=
&#39;s) work already done on initial handshake and negotiation.<br>
<br>
Plan E: this might be interesting, since<br>
<br>
1) HTTP: provides apps with independent request/response message exchange p=
attern<br>
2) WS: provides apps with strictly ordered message stream<br>
<br>
obviously does not cover all message exchange patterns apps might want.<br>
<br>
I agree with Robert on that (if I understood his other mail correctly).<br>
<br>
But this is a bigger undertaking than &quot;tunnel WS over SDPY&quot;.<br>
<br>
It involves not only protocol, but also API for app to express their desire=
d message exchange pattern. etc etc ..<br>
<br>
Also, there is work being done on WebRTC Data Channels, which provide apps =
with all combinations of<br>
<br>
- ordered/unordered<br>
- reliable/unreliable<br>
<br>
message delivery.<br>
<br>
In fact, WebRTC has a complete, independent protocol stack on top of UDP.<b=
r>
<br>
So as far as I can see, there are at least 3 efforts now on the Web data ha=
ve business in &quot;message-based&quot; protocols:<br>
<br>
- WebSocket<br>
- WebRTC<br>
- SPDY<br>
<span class=3D"HOEnZb"><font color=3D"#888888"><br>
/Tobias<br>
</font></span><div class=3D"HOEnZb"><div class=3D"h5"><br>
<br>
<br>
<br>
On Thu, Feb 6, 2014 at 12:46 AM, Peter Thorson &lt;<a href=3D"mailto:webmas=
ter@zaphoyd.com">webmaster@zaphoyd.com</a>&gt; wrote:<br>
<br>
On Feb 5, 2014, at 4:56, Yutaka Hirano &lt;<a href=3D"mailto:yhirano@chromi=
um.org">yhirano@chromium.org</a>&gt; wrote:<br>
<br>
<br>
Let me talk about PlanA, WS-TUNNEL with MSG_DONE.<br>
Roberto says he is interested in loadbalancing ***without needing to unders=
tand application-layer semantics***. (1)<br>
I say we can&#39;t detect message boundary ***without understanding WebSock=
et level semantics***. (2)<br>
You say we can detect message boundary ***with the RFC6455 parser***. (3)<b=
r>
<br>
Year, I think (2) and (3) are both true.<br>
<br>
I thought that Roberto didn&#39;t want to implement an RFC6455 paser for th=
e loadbalancer.<br>
If it is wrong: Yay! PlanA (with MSG_DONE) seems good.<br>
<br>
If it is correct:<br>
I still think MSG_DONE is insufficient for the loadbalancing.<br>
Imagine a WebSocket connection consisting of three WebSocket frames, f1 [wi=
thout FIN], f2 [with FIN] and f3 [with FIN].<br>
Assume that there is no extensions.<br>
Assume that one HTTP/2.0 frame is mapped to one WebSocket frame for simplic=
ity.<br>
Let the HTTP/2.0 frames be hf1 [without MSG_DONE], hf2 [with MSG_DONE] and =
hf3 [with MSG_DONE].<br>
If the loadbalancer does NOT understand RFC6455, there are two possibilitie=
s,<br>
=C2=A0- A message consists of (hf1, hf2) and another message consists of (h=
f3).<br>
=C2=A0- A data message consists of (hf1, hf3) and a control message consist=
s of (hf2).<br>
Hence it fails to recognize messages.<br>
<br>
Note: For enabling intermediaries to transfer WS over HTTP/2.0 data without=
 understanding RFC6455, MSG_DONE is sufficient.=C2=A0<br>
<br>
Let me define &quot;Plan D&quot; more precisely:<br>
HTTP/2.0 DATA frames are used as a dumb transport for an RFC6455 byte strea=
m. There is no 1:1 frame mapping between WS frames and HTTP/2.0 frames. A b=
rowser could send two WebSocket frames back to back in one HTTP/2.0 frame. =
An intermediary could split this frame evenly into four HTTP/2.0 frames eac=
h containing half of an RFC6455 frame. The remote endpoint would accept an =
HTTP/2.0 frame, feed its payload bytes straight to an RFC6455 parser which =
would handle the rest. A MSG_DONE flag is not required at all even for full=
 ignorant intermediary compatibility.<br>

<br>
Benefits:<br>
- 100% RFC6455 compatibility present and future<br>
- Simple implementation:<br>
-- Nothing needs to be added to HTTP/2.0 or RFC6455.=C2=A0<br>
-- Browsers leverage existing RFC6455 parsers and just add logic to choose =
whether to use an HTTP/2.0 stream or a raw TCP stream. Data centers demux t=
he streams into RFC6455 message processors.<br>
-- Everyone gets trivial backwards compatibility with existing RFC6455 soft=
ware and hardware, both on the client and server side.<br>
- Intermediaries that don&#39;t care can safely ignore or re-fragment howev=
er they like.<br>
<br>
Drawbacks<br>
- Requires intermediaries and endpoints that want to understand messages to=
 read RFC6455. (I don&#39;t think this is an issue, but it seems important =
to Roberto)<br>
- Retains unnecessary/redundant features (masking, control messages, frame =
length fields)<br>
<br>
I think this is the best option if we value RFC6455 compatibility or desire=
 ease of integration into existing software and environments. Otherwise, I =
would prefer that we go the &quot;Plan B extreme&quot; route that ditches R=
FC6455 entirely and builds a completely native HTTP/2.0 message stream. MSG=
_DONE/type bits, no masking, no opcode, no rsv, no control frames.<br>

<br>
Benefits:<br>
- HTTP/2.0 software doesn&#39;t need to know about RFC6455 unless it wants =
backwards compatibility with non-HTTP/2.0 software<br>
- Doesn&#39;t re-create lots of irrelevant/duplicated features from RFC6455=
<br>
- Optimal wire efficiency.<br>
<br>
Drawbacks:<br>
- Intermediaries need to implement &quot;HTTP/2.0 messaging&quot;. (it is s=
imple enough now though that that shouldn&#39;t be a great hardship.)<br>
- Limited RFC6455 extension compatibility.<br>
<br>
I don&#39;t have a strong preference for either of these options. I do have=
 a strong preference that we not do something in between. RFC6455 compatibi=
lity is useful for me because it allows sharing code/infrastructure and bac=
kwards compatibility. Not because the specific features of RFC6455 are pres=
erved.<br>

<br>
That is the reason I think adding MSG_DONE to WS-TUNNEL doesn&#39;t make se=
nse. It puts us in a situation where *all* HTTP/2.0 implementations must im=
plement a bunch of message related functionality that is already in RFC6455=
. As Yukata points out MSG_DONE alone is not sufficient to detect messages =
in a tunneled RFC6455 stream. You&#39;d need to throw in an opcode as well.=
 At this point we end up re-implementing half of RFC6455 in HTTP/2.0 in ord=
er to avoid using RFC6455 in a product that already has an RFC6455 parser??=
 The only gain I see is certain types of limited and specialized intermedia=
ries can drop RFC6455. That doesn&#39;t seem like a good tradeoff given tha=
t we could just choose Plan B and get rid of RFC6455 for everybody.<br>

<br>
Regarding message based load balancing for WebSocket:<br>
I agree with Tobias that this doesn&#39;t make sense. An application that w=
ants to send independent messages can send HTTP requests. Such requests can=
 be easily muxed, demuxed, or load balanced. Each message has headers to st=
ore application specific information necessary to correctly load balance. T=
his can all be done efficiently with the functionality HTTP/2.0 provides al=
ready.<br>

<br>
<br>
</div></div></blockquote></div><br></div>

--001a1134f2b89d0eb904f1bc111e--

From tobias.oberstein@tavendo.de  Thu Feb  6 04:44:08 2014
Return-Path: <tobias.oberstein@tavendo.de>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 540151A03AF for <hybi@ietfa.amsl.com>; Thu,  6 Feb 2014 04:44:08 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.9
X-Spam-Level: 
X-Spam-Status: No, score=-1.9 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, RCVD_IN_DNSWL_NONE=-0.0001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id IAk8l5lefG5A for <hybi@ietfa.amsl.com>; Thu,  6 Feb 2014 04:44:06 -0800 (PST)
Received: from EXHUB020-5.exch020.serverdata.net (exhub020-5.exch020.serverdata.net [206.225.164.32]) by ietfa.amsl.com (Postfix) with ESMTP id 2AEC01A00F5 for <hybi@ietf.org>; Thu,  6 Feb 2014 04:44:06 -0800 (PST)
Received: from EXVMBX020-12.exch020.serverdata.net ([169.254.3.11]) by EXHUB020-5.exch020.serverdata.net ([206.225.164.32]) with mapi; Thu, 6 Feb 2014 04:44:04 -0800
From: Tobias Oberstein <tobias.oberstein@tavendo.de>
To: Roberto Peon <fenix@google.com>
Date: Thu, 6 Feb 2014 04:44:03 -0800
Thread-Topic: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
Thread-Index: Ac8inrz1my6DtuK4SjisjvIaKuG+twAln9aQ
Message-ID: <634914A010D0B943A035D226786325D4446C295D80@EXVMBX020-12.exch020.serverdata.net>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <CAGzyod6H-To9EhaPqW5C0c9xyy5XnRg_851GuT3q=X+KoXe7Wg@mail.gmail.com> <0533B942-C1DA-4BFE-BE80-4722C0BD740D@zaphoyd.com> <CAGzyod7xtCZqUvcb0gX0PYKyikyd4kKc=kOzwzgfAmUFExCq3g@mail.gmail.com> <634914A010D0B943A035D226786325D4446C29584C@EXVMBX020-12.exch020.serverdata.net> <CAGzyod6LZ=wXj3jXYQ4iCo+jtFEQHfm+NEnei9N6AoaLrYyH1g@mail.gmail.com>
In-Reply-To: <CAGzyod6LZ=wXj3jXYQ4iCo+jtFEQHfm+NEnei9N6AoaLrYyH1g@mail.gmail.com>
Accept-Language: de-DE, en-US
Content-Language: de-DE
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
acceptlanguage: de-DE, en-US
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: base64
MIME-Version: 1.0
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 06 Feb 2014 12:44:08 -0000
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From tobias.oberstein@tavendo.de  Thu Feb  6 05:03:55 2014
Return-Path: <tobias.oberstein@tavendo.de>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 767C91A00F6 for <hybi@ietfa.amsl.com>; Thu,  6 Feb 2014 05:03:55 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.9
X-Spam-Level: 
X-Spam-Status: No, score=-1.9 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, RCVD_IN_DNSWL_NONE=-0.0001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id M_S54lAwBOwC for <hybi@ietfa.amsl.com>; Thu,  6 Feb 2014 05:03:54 -0800 (PST)
Received: from EXHUB020-3.exch020.serverdata.net (exhub020-3.exch020.serverdata.net [206.225.164.30]) by ietfa.amsl.com (Postfix) with ESMTP id C80E61A00F5 for <hybi@ietf.org>; Thu,  6 Feb 2014 05:03:53 -0800 (PST)
Received: from EXVMBX020-12.exch020.serverdata.net ([169.254.3.11]) by EXHUB020-3.exch020.serverdata.net ([206.225.164.30]) with mapi; Thu, 6 Feb 2014 05:03:52 -0800
From: Tobias Oberstein <tobias.oberstein@tavendo.de>
To: Yutaka Hirano <yhirano@chromium.org>
Date: Thu, 6 Feb 2014 05:03:49 -0800
Thread-Topic: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
Thread-Index: Ac8jN4aS62vC15zoSTCa+AEhp6713AAAltqA
Message-ID: <634914A010D0B943A035D226786325D4446C295D86@EXVMBX020-12.exch020.serverdata.net>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <3274C7C2-41D4-49AD-971E-DA3D85898BC2@zaphoyd.com> <CABihn6GGezUU2aSboK0wsxYX3TBWTz5Av6-ZFJTPy6PQ50LrTQ@mail.gmail.com> <740FF040-F1E9-43E6-ACDE-0014FE4D012F@zaphoyd.com> <CABihn6HM_yXSB9RapNEQEG0_9OYfpHG3uH+=wnDwRJAFbh9EJA@mail.gmail.com> <634914A010D0B943A035D226786325D4446C295D7E@EXVMBX020-12.exch020.serverdata.net> <CABihn6HaRGNqUqug-FTdzY=ya1CfUrEixSPuYBAz4VZTEHcxfg@mail.gmail.com>
In-Reply-To: <CABihn6HaRGNqUqug-FTdzY=ya1CfUrEixSPuYBAz4VZTEHcxfg@mail.gmail.com>
Accept-Language: de-DE, en-US
Content-Language: de-DE
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
acceptlanguage: de-DE, en-US
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: base64
MIME-Version: 1.0
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 06 Feb 2014 13:03:55 -0000
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From webmaster@zaphoyd.com  Thu Feb  6 07:21:38 2014
Return-Path: <webmaster@zaphoyd.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 211A01A03DF for <hybi@ietfa.amsl.com>; Thu,  6 Feb 2014 07:21:38 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.899
X-Spam-Level: 
X-Spam-Status: No, score=-1.899 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HTML_MESSAGE=0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id PnFJW9Xr_IVF for <hybi@ietfa.amsl.com>; Thu,  6 Feb 2014 07:21:35 -0800 (PST)
Received: from authsmtp00.uchicago.edu (authsmtp00.uchicago.edu [128.135.12.120]) by ietfa.amsl.com (Postfix) with ESMTP id F39041A03D5 for <hybi@ietf.org>; Thu,  6 Feb 2014 07:21:34 -0800 (PST)
Received: from [10.0.1.89] (c-68-51-76-178.hsd1.il.comcast.net [68.51.76.178]) (Authenticated sender: zaphoyd) by authsmtp00.uchicago.edu (Postfix) with ESMTP id 69D8F4980A0; Thu,  6 Feb 2014 09:21:33 -0600 (CST)
Content-Type: multipart/alternative; boundary="Apple-Mail=_0A751847-539B-4CB8-B943-71A4D065CCCF"
Mime-Version: 1.0 (Mac OS X Mail 7.0 \(1822\))
From: Peter Thorson <webmaster@zaphoyd.com>
In-Reply-To: <CABihn6HM_yXSB9RapNEQEG0_9OYfpHG3uH+=wnDwRJAFbh9EJA@mail.gmail.com>
Date: Thu, 6 Feb 2014 09:26:19 -0600
Message-Id: <0F978119-88F2-49CE-98C6-2A640527AA64@zaphoyd.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <3274C 7C2-41D4-49AD-971E-DA3D85898BC2@zaphoyd.com> <CABihn6GGezUU2aSboK0wsxYX3TBWTz5Av6-ZFJTPy6PQ50LrTQ@mail.gmail.com> <740FF040-F1E9-43E6-ACDE-0014FE4D012F@zaphoyd.com> <CABihn6HM_yXSB9RapNEQEG0_9OYfpHG3uH+=wnDwRJAFbh9EJA@mail.gmail.com>
To: Yutaka Hirano <yhirano@chromium.org>
X-Mailer: Apple Mail (2.1822)
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 06 Feb 2014 15:21:38 -0000

--Apple-Mail=_0A751847-539B-4CB8-B943-71A4D065CCCF
Content-Transfer-Encoding: quoted-printable
Content-Type: text/plain;
	charset=windows-1252


On Feb 6, 2014, at 0:04, Yutaka Hirano <yhirano@chromium.org> wrote:

> As I said before, enabling intermediaries to transfer WS over HTTP/2.0 =
*without an RFC6455 parser* is important, I think.
>> Agree. And Plan D allows this.
> I'm sorry, I don't understand how. Can you explain?

Specifically for Plan D..
An RFC6455 parser is required to:
- Add or remove messages from the stream
- Determine WS frame and message boundaries
- Read message contents
An RFC6455 parser is NOT required to:
- Proxy the stream
- Load balance the stream
- Mux a WS stream into an HTTP/2.0 stream
- Demux a WS stream out of an HTTP/2.0 stream
- Flush a stream using an HTTP/2.0 level FLUSH_HINT

If it is important that intermediaries or endpoints be able to read and =
manipulate these streams on a per-message or per-frame level without an =
RFC6455 parser then a tunneled solution is a bad choice. If all the =
intermediaries need to do is move the bytes around then it is fine.

> No matter what protocol we choose, we must observe the HTTP/2.0 =
framing protocol, which means:
>   - The size of a frame must not exceed the fixed limit.
>   - We must set and respect the END_STREAM flag appropriately.
>   - ...
> Adding one rule (MSG_DONE) is harmless, and I think it has a merit.

I agree that the MSG_DONE concept in general is a useful general purpose =
flag to have in HTTP/2.0. I disagree that it is harmless to use it as a =
1:1 message delimiter in a tunneled RFC6455 stream.

>> MSG_DONE is used just as a flush signal (at HTTP/2.0 frame level) in =
Plan A.
>> I don't think it breaks permessage-priority.
>=20
> If it's intended use is a Flush signal to intermedaries, then we =
should probably
>=20
> a) call it FLUSH_HINT, and
> b) allow multiple WS frames with FIN bit inside a single SPDY data =
frame.
>=20
> FLUSH_HINT would be set if _any_ WS frame inside the SPDY data frame =
would have it's FIN bit set.
>=20
> This would do away with one additional limitation of Plan A over Plan =
D:
>=20
> As it stands now, Plan A does not allow batching of multiple WS =
messages inside a single SPDY data frame.

I am fine with a =93flush hint=94 flag where the sending endpoint =
*may/should* add it to HTTP/2.0 frames that the application prefers not =
to have buffered/delayed by intermediaries. I am fine with a separate =
bit named FLUSH_HINT (useful for either Plan D or native messaging) or =
re-using MSG_DONE to imply =93flush hint=94 for Plan D streams if =
MSG_DONE is used elsewhere by HTTP/2.0. I would not want to define =
specific scenarios where a flush must occur.

Re: Load balancing

Robertos point that HTTP messages traditionally must be answered in =
order certainly makes HTTP less appealing for the unordered, unrelated =
RPC style message case. I can see how an application might use WebSocket =
today in order to get out of order message return delivery.

I am curious if there has there been any discussion to relaxing that =
requirement in HTTP/2.0?

Nothing in Plan D or Native Messaging would prevent the creation of a =
per-message WebSocket load balancer. In fact you could also do one today =
using RFC6455 alone. Under Plan D, said balancer would require an =
RFC6455 parser as it would need to add/remove messages from a stream. =
Under native messaging obviously not. The load balancer would clearly =
need to understand the application semantics of that specific stream =
(i.e. this is a loosely coupled message stream that can be balanced). =
Information in the URL, subprotocol, headers, etc could be used to =
communicate this application specific information.=

--Apple-Mail=_0A751847-539B-4CB8-B943-71A4D065CCCF
Content-Transfer-Encoding: quoted-printable
Content-Type: text/html;
	charset=windows-1252

<html><head><meta http-equiv=3D"Content-Type" content=3D"text/html =
charset=3Dwindows-1252"></head><body style=3D"word-wrap: break-word; =
-webkit-nbsp-mode: space; -webkit-line-break: =
after-white-space;"><br><div><div>On Feb 6, 2014, at 0:04, Yutaka Hirano =
&lt;<a href=3D"mailto:yhirano@chromium.org">yhirano@chromium.org</a>&gt; =
wrote:</div><br><blockquote type=3D"cite"><div =
dir=3D"ltr"><div><blockquote class=3D"gmail_quote" style=3D"margin:0px =
0px 0px =
0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left=
-style:solid;padding-left:1ex">As I said before, enabling intermediaries =
to transfer WS over HTTP/2.0 *without an RFC6455 parser* is important, I =
think.</blockquote></div></div></blockquote><blockquote =
type=3D"cite"><blockquote type=3D"cite">Agree. And Plan D allows =
this.<br></blockquote>I'm sorry, I don't understand how. Can you =
explain?</blockquote><div><br></div><div>Specifically for Plan =
D..</div><div>An RFC6455 parser is required to:</div><div>- Add or =
remove messages from the stream</div><div>- Determine WS frame and =
message boundaries</div><div>- Read message contents</div><div>An =
RFC6455 parser is NOT required to:</div><div>- Proxy the =
stream</div><div>- Load balance the stream</div><div>- Mux a WS stream =
into an HTTP/2.0 stream</div><div>- Demux a WS stream out of an HTTP/2.0 =
stream</div><div>- Flush a stream using an HTTP/2.0 level =
FLUSH_HINT</div><div><br></div><div>If it is important that =
intermediaries or endpoints be able to read and manipulate these streams =
on a per-message or per-frame level without an RFC6455 parser then a =
tunneled solution is a bad choice. If all the intermediaries need to do =
is move the bytes around then it is fine.</div><br><blockquote =
type=3D"cite"><div dir=3D"ltr"><div>No matter what protocol we choose, =
we must observe the HTTP/2.0 framing protocol, which means:</div>
<div>&nbsp; - The size of a frame must not exceed the fixed =
limit.</div><div>&nbsp; - We must set and respect the END_STREAM flag =
appropriately.</div><div>&nbsp; - ...</div><div>Adding one rule =
(MSG_DONE) is harmless, and I think it has a =
merit.</div></div></blockquote><div><br></div><div>I agree that the =
MSG_DONE concept in general is a useful general purpose flag to have in =
HTTP/2.0. I disagree that it is harmless to use it as a 1:1 message =
delimiter in a tunneled RFC6455 =
stream.</div><div><br></div><div><blockquote =
type=3D"cite"></blockquote><blockquote type=3D"cite"><blockquote =
type=3D"cite">MSG_DONE is used just as a flush signal (at HTTP/2.0 frame =
level) in Plan A.<br>I don't think it breaks =
permessage-priority.<br></blockquote><br>If it's intended use is a Flush =
signal to intermedaries, then we should probably<br><br>a) call it =
FLUSH_HINT, and<br>b) allow multiple WS frames with FIN bit inside a =
single SPDY data frame.<br><br>FLUSH_HINT would be set if _any_ WS frame =
inside the SPDY data frame would have it's FIN bit set.<br><br>This =
would do away with one additional limitation of Plan A over Plan =
D:<br><br>As it stands now, Plan A does not allow batching of multiple =
WS messages inside a single SPDY data =
frame.</blockquote><br></div><div>I am fine with a =93flush hint=94 flag =
where the sending endpoint *may/should* add it to HTTP/2.0 frames that =
the application prefers not to have buffered/delayed by intermediaries. =
I am fine with a separate bit named FLUSH_HINT (useful for either Plan D =
or native messaging) or re-using MSG_DONE to imply =93flush hint=94 for =
Plan D streams if MSG_DONE is used elsewhere by HTTP/2.0. I would not =
want to define specific scenarios where a flush must =
occur.</div><div><br></div><div>Re: Load =
balancing</div><div><br></div><div>Robertos point that HTTP messages =
traditionally must be answered in order certainly makes HTTP less =
appealing for the unordered, unrelated RPC style message case. I can see =
how an application might use WebSocket today in order to get out of =
order message return delivery.</div><div><br></div><div>I am curious if =
there has there been any discussion to relaxing that requirement in =
HTTP/2.0?</div><div><br></div><div>Nothing in Plan D or Native Messaging =
would prevent the creation of a per-message WebSocket load balancer. In =
fact you could also do one today using RFC6455 alone. Under Plan D, said =
balancer would require an RFC6455 parser as it would need to add/remove =
messages from a stream. Under native messaging obviously not. The load =
balancer would clearly need to understand the application semantics of =
that specific stream (i.e. this is a loosely coupled message stream that =
can be balanced). Information in the URL, subprotocol, headers, etc =
could be used to communicate this application specific =
information.</div></div></body></html>=

--Apple-Mail=_0A751847-539B-4CB8-B943-71A4D065CCCF--

From fenix@google.com  Thu Feb  6 11:23:50 2014
Return-Path: <fenix@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id EBB331A041C for <hybi@ietfa.amsl.com>; Thu,  6 Feb 2014 11:23:49 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.913
X-Spam-Level: 
X-Spam-Status: No, score=-1.913 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.535, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id i0MFyWdkNogt for <hybi@ietfa.amsl.com>; Thu,  6 Feb 2014 11:23:46 -0800 (PST)
Received: from mail-oa0-x236.google.com (mail-oa0-x236.google.com [IPv6:2607:f8b0:4003:c02::236]) by ietfa.amsl.com (Postfix) with ESMTP id 7A1B41A03EE for <hybi@ietf.org>; Thu,  6 Feb 2014 11:23:46 -0800 (PST)
Received: by mail-oa0-f54.google.com with SMTP id i4so2935732oah.27 for <hybi@ietf.org>; Thu, 06 Feb 2014 11:23:45 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=MAO6TaWgAeQMM3X4a5oWHr047laAoht+/RsGByXIHnk=; b=dAUI+dhMsSBqW1kP1PxQ093VZ/oXkA7LfXhTPwpvA5VsXa9o1Fd/oISs/2Q86Y4GRJ ImOsNEwlaqlSdLEhohzhOljQ0LHoQvlaGtSTrN6VaFzNPiVvgbOFTEL8YhyWlo6doK/o SrQpBh/eLY7HFyCku5SsErrMCbTEC+b9Dn/HneldLmFFSNVYV6ngp15eW+NZ8PlqCOml RSJwbBpueDcFR22sz/QNSiMiOrGxG1GmEQoIL6zchPGnQgM7F/l/jeSmaiUf/MGqY2z7 8QcZPtWo5Ue/ZWU9NSzoLX/4m7xCTfmRIO1r24FCR1KDtkILwlyLh31A5gAjKnAd1kdc aFsA==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=MAO6TaWgAeQMM3X4a5oWHr047laAoht+/RsGByXIHnk=; b=S10fj77G6RbxYI8eF0enHTxCIqYuqEAvod+79Auqblt040NOCEyZ+xVFJtgnegdxft KHv5EWVV9vaw3p6zGWsSsUye8ZxWP/HUv8WEsihjVzKbU9+E+D9rAJfrgQR70pGd1E8T QTSkICphHxNV5UuJ8nBUEgCiE454OX1N9gq3t6rj9SuOiJQ3G1wMAZghkKzqzYYPKShg u66l9T8cd2W2ubdpPiepbyM3GXHSAGYf+zP/XPBkHkVpx+hkLzQ0/3oKwSa+mLEYDBL2 CqtQK9Q0JvnswJBlhiTkzldtrMfK+JU/p+zEpcxptU9RsSM9bBqaxSyHekldRZda+vVE Mzrw==
X-Gm-Message-State: ALoCoQnDUMzB2D6xvky2DFVzQ8LJTQaMVl5K+kCyTH3H+sZyJOfa3yHex1Ix8y7gn7ckcBhUbXzd8D5qpQWd++LEPNwnqcPz7/EYYdhK4HrXBvCMHDOS2hdJaifyU7AZ6tHhndCt/PCootZI8iClrujOHzb8ZWoQ8vJbt166eINbJAkWC/3LLbomu9ZA9o0RMdwPB8XBdCvJ
MIME-Version: 1.0
X-Received: by 10.182.128.138 with SMTP id no10mr8829291obb.32.1391714625035;  Thu, 06 Feb 2014 11:23:45 -0800 (PST)
Received: by 10.182.221.231 with HTTP; Thu, 6 Feb 2014 11:23:44 -0800 (PST)
In-Reply-To: <0F978119-88F2-49CE-98C6-2A640527AA64@zaphoyd.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <CABihn6GGezUU2aSboK0wsxYX3TBWTz5Av6-ZFJTPy6PQ50LrTQ@mail.gmail.com> <740FF040-F1E9-43E6-ACDE-0014FE4D012F@zaphoyd.com> <CABihn6HM_yXSB9RapNEQEG0_9OYfpHG3uH+=wnDwRJAFbh9EJA@mail.gmail.com> <0F978119-88F2-49CE-98C6-2A640527AA64@zaphoyd.com>
Date: Thu, 6 Feb 2014 11:23:44 -0800
Message-ID: <CAGzyod7BA_UXjN4r+Qr+P33QsPpeWWotxiScTpzqyhvbtPgNYQ@mail.gmail.com>
From: Roberto Peon <fenix@google.com>
To: Peter Thorson <webmaster@zaphoyd.com>
Content-Type: multipart/alternative; boundary=089e0153852edab64304f1c1cfa7
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 06 Feb 2014 19:23:50 -0000

--089e0153852edab64304f1c1cfa7
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

On Thu, Feb 6, 2014 at 7:26 AM, Peter Thorson <webmaster@zaphoyd.com> wrote=
:

>
> On Feb 6, 2014, at 0:04, Yutaka Hirano <yhirano@chromium.org> wrote:
>
> As I said before, enabling intermediaries to transfer WS over HTTP/2.0
>> *without an RFC6455 parser* is important, I think.
>
> Agree. And Plan D allows this.
>
> I'm sorry, I don't understand how. Can you explain?
>
>
> Specifically for Plan D..
> An RFC6455 parser is required to:
> - Add or remove messages from the stream
> - Determine WS frame and message boundaries
> - Read message contents
> An RFC6455 parser is NOT required to:
> - Proxy the stream
> - Load balance the stream
> - Mux a WS stream into an HTTP/2.0 stream
> - Demux a WS stream out of an HTTP/2.0 stream
> - Flush a stream using an HTTP/2.0 level FLUSH_HINT
>
> If it is important that intermediaries or endpoints be able to read and
> manipulate these streams on a per-message or per-frame level without an
> RFC6455 parser then a tunneled solution is a bad choice. If all the
> intermediaries need to do is move the bytes around then it is fine.
>

Exactly.


>
> No matter what protocol we choose, we must observe the HTTP/2.0 framing
> protocol, which means:
>   - The size of a frame must not exceed the fixed limit.
>   - We must set and respect the END_STREAM flag appropriately.
>   - ...
> Adding one rule (MSG_DONE) is harmless, and I think it has a merit.
>
>
> I agree that the MSG_DONE concept in general is a useful general purpose
> flag to have in HTTP/2.0. I disagree that it is harmless to use it as a 1=
:1
> message delimiter in a tunneled RFC6455 stream.
>

> MSG_DONE is used just as a flush signal (at HTTP/2.0 frame level) in Plan
> A.
> I don't think it breaks permessage-priority.
>
>
> If it's intended use is a Flush signal to intermedaries, then we should
> probably
>
> a) call it FLUSH_HINT, and
> b) allow multiple WS frames with FIN bit inside a single SPDY data frame.
>
> FLUSH_HINT would be set if _any_ WS frame inside the SPDY data frame woul=
d
> have it's FIN bit set.
>
> This would do away with one additional limitation of Plan A over Plan D:
>
> As it stands now, Plan A does not allow batching of multiple WS messages
> inside a single SPDY data frame.
>
> I wouldn't really find FLUSH_HINT useful for anything unless the
application-logic in the webpage could affect it.
But.. there is no such capability in the API right now.



>
> I am fine with a =E2=80=9Cflush hint=E2=80=9D flag where the sending endp=
oint *may/should*
> add it to HTTP/2.0 frames that the application prefers not to have
> buffered/delayed by intermediaries. I am fine with a separate bit named
> FLUSH_HINT (useful for either Plan D or native messaging) or re-using
> MSG_DONE to imply =E2=80=9Cflush hint=E2=80=9D for Plan D streams if MSG_=
DONE is used
> elsewhere by HTTP/2.0. I would not want to define specific scenarios wher=
e
> a flush must occur.
>


HTTP2's MSG_DONE (under a different name) is in the next implementation
draft, coming out shortly.
For HTTP on HTTP/2 it is use for delimiting chunk boundaries on an HTTP
stream, since the lack of this has proven painful for some. Intermediaries
are disallowed from coalescing data across frame where this is set.


>
> Re: Load balancing
>
> Robertos point that HTTP messages traditionally must be answered in order
> certainly makes HTTP less appealing for the unordered, unrelated RPC styl=
e
> message case. I can see how an application might use WebSocket today in
> order to get out of order message return delivery.
>
> I am curious if there has there been any discussion to relaxing that
> requirement in HTTP/2.0?
>

Unfortunately, one can't relax that requirement and still have the
backwards-compatible API that exists today: while HTTP/2 can deliver on
this, HTTP/1.X cannot.
That one need not change their content to deploy HTTP/2 (etc.) is a major
driver of deployment, so it isn't something that could be lightly changed.
IMHO, we'd have to invent a new API (which, honestly I'd be all for, but
that is a completely different set of work).


>
> Nothing in Plan D or Native Messaging would prevent the creation of a
> per-message WebSocket load balancer. In fact you could also do one today
> using RFC6455 alone. Under Plan D, said balancer would require an RFC6455
> parser as it would need to add/remove messages from a stream. Under nativ=
e
> messaging obviously not. The load balancer would clearly need to understa=
nd
> the application semantics of that specific stream (i.e. this is a loosely
> coupled message stream that can be balanced). Information in the URL,
> subprotocol, headers, etc could be used to communicate this application
> specific information.
>

Yup.

-=3DR

--089e0153852edab64304f1c1cfa7
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><br><div class=3D"gmail_extra"><br><br><div class=3D"gmail=
_quote">On Thu, Feb 6, 2014 at 7:26 AM, Peter Thorson <span dir=3D"ltr">&lt=
;<a href=3D"mailto:webmaster@zaphoyd.com" target=3D"_blank">webmaster@zapho=
yd.com</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div style=3D"word-wrap:break-word"><br><div=
><div class=3D"im"><div>On Feb 6, 2014, at 0:04, Yutaka Hirano &lt;<a href=
=3D"mailto:yhirano@chromium.org" target=3D"_blank">yhirano@chromium.org</a>=
&gt; wrote:</div>
<br><blockquote type=3D"cite"><div dir=3D"ltr"><div><blockquote class=3D"gm=
ail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-l=
eft-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">As I s=
aid before, enabling intermediaries to transfer WS over HTTP/2.0 *without a=
n RFC6455 parser* is important, I think.</blockquote>
</div></div></blockquote></div><blockquote type=3D"cite"><blockquote type=
=3D"cite">Agree. And Plan D allows this.<br></blockquote><div class=3D"im">=
I&#39;m sorry, I don&#39;t understand how. Can you explain?</div></blockquo=
te><div>
<br></div><div>Specifically for Plan D..</div><div>An RFC6455 parser is req=
uired to:</div><div>- Add or remove messages from the stream</div><div>- De=
termine WS frame and message boundaries</div><div>- Read message contents</=
div>
<div>An RFC6455 parser is NOT required to:</div><div>- Proxy the stream</di=
v><div>- Load balance the stream</div><div>- Mux a WS stream into an HTTP/2=
.0 stream</div><div>- Demux a WS stream out of an HTTP/2.0 stream</div>
<div>- Flush a stream using an HTTP/2.0 level FLUSH_HINT</div><div><br></di=
v><div>If it is important that intermediaries or endpoints be able to read =
and manipulate these streams on a per-message or per-frame level without an=
 RFC6455 parser then a tunneled solution is a bad choice. If all the interm=
ediaries need to do is move the bytes around then it is fine.</div>
</div></div></blockquote><div><br></div><div>Exactly.</div><div>=C2=A0</div=
><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1=
px #ccc solid;padding-left:1ex"><div style=3D"word-wrap:break-word"><div><d=
iv class=3D"im">
<br><blockquote type=3D"cite"><div dir=3D"ltr"><div>No matter what protocol=
 we choose, we must observe the HTTP/2.0 framing protocol, which means:</di=
v>
<div>=C2=A0 - The size of a frame must not exceed the fixed limit.</div><di=
v>=C2=A0 - We must set and respect the END_STREAM flag appropriately.</div>=
<div>=C2=A0 - ...</div><div>Adding one rule (MSG_DONE) is harmless, and I t=
hink it has a merit.<span style=3D"color:rgb(34,34,34)">=C2=A0</span></div>
</div></blockquote></div></div></div></blockquote><blockquote class=3D"gmai=
l_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left=
:1ex"><div style=3D"word-wrap:break-word"><div><div class=3D"im"><div><br><=
/div>
</div><div>I agree that the MSG_DONE concept in general is a useful general=
 purpose flag to have in HTTP/2.0. I disagree that it is harmless to use it=
 as a 1:1 message delimiter in a tunneled RFC6455 stream.=C2=A0</div></div>=
</div>
</blockquote><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;b=
order-left:1px #ccc solid;padding-left:1ex"><div style=3D"word-wrap:break-w=
ord"><div><div class=3D"im"><div><br></div><div><blockquote type=3D"cite"><=
/blockquote>
<blockquote type=3D"cite"><blockquote type=3D"cite">MSG_DONE is used just a=
s a flush signal (at HTTP/2.0 frame level) in Plan A.<br>I don&#39;t think =
it breaks permessage-priority.<br></blockquote><br>If it&#39;s intended use=
 is a Flush signal to intermedaries, then we should probably<br>
<br>a) call it FLUSH_HINT, and<br>b) allow multiple WS frames with FIN bit =
inside a single SPDY data frame.<br><br>FLUSH_HINT would be set if _any_ WS=
 frame inside the SPDY data frame would have it&#39;s FIN bit set.<br><br>
This would do away with one additional limitation of Plan A over Plan D:<br=
><br>As it stands now, Plan A does not allow batching of multiple WS messag=
es inside a single SPDY data frame.</blockquote></div></div></div></div>
</blockquote><div>I wouldn&#39;t really find FLUSH_HINT useful for anything=
 unless the application-logic in the webpage could affect it.</div><div>But=
.. there is no such capability in the API right now.</div><div><br></div>
<div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8=
ex;border-left:1px #ccc solid;padding-left:1ex"><div style=3D"word-wrap:bre=
ak-word"><div><div class=3D"im"><div><br></div></div><div>I am fine with a =
=E2=80=9Cflush hint=E2=80=9D flag where the sending endpoint *may/should* a=
dd it to HTTP/2.0 frames that the application prefers not to have buffered/=
delayed by intermediaries. I am fine with a separate bit named FLUSH_HINT (=
useful for either Plan D or native messaging) or re-using MSG_DONE to imply=
 =E2=80=9Cflush hint=E2=80=9D for Plan D streams if MSG_DONE is used elsewh=
ere by HTTP/2.0. I would not want to define specific scenarios where a flus=
h must occur.</div>
</div></div></blockquote><div><br></div><div><br></div><div>HTTP2&#39;s MSG=
_DONE (under a different name) is in the next implementation draft, coming =
out shortly.</div><div>For HTTP on HTTP/2 it is use for delimiting chunk bo=
undaries on an HTTP stream, since the lack of this has proven painful for s=
ome. Intermediaries are disallowed from coalescing data across frame where =
this is set.</div>
<div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8=
ex;border-left:1px #ccc solid;padding-left:1ex"><div style=3D"word-wrap:bre=
ak-word"><div><div><br></div><div>Re: Load balancing</div><div><br></div><d=
iv>Robertos point that HTTP messages traditionally must be answered in orde=
r certainly makes HTTP less appealing for the unordered, unrelated RPC styl=
e message case. I can see how an application might use WebSocket today in o=
rder to get out of order message return delivery.</div>
<div><br></div><div>I am curious if there has there been any discussion to =
relaxing that requirement in HTTP/2.0?</div></div></div></blockquote><div><=
br></div><div>Unfortunately, one can&#39;t relax that requirement and still=
 have the backwards-compatible API that exists today: while HTTP/2 can deli=
ver on this, HTTP/1.X cannot.</div>
<div>That one need not change their content to deploy HTTP/2 (etc.) is a ma=
jor driver of deployment, so it isn&#39;t something that could be lightly c=
hanged.</div><div>IMHO, we&#39;d have to invent a new API (which, honestly =
I&#39;d be all for, but that is a completely different set of work).</div>
<div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8=
ex;border-left:1px #ccc solid;padding-left:1ex"><div style=3D"word-wrap:bre=
ak-word"><div><div><br></div><div>Nothing in Plan D or Native Messaging wou=
ld prevent the creation of a per-message WebSocket load balancer. In fact y=
ou could also do one today using RFC6455 alone. Under Plan D, said balancer=
 would require an RFC6455 parser as it would need to add/remove messages fr=
om a stream. Under native messaging obviously not. The load balancer would =
clearly need to understand the application semantics of that specific strea=
m (i.e. this is a loosely coupled message stream that can be balanced). Inf=
ormation in the URL, subprotocol, headers, etc could be used to communicate=
 this application specific information.</div>
</div></div></blockquote><div><br></div><div>Yup.</div><div><br></div><div>=
-=3DR</div></div></div></div>

--089e0153852edab64304f1c1cfa7--

From webmaster@zaphoyd.com  Thu Feb  6 15:18:20 2014
Return-Path: <webmaster@zaphoyd.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 822F61A051A for <hybi@ietfa.amsl.com>; Thu,  6 Feb 2014 15:18:20 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.9
X-Spam-Level: 
X-Spam-Status: No, score=-1.9 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, RCVD_IN_DNSWL_NONE=-0.0001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 8LiYd6VSlbc8 for <hybi@ietfa.amsl.com>; Thu,  6 Feb 2014 15:18:18 -0800 (PST)
Received: from authsmtp01.uchicago.edu (authsmtp01.uchicago.edu [128.135.12.121]) by ietfa.amsl.com (Postfix) with ESMTP id B18AF1A0517 for <hybi@ietf.org>; Thu,  6 Feb 2014 15:18:18 -0800 (PST)
Received: from ranna.uchicago.edu (ranna.uchicago.edu [128.135.45.206]) (Authenticated sender: zaphoyd) by authsmtp01.uchicago.edu (Postfix) with ESMTP id 0870C49806B; Thu,  6 Feb 2014 17:18:16 -0600 (CST)
Content-Type: text/plain; charset=windows-1252
Mime-Version: 1.0 (Mac OS X Mail 7.1 \(1827\))
From: Peter Thorson <webmaster@zaphoyd.com>
In-Reply-To: <CAGzyod7BA_UXjN4r+Qr+P33QsPpeWWotxiScTpzqyhvbtPgNYQ@mail.gmail.com>
Date: Thu, 6 Feb 2014 17:18:16 -0600
Content-Transfer-Encoding: quoted-printable
Message-Id: <2B0B9332-D002-460A-8D50-B90B55999F42@zaphoyd.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <CABih n6GGezUU2aSboK0wsxYX3TBWTz5Av6-ZFJTPy6PQ50LrTQ@mail.gmail.com> <740FF040-F1E9-43E6-ACDE-0014FE4D012F@zaphoyd.com> <CABihn6HM_yXSB9RapNEQEG0_9OYfpHG3uH+=wnDwRJAFbh9EJA@mail.gmail.com> <0F978119-88F2-49CE-98C6-2A640527AA64@zaphoyd.com> <CAGzyod7BA_UXjN4r+Qr+P33QsPpeWWotxiScTpzqyhvbtPgNYQ@mail.gmail.com>
To: Roberto Peon <fenix@google.com>
X-Mailer: Apple Mail (2.1827)
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 06 Feb 2014 23:18:20 -0000

On Feb 6, 2014, at 13:23 , Roberto Peon <fenix@google.com> wrote:

> Unfortunately, one can't relax that requirement and still have the =
backwards-compatible API that exists today: while HTTP/2 can deliver on =
this, HTTP/1.X cannot.
> That one need not change their content to deploy HTTP/2 (etc.) is a =
major driver of deployment, so it isn't something that could be lightly =
changed.
> IMHO, we'd have to invent a new API (which, honestly I'd be all for, =
but that is a completely different set of work).

This statement gets at the heart of the present discussion.

I see the "plan D" option treating WebSocket as HTTP/1.1 is treated by =
the rest of HTTP/2.0, an existing protocol to be muxed mostly "as is" =
for the benefit of backwards compatibility with existing systems. I see =
a Native Messaging option as the "new API=94 that would allow a cleaner =
interface for pure HTTP/2.0 applications going forward.

For me, the reality is that any piece of HTTP/2.0 software that I write =
or deploy in the foreseeable future is going to demux the message =
stream, convert to RFC6455, and forward to existing systems. This is =
both due to backwards compatibility as well as the nature of a large =
chunk of my user base (particularly the embedded non-browser clients) =
having no use for any of the features that HTTP/2.0 adds over RFC6455 =
and every reason to want to avoid additional overhead.

Plan D provides a straightforward path for this sort of operation and =
simplifies things considerably for a good chunk of the =
infrastructure-type intermediaries without imposing any additional =
limitations on anyone (other than perhaps additional library/protocol =
dependencies for the more ambitious intermediaries).

That said, even if I plan to proxy HTTP/2.0 messages right back into =
RFC6455. I also see the future benefit to having a cleaner messaging =
option that is aware of HTTP/2.0's new features. My first choice is for =
this option as it seems more sane long term.. as long as the new =
protocol is something worth breaking backwards compatibility for. The =
Plan D case is strong as well and I would not oppose that if the =
consensus says that backwards compatibility is more important.=

From yhirano@google.com  Thu Feb  6 23:05:48 2014
Return-Path: <yhirano@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 743D21A05B8 for <hybi@ietfa.amsl.com>; Thu,  6 Feb 2014 23:05:48 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.913
X-Spam-Level: 
X-Spam-Status: No, score=-1.913 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.535, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id uXIWr9bQQAdl for <hybi@ietfa.amsl.com>; Thu,  6 Feb 2014 23:05:45 -0800 (PST)
Received: from mail-qc0-x235.google.com (mail-qc0-x235.google.com [IPv6:2607:f8b0:400d:c01::235]) by ietfa.amsl.com (Postfix) with ESMTP id 3AC971A05B7 for <hybi@ietf.org>; Thu,  6 Feb 2014 23:05:45 -0800 (PST)
Received: by mail-qc0-f181.google.com with SMTP id e9so5169526qcy.40 for <hybi@ietf.org>; Thu, 06 Feb 2014 23:05:43 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=A++8z6DzNfJXcX+4Dbm8Pb0JQJ6ZeUxWVJSHtP+lojo=; b=bLpSuHKA2u0IdoUotaMSZpdJHOYmlRe+sucqdqnOzF1l1bMS7mXgagJllEGMWXPSrP a2TmkOjFe4VGOM5M0J7kbz3mctGsZc/EJNhMlbzVOMl/3oeKp5fTx1x9R9c012ETMLJi rKbMrgmLtQVQfqvldBV4Kpp/hEDqISAc4kW3V/wX9TqVn/zo7s024fI/HXN25aKdoATF xWE9lLAnu153VSD9PCm6qvcYajRMnefnGhvYG7rmodEEi/WHjgYgTXmnX15ugAHsUAs7 I77hrfm3CqZBVCFmVnpyHPmRGX842ZoHQUnJQ/wahyFD7/iAyGUNbmpkZ66pFqQhcokI +N9A==
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=chromium.org; s=google; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=A++8z6DzNfJXcX+4Dbm8Pb0JQJ6ZeUxWVJSHtP+lojo=; b=KuMwJfzKly2bBwgolUVcl9Aiq5xLoKJaWvACl/sfpH1sf1Nh2JntgZcKRBUKKERK2S w7ZIvKqr/nnRsiJbeiyV1FZcUzliez/IHLyvdpCP3yI4pYoPDtWc3HRUIOwqpqKNZyXu lnOJxTXrxsu98df7VKkNM+/Z0o+p3w18FQ7cc=
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:sender:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=A++8z6DzNfJXcX+4Dbm8Pb0JQJ6ZeUxWVJSHtP+lojo=; b=eAIgKppvKOYlc9ezMNzwn21GnxMEOE/XFdd8TK6BJDqjHLAtR2BmmQPE7lu4HB77aD Z9Pow1J5jyCpejc77h+/J4zu8tRKOLb+b69DAkn6ib5w6IkSnwJjCuu+bwd48Fm9gEJn 0uQEBgMAi3bfSa1CJL3+SvwTSeiejgye+bx82oMNNgIAJYrstfBh9Wc6rhr7mNlDQZxK ZvBsPeKWWmDoRc7Rrh3o47wJquhelnQT/8IoOE/FLNkOV+UrUo9UnjEAzJCnWUJkAQP9 6IaiHR8Mlt2p6/UUkEWpvi/LHueD15vwEyDhDNNqDH1W6pulUgi87Xr14TTP4u70fEzL nC3A==
X-Gm-Message-State: ALoCoQkb/tmdEzkQ0va6jWCDZloJn6SN5iT6Ko5I7f2/p5lANE3+V0VMgJ6jFMGXk4kavs35g/KeWLvcUne0DHH/kyp3H7/LiLcaI79GFwijEQDClXZ/V5395vrZfJVGpijOjkiBROX8/f+Jg25HPnY2mf3DsnecZuGQZkrujEaUVFHX0oAeqKXI2o7JTMNjpSRVbhVSBMHH
MIME-Version: 1.0
X-Received: by 10.140.22.145 with SMTP id 17mr18171640qgn.0.1391756743704; Thu, 06 Feb 2014 23:05:43 -0800 (PST)
Sender: yhirano@google.com
Received: by 10.140.109.202 with HTTP; Thu, 6 Feb 2014 23:05:43 -0800 (PST)
In-Reply-To: <2B0B9332-D002-460A-8D50-B90B55999F42@zaphoyd.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <740FF040-F1E9-43E6-ACDE-0014FE4D012F@zaphoyd.com> <CABihn6HM_yXSB9RapNEQEG0_9OYfpHG3uH+=wnDwRJAFbh9EJA@mail.gmail.com> <0F978119-88F2-49CE-98C6-2A640527AA64@zaphoyd.com> <CAGzyod7BA_UXjN4r+Qr+P33QsPpeWWotxiScTpzqyhvbtPgNYQ@mail.gmail.com> <2B0B9332-D002-460A-8D50-B90B55999F42@zaphoyd.com>
Date: Fri, 7 Feb 2014 16:05:43 +0900
X-Google-Sender-Auth: wVwuYCYz0wd-2lRB_3oidPHT-Cg
Message-ID: <CABihn6Hykczbu0YmUDuFcjVt2+xyeYzmdOpEkiZmP-2Vzuiuow@mail.gmail.com>
From: Yutaka Hirano <yhirano@chromium.org>
To: Peter Thorson <webmaster@zaphoyd.com>
Content-Type: multipart/alternative; boundary=001a11c13010529eb204f1cb9e7e
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 07 Feb 2014 07:05:48 -0000

--001a11c13010529eb204f1cb9e7e
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

Peter, Roberto, Tobias,
I think we should shrink the battle line.
I propose to choose if we preserve the RFC6455 semantics or not (for
framing).

*Preserve the RFC6455 semantics*
We won't change the RFC6455 semantics, which means we will discuss how to
encapsulate WebSocket frame(s) into HTTP/2.0 frame(s).
All RFC6455 extensions will be supported.
To preserve the RFC6455 semantics, I modify the proposed plans.

Plan A (WS-TUNNEL with MSG_DONE):
 - MSG_DONE is used as a FLUSH signal
 - multiple WebSocket message can be batched in a single HTTP/2.0 frame
I believe that this preserves the RFC6455 semantics.
I support this plan.

Plan C (HEADERS + DATAs for a WebSocket frame):
 - All headers including FIN and length are included in HEADERS.
 - MSG_DONE is mapped to the end of a **WebSocket frame**
  - This means that a WebSocket connection is a stream-of-WS_frames, which
is completely correct.
Hence HEADERS + DATAs represent a complete WebSocket frame.
I support this plan.
I believe Roberto supports this plan.

Plan D (WS-TUNNEL only): as is
Tobias and Peter support this plan.

Choosing "Preserve the RFC6455 semantics", means we choose framing method
from these three plans.
I know that all of us hate some of the plans, but I think this option is
better, because
 - All of us like some of the plans.
 - We can simplify the discussion.
 - if "Not preserve the RFC6455 semantics" seems appealing now, at least
partially that is because it is not concrete.

As the only supporter of Plan A, I promise that I will retract the plan if
a critical problem is found on it (it is not compatible with RFC6455,
MSG_DONE can't be used for the flushing purpose, FLUSH can't be introduced
and so on).

*Not preserve the RFC6455 semantics*
Choosing this option means we abandon Plan A, C and D.

*Questions*
As you notice, there are two questions from me.
1: Are you OK with deciding NOW if we should preserve the RFC6455 semantics
or not?
2: (If the answer for 1 is yes) Should we preserve the RFC6455 semantics?

My answer is, yes & yes.



On Fri, Feb 7, 2014 at 8:18 AM, Peter Thorson <webmaster@zaphoyd.com> wrote=
:

>
> On Feb 6, 2014, at 13:23 , Roberto Peon <fenix@google.com> wrote:
>
> > Unfortunately, one can't relax that requirement and still have the
> backwards-compatible API that exists today: while HTTP/2 can deliver on
> this, HTTP/1.X cannot.
> > That one need not change their content to deploy HTTP/2 (etc.) is a
> major driver of deployment, so it isn't something that could be lightly
> changed.
> > IMHO, we'd have to invent a new API (which, honestly I'd be all for, bu=
t
> that is a completely different set of work).
>
> This statement gets at the heart of the present discussion.
>
> I see the "plan D" option treating WebSocket as HTTP/1.1 is treated by th=
e
> rest of HTTP/2.0, an existing protocol to be muxed mostly "as is" for the
> benefit of backwards compatibility with existing systems. I see a Native
> Messaging option as the "new API=E2=80=9D that would allow a cleaner inte=
rface for
> pure HTTP/2.0 applications going forward.
>
> For me, the reality is that any piece of HTTP/2.0 software that I write o=
r
> deploy in the foreseeable future is going to demux the message stream,
> convert to RFC6455, and forward to existing systems. This is both due to
> backwards compatibility as well as the nature of a large chunk of my user
> base (particularly the embedded non-browser clients) having no use for an=
y
> of the features that HTTP/2.0 adds over RFC6455 and every reason to want =
to
> avoid additional overhead.
>
> Plan D provides a straightforward path for this sort of operation and
> simplifies things considerably for a good chunk of the infrastructure-typ=
e
> intermediaries without imposing any additional limitations on anyone (oth=
er
> than perhaps additional library/protocol dependencies for the more
> ambitious intermediaries).
>
> That said, even if I plan to proxy HTTP/2.0 messages right back into
> RFC6455. I also see the future benefit to having a cleaner messaging opti=
on
> that is aware of HTTP/2.0's new features. My first choice is for this
> option as it seems more sane long term.. as long as the new protocol is
> something worth breaking backwards compatibility for. The Plan D case is
> strong as well and I would not oppose that if the consensus says that
> backwards compatibility is more important.

--001a11c13010529eb204f1cb9e7e
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div>Peter, Roberto, Tobias,<br></div><div>I think we shou=
ld shrink the battle line.</div><div>I propose to choose if we preserve the=
 RFC6455 semantics or not (for framing).</div><div><br></div><div><b>Preser=
ve the RFC6455 semantics</b></div>

<div>We won&#39;t change the RFC6455 semantics, which means we will discuss=
 how to encapsulate WebSocket frame(s) into HTTP/2.0 frame(s).</div><div>Al=
l RFC6455 extensions will be supported.</div><div>To preserve the RFC6455 s=
emantics, I modify the proposed plans.</div>

<div><br></div><div>Plan A (WS-TUNNEL with MSG_DONE):</div><div>=C2=A0- MSG=
_DONE is used as a FLUSH signal</div><div>=C2=A0- multiple WebSocket messag=
e can be batched in a single HTTP/2.0 frame</div><div>I believe that this p=
reserves the RFC6455 semantics.</div>

<div><div><div>I support this plan.</div></div></div><div><br></div><div>Pl=
an C (HEADERS + DATAs for a WebSocket frame):</div><div>=C2=A0- All headers=
 including FIN and length are included in HEADERS.</div><div>=C2=A0- MSG_DO=
NE is mapped to the end of a **WebSocket frame**</div>

<div>=C2=A0 - This means that a WebSocket connection is a stream-of-WS_fram=
es, which is completely correct.</div><div>Hence HEADERS + DATAs represent =
a complete WebSocket frame.</div><div><div>I support this plan.</div><div>I=
 believe Roberto supports this plan.</div>

<div><br></div></div><div>Plan D (WS-TUNNEL only): as is</div><div>Tobias a=
nd Peter support this plan.</div><div><br></div><div>Choosing &quot;Preserv=
e the RFC6455 semantics&quot;, means we choose framing method from these th=
ree plans.</div>

<div>I know that all of us hate some of the plans, but I think this option =
is better, because</div><div>=C2=A0- All of us like some of the plans.</div=
><div>=C2=A0- We can simplify the discussion.</div><div>=C2=A0- if &quot;No=
t preserve the RFC6455 semantics&quot; seems appealing now, at least partia=
lly that is because it is not concrete.</div>

<div><br></div><div>As the only supporter of Plan A, I promise that I will =
retract the plan if a critical problem is found on it (it is not compatible=
 with RFC6455, MSG_DONE can&#39;t be used for the flushing purpose, FLUSH c=
an&#39;t be introduced and so on).</div>

<div><br></div><div><b>Not preserve the RFC6455 semantics</b><br></div><div=
>Choosing this option means we abandon Plan A, C and D.</div><div><br></div=
><div><b>Questions</b></div><div>As you notice, there are two questions fro=
m me.</div>
<div>1: Are you OK with deciding NOW if we should preserve the RFC6455 sema=
ntics or not?</div>
<div>2: (If the answer for 1 is yes) Should we preserve the RFC6455 semanti=
cs?</div><div><br></div><div>My answer is, yes &amp; yes.</div><div><br></d=
iv></div><div class=3D"gmail_extra"><br><br><div class=3D"gmail_quote">On F=
ri, Feb 7, 2014 at 8:18 AM, Peter Thorson <span dir=3D"ltr">&lt;<a href=3D"=
mailto:webmaster@zaphoyd.com" target=3D"_blank">webmaster@zaphoyd.com</a>&g=
t;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div class=3D""><br>
On Feb 6, 2014, at 13:23 , Roberto Peon &lt;<a href=3D"mailto:fenix@google.=
com">fenix@google.com</a>&gt; wrote:<br>
<br>
&gt; Unfortunately, one can&#39;t relax that requirement and still have the=
 backwards-compatible API that exists today: while HTTP/2 can deliver on th=
is, HTTP/1.X cannot.<br>
&gt; That one need not change their content to deploy HTTP/2 (etc.) is a ma=
jor driver of deployment, so it isn&#39;t something that could be lightly c=
hanged.<br>
&gt; IMHO, we&#39;d have to invent a new API (which, honestly I&#39;d be al=
l for, but that is a completely different set of work).<br>
<br>
</div>This statement gets at the heart of the present discussion.<br>
<br>
I see the &quot;plan D&quot; option treating WebSocket as HTTP/1.1 is treat=
ed by the rest of HTTP/2.0, an existing protocol to be muxed mostly &quot;a=
s is&quot; for the benefit of backwards compatibility with existing systems=
. I see a Native Messaging option as the &quot;new API=E2=80=9D that would =
allow a cleaner interface for pure HTTP/2.0 applications going forward.<br>

<br>
For me, the reality is that any piece of HTTP/2.0 software that I write or =
deploy in the foreseeable future is going to demux the message stream, conv=
ert to RFC6455, and forward to existing systems. This is both due to backwa=
rds compatibility as well as the nature of a large chunk of my user base (p=
articularly the embedded non-browser clients) having no use for any of the =
features that HTTP/2.0 adds over RFC6455 and every reason to want to avoid =
additional overhead.<br>

<br>
Plan D provides a straightforward path for this sort of operation and simpl=
ifies things considerably for a good chunk of the infrastructure-type inter=
mediaries without imposing any additional limitations on anyone (other than=
 perhaps additional library/protocol dependencies for the more ambitious in=
termediaries).<br>

<br>
That said, even if I plan to proxy HTTP/2.0 messages right back into RFC645=
5. I also see the future benefit to having a cleaner messaging option that =
is aware of HTTP/2.0&#39;s new features. My first choice is for this option=
 as it seems more sane long term.. as long as the new protocol is something=
 worth breaking backwards compatibility for. The Plan D case is strong as w=
ell and I would not oppose that if the consensus says that backwards compat=
ibility is more important.</blockquote>
</div><br></div>

--001a11c13010529eb204f1cb9e7e--

From tyoshino@google.com  Thu Feb  6 23:30:24 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id B69701A05B7 for <hybi@ietfa.amsl.com>; Thu,  6 Feb 2014 23:30:24 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.613
X-Spam-Level: 
X-Spam-Status: No, score=-1.613 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, MIME_8BIT_HEADER=0.3, RP_MATCHES_RCVD=-0.535, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id B-as5ymqlAuJ for <hybi@ietfa.amsl.com>; Thu,  6 Feb 2014 23:30:22 -0800 (PST)
Received: from mail-wg0-x229.google.com (mail-wg0-x229.google.com [IPv6:2a00:1450:400c:c00::229]) by ietfa.amsl.com (Postfix) with ESMTP id 833291A05AC for <hybi@ietf.org>; Thu,  6 Feb 2014 23:30:22 -0800 (PST)
Received: by mail-wg0-f41.google.com with SMTP id n12so501832wgh.4 for <hybi@ietf.org>; Thu, 06 Feb 2014 23:30:20 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:from:date:message-id:subject:to:cc:content-type; bh=mjiXlSM2CrWPP1DKhKwcg+XjK+JWMIRPDmcw3QX4alg=; b=KBfwWqI8esVwYJVOqdxkdMp/BeFEtmyidRCLK5Ne6yo5MPS84QJ0VVJMWtbHR/4EOY wOJmLgeTB0KVZvuFLk9ZiorL37E+Gc+1KHIyKOhbei2wAnn7wwZAkGn5LTcQZGa7g2Jm ncrRs+GyEMcw9j6kxoW7/GWeygBcD4Hqpvg8PFIUX71Edjsm1FRijhkegp1l9nLaXDnL hYOXRtW/aTu0C/uYqlDF7mG7ECH89FzG0+Q405P3gynmjFg1DaKP10jYH94N4Bclsp/K Rwcw9UcZHvd9Ri6HsWo2hnF5t5hOP4mSix0v3+Bj7P481kp/nI9oL0sp3EyhOuQH893j E4XQ==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:from:date:message-id:subject:to:cc :content-type; bh=mjiXlSM2CrWPP1DKhKwcg+XjK+JWMIRPDmcw3QX4alg=; b=l/3hnw3E3wZje0SciK7OepFopz/BdMcgfp4OdbEm/54ICeB1c6QB0dyIlrdHGPkQWz 6JYH0qULc+8WfUH7T8Bnne4lyFToXaaifONcc+iJgO5KKazvWohF0biTMQ9HuBEOtQs4 MWkE0VKTrlO0ayYMAgKUJ7J4CMH2WgtIkz2HJ1CMtzno6vHke1lfPaW4mwndW3UDN0HM 9oe6PuQJtDjByI2iHue9G9v7wVb+huDcZ09/w78lYFAGhvKS35nzIe2NxKo60tSJ6zQv 444lX6Kpcs3XLp1wrltvSJ7NYGN2x7Mq9299t26ov8jRnrghrQu3y03ViA1cZw+wllOJ 6P7Q==
X-Gm-Message-State: ALoCoQnDUuBUQl/BUpYIwipOkMbtVpFNi8/tQG0JXIN6OjLwfPB2IepYNC4GIaApT1oLixdmVGecDbE5ZcxIqQNRSxL+GNIXAUCgzTMwS51sY3UE4HkQB155/uRloTkHHIvpMr7UmXxg0UAi1otDjLzNQaHrG4ue26fjbuFb9pZsDtRWrCUsW2e9grAg4skhLB+lTYHi7pve
X-Received: by 10.180.10.105 with SMTP id h9mr2610624wib.11.1391758220712; Thu, 06 Feb 2014 23:30:20 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Thu, 6 Feb 2014 23:30:00 -0800 (PST)
From: Takeshi Yoshino <tyoshino@google.com>
Date: Fri, 7 Feb 2014 16:30:00 +0900
Message-ID: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com>
To: =?UTF-8?Q?Martin_J=2E_D=C3=BCrst?= <duerst@it.aoyama.ac.jp>
Content-Type: multipart/alternative; boundary=001a11c263665c2cb104f1cbf6bc
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 07 Feb 2014 07:30:24 -0000

--001a11c263665c2cb104f1cbf6bc
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

I remember that in early days of the standardization, somebody was
proposing (*) that we should have a message length field in the first
fragment to allow for allocating buffer for full message at the beginning.

On Thu, Jan 16, 2014 at 3:27 PM, "Martin J. D=C3=BCrst"
<duerst@it.aoyama.ac.jp>wrote:

> On 2014/01/16 15:05, Yutaka Hirano wrote:
>
>> Hi Joakim,
>>
>> Thanks for the comment.
>>
>> Knowing the size of a frame at the start of the frame, traditional
>>
>>> websocket behavior, is desired, as it allows us to fast fail for the 10=
09
>>> (message too big) use case.
>>>
>>
>> Do you know how effective it is?
>>
>> Having no prior knowledge of frame size is undesirable to us.
>>
>> On the other hand, the frame sender have more freedom because they don't
>> have to know the size of the frame in front.
>>
>
> My understanding of why messages are split into frames was that this woul=
d
> allow to start a message without knowing its length. But if you don't kno=
w
> the size of a frame before sending it, then the right solution is to make
> it smaller. Frames are just arbitrary pieces of a message.
>

We should be clear if we want to make it able to
(a) know the length of WebSocket frame on receiving the header of each
fragment
(b) know the length of WebSocket message on receiving the first fragment

As the proposal (*) has been rejected and not in the RFC6455 header, (b) is
not true even in RFC6455. What Joakim said should be to keep (a) which is
true now in RFC6455.

But if a message is fragmented into smaller frames, we cannot reject (and
send 1009) the message on receiving the first frame. "Whether we can
fast-reject a big message on receiving some non-final frame" depends on how
the peer fragments a message.

In some plans discussed in the main thread, it's planned to drop frame
length info for efficiency. But something else such as the length field of
HTTP/2.0 DATA frame tells how much data the fragment (when multiple DATAs
convey a WebSocket frame, each DATA would be called fragments of a frame)
contains. So, we won't be able to know the length of fragments
(original/encapsulated WebSocket frames) on receiving each of them, but
still be able to know the length of sub-fragments (e.g. HTTP/2.0 data
frames).

I think this change is acceptable and shouldn't be considered as breaking
of RFC6455 semantics. Even in the mux extension I was editing (
http://tools.ietf.org/html/draft-ietf-hybi-websocket-multiplexing-11), I
made the same change. The length of the encapsulated frame is unknown until
receiving all the fragments of the encapsulating message. No one was
objecting to that when we were discussing the spec, IIRC.

--001a11c263665c2cb104f1cbf6bc
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_extra">I remember that in early days o=
f the standardization, somebody was proposing (*) that we should have a mes=
sage length field in the first fragment to allow for allocating buffer for =
full message at the beginning.<br>

</div><div class=3D"gmail_extra"><br><div class=3D"gmail_quote">On Thu, Jan=
 16, 2014 at 3:27 PM, &quot;Martin J. D=C3=BCrst&quot; <span dir=3D"ltr">&l=
t;<a href=3D"mailto:duerst@it.aoyama.ac.jp" target=3D"_blank">duerst@it.aoy=
ama.ac.jp</a>&gt;</span> wrote:<br>

<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div class=3D"im">On 2014/01/16 15:05, Yutaka Hirano wrote=
:<br>


<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex">
Hi Joakim,<br>
<br>
Thanks for the comment.<br>
<br>
Knowing the size of a frame at the start of the frame, traditional<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex">
websocket behavior, is desired, as it allows us to fast fail for the 1009<b=
r>
(message too big) use case.<br>
</blockquote>
<br>
Do you know how effective it is?<br>
<br>
Having no prior knowledge of frame size is undesirable to us.<br>
<br>
On the other hand, the frame sender have more freedom because they don&#39;=
t<br>
have to know the size of the frame in front.<br>
</blockquote>
<br></div>
My understanding of why messages are split into frames was that this would =
allow to start a message without knowing its length. But if you don&#39;t k=
now the size of a frame before sending it, then the right solution is to ma=
ke it smaller. Frames are just arbitrary pieces of a message.<br>

</blockquote><div><br></div><div>We should be clear if we want to make it a=
ble to</div><div>(a) know the length of WebSocket frame on receiving the he=
ader of each fragment</div><div>(b) know the length of WebSocket message on=
 receiving the first fragment</div>

<div>=C2=A0</div><div>As the proposal (*) has been rejected and not in the =
RFC6455 header, (b) is not true even in RFC6455. What Joakim said should be=
 to keep (a) which is true now in RFC6455.</div><div><br></div><div>But if =
a message is fragmented into smaller frames, we cannot reject (and send 100=
9) the message on receiving the first frame. &quot;Whether we can fast-reje=
ct a big message on receiving some non-final frame&quot; depends on how the=
 peer fragments a message.</div>

<div><br></div><div>In some plans discussed in the main thread, it&#39;s pl=
anned to drop frame length info for efficiency. But something else such as =
the length field of HTTP/2.0 DATA frame tells how much data the fragment (w=
hen multiple DATAs convey a WebSocket frame, each DATA would be called frag=
ments of a frame) contains. So, we won&#39;t be able to know the length of =
fragments (original/encapsulated WebSocket frames) on receiving each of the=
m, but still be able to know the length of sub-fragments (e.g. HTTP/2.0 dat=
a frames).</div>

<div><br></div><div>I think this change is acceptable and shouldn&#39;t be =
considered as breaking of RFC6455 semantics. Even in the mux extension I wa=
s editing (<a href=3D"http://tools.ietf.org/html/draft-ietf-hybi-websocket-=
multiplexing-11">http://tools.ietf.org/html/draft-ietf-hybi-websocket-multi=
plexing-11</a>), I made the same change. The length of the encapsulated fra=
me is unknown until receiving all the fragments of the encapsulating messag=
e. No one was objecting to that when we were discussing the spec, IIRC.</di=
v>

<div><br></div></div></div></div>

--001a11c263665c2cb104f1cbf6bc--

From tobias.oberstein@tavendo.de  Fri Feb  7 02:16:23 2014
Return-Path: <tobias.oberstein@tavendo.de>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id A3A9C1A036B for <hybi@ietfa.amsl.com>; Fri,  7 Feb 2014 02:16:23 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.9
X-Spam-Level: 
X-Spam-Status: No, score=-1.9 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, RCVD_IN_DNSWL_NONE=-0.0001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 55L2ijxkf-3U for <hybi@ietfa.amsl.com>; Fri,  7 Feb 2014 02:16:22 -0800 (PST)
Received: from EXHUB020-1.exch020.serverdata.net (exhub020-1.exch020.serverdata.net [206.225.164.28]) by ietfa.amsl.com (Postfix) with ESMTP id 6368F1A01FB for <hybi@ietf.org>; Fri,  7 Feb 2014 02:16:22 -0800 (PST)
Received: from EXVMBX020-12.exch020.serverdata.net ([169.254.3.11]) by EXHUB020-1.exch020.serverdata.net ([206.225.164.28]) with mapi; Fri, 7 Feb 2014 02:16:20 -0800
From: Tobias Oberstein <tobias.oberstein@tavendo.de>
To: Yutaka Hirano <yhirano@chromium.org>, Peter Thorson <webmaster@zaphoyd.com>
Date: Fri, 7 Feb 2014 02:16:17 -0800
Thread-Topic: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
Thread-Index: Ac8j0wVm2V/ehCEOSxmDr1gpUVPXbwAGW1lA
Message-ID: <634914A010D0B943A035D226786325D4446C36C47C@EXVMBX020-12.exch020.serverdata.net>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <740FF040-F1E9-43E6-ACDE-0014FE4D012F@zaphoyd.com> <CABihn6HM_yXSB9RapNEQEG0_9OYfpHG3uH+=wnDwRJAFbh9EJA@mail.gmail.com> <0F978119-88F2-49CE-98C6-2A640527AA64@zaphoyd.com> <CAGzyod7BA_UXjN4r+Qr+P33QsPpeWWotxiScTpzqyhvbtPgNYQ@mail.gmail.com> <2B0B9332-D002-460A-8D50-B90B55999F42@zaphoyd.com> <CABihn6Hykczbu0YmUDuFcjVt2+xyeYzmdOpEkiZmP-2Vzuiuow@mail.gmail.com>
In-Reply-To: <CABihn6Hykczbu0YmUDuFcjVt2+xyeYzmdOpEkiZmP-2Vzuiuow@mail.gmail.com>
Accept-Language: de-DE, en-US
Content-Language: de-DE
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
acceptlanguage: de-DE, en-US
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: base64
MIME-Version: 1.0
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 07 Feb 2014 10:16:23 -0000
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From tyoshino@google.com  Fri Feb  7 04:07:55 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 264321A1DBD for <hybi@ietfa.amsl.com>; Fri,  7 Feb 2014 04:07:55 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.913
X-Spam-Level: 
X-Spam-Status: No, score=-1.913 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.535, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Acg8DOpi1gm8 for <hybi@ietfa.amsl.com>; Fri,  7 Feb 2014 04:07:52 -0800 (PST)
Received: from mail-wi0-x230.google.com (mail-wi0-x230.google.com [IPv6:2a00:1450:400c:c05::230]) by ietfa.amsl.com (Postfix) with ESMTP id 3314E1A1DBC for <hybi@ietf.org>; Fri,  7 Feb 2014 04:07:52 -0800 (PST)
Received: by mail-wi0-f176.google.com with SMTP id hi5so760113wib.9 for <hybi@ietf.org>; Fri, 07 Feb 2014 04:07:51 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:from:date:message-id:subject:to:cc:content-type; bh=x9cA1XOr13srVWXWLM68MVfJoDz2lUOVOP7nx7K9kp4=; b=OCiwZklbPSAeuiaabfR2zEqzlHw5+WATTvgqSrc7TTlVKzMgsRi3WbgiXIANr61fHJ J9ZL+v1mvAtJ/KyjXR+CrtqamsoUH90fdM/0llQuPthI39o30CyNOx1U9gy9J522JnZh 9QlQmg9LyVg7+6yCsQNTqdaz/0DLzijDoO2qW8CmZCZDLAnKK/y73BzpHUzgEmZSjfiu zPu8UHikBEIEXpLecNisnD+QebbW6Gv9OdrV0S7JVmnAibehWkdeOzQfPnr+oUUh3Kyt 5jGLVTy4aPl1wTAj1yv3duVPlHJ749aJZRosZzKIM/b+lIWQQ+EbU3uzmE1rbMJhOrJY ChSA==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:from:date:message-id:subject:to:cc :content-type; bh=x9cA1XOr13srVWXWLM68MVfJoDz2lUOVOP7nx7K9kp4=; b=lGvJImNbTUl/XNk9PMOumpG46NIb1Pxv/xMOrlkRaZf3pREzZUfKSRJ4ZsflWvNswv nF46gnzuokyGgO0Jfi3i2+bng8PYUAC5YYp8Gx2da0BwjUAXshpEQGKZYLsdkRlIQfx9 bKOInv3mOdqinsua3ff0AWcF3xdFWh14GRWRSK59aN1l/VrLo/jUM11H/p3tc6LQGuVP 5UFQRFdEMe65Jl1Kjm6tiw6ucXoTysWwcURo67DpNNASm6gQakGotD2z+2N5+MTKN3/m 7ZOXQNWjl7lJliz/NVQW5mMx69N6QiYyDd946DQlxwN6lOD+VI5yc4JRjWGxCXD6byy2 tGmw==
X-Gm-Message-State: ALoCoQlXIR27PfXBgJ5dJyGrQwdoh/UHyOHOTAaaroZu/FJsNCfvBAuep3sfWzQh/jNlWty8qQzITk1t+E7QfcgROmYb7K113vOan9zbjqAe9AfekT/2uRzdvvt24FkfACy6PmJsgi7A5CJEMd9/2veSVIZFqFNIMzi5iyiZpmDYEmfUnytcN+/Ll6m3IpQR4Y/MpD/Dfe0S
X-Received: by 10.194.71.47 with SMTP id r15mr10410547wju.19.1391774871462; Fri, 07 Feb 2014 04:07:51 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Fri, 7 Feb 2014 04:07:31 -0800 (PST)
From: Takeshi Yoshino <tyoshino@google.com>
Date: Fri, 7 Feb 2014 21:07:31 +0900
Message-ID: <CAH9hSJaYgxHsxM12BC7a-=TG+Ga+x2EUVCcXvsfcYcgrKsvkxQ@mail.gmail.com>
To: Roberto Peon <fenix@google.com>
Content-Type: multipart/alternative; boundary=047d7bfd0172d2518504f1cfd655
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: [hybi] Replacing encapsulation method of the mux extension with END_SEGMENT (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 07 Feb 2014 12:07:55 -0000

--047d7bfd0172d2518504f1cfd655
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

Just FYI, here's how the WebSocket specific mux extension (which we were
developing in HyBi) would be like if its encapsulation method is replaced
with HTTP/2.0 DATA frames and END_SEGMENT (formerly MSG_DONE) signal.

If the WebSocket header fields are represented as HEADERS HTTP/2.0 frame,
it'll be the same as yhirano's plan C.

=3D=3D=3D=3D

Data frame 1
- 1bit: FIN of the encapsulated WebSocket frame
- 3bits: RSV1-3 of ...
- 4bits: Opcode of ...
- Rest: 1st part of payload of ...

Data frame 2
- 2nd part of payload of ...

...

Data frame N (with END_SEGMENT set)
- Last part of payload of ...

=3D=3D=3D=3D

See also http://www.ietf.org/mail-archive/web/hybi/current/msg10404.htmlfor
the reason why I think it's fine to exclude frame length.


On Fri, Feb 7, 2014 at 4:23 AM, Roberto Peon <fenix@google.com> wrote:

>
>
>
> On Thu, Feb 6, 2014 at 7:26 AM, Peter Thorson <webmaster@zaphoyd.com>wrot=
e:
>
>>
>> I am fine with a =E2=80=9Cflush hint=E2=80=9D flag where the sending end=
point
>> *may/should* add it to HTTP/2.0 frames that the application prefers not =
to
>> have buffered/delayed by intermediaries. I am fine with a separate bit
>> named FLUSH_HINT (useful for either Plan D or native messaging) or re-us=
ing
>> MSG_DONE to imply =E2=80=9Cflush hint=E2=80=9D for Plan D streams if MSG=
_DONE is used
>> elsewhere by HTTP/2.0. I would not want to define specific scenarios whe=
re
>> a flush must occur.
>>
>
>
> HTTP2's MSG_DONE (under a different name) is in the next implementation
> draft, coming out shortly.
> For HTTP on HTTP/2 it is use for delimiting chunk boundaries on an HTTP
> stream, since the lack of this has proven painful for some. Intermediarie=
s
> are disallowed from coalescing data across frame where this is set.
>

--047d7bfd0172d2518504f1cfd655
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote">Just=
 FYI, here&#39;s how the WebSocket specific mux extension (which we were de=
veloping in HyBi) would be like if its encapsulation method is replaced wit=
h HTTP/2.0 DATA frames and END_SEGMENT (formerly MSG_DONE) signal.</div>

<div class=3D"gmail_quote"><br></div><div class=3D"gmail_quote">If the WebS=
ocket header fields are represented as HEADERS HTTP/2.0 frame, it&#39;ll be=
 the same as yhirano&#39;s plan C.</div><div class=3D"gmail_quote"><br></di=
v>

<div class=3D"gmail_quote">=3D=3D=3D=3D</div><div class=3D"gmail_quote"><br=
></div><div class=3D"gmail_quote">Data frame 1</div><div class=3D"gmail_quo=
te">- 1bit: FIN of the encapsulated WebSocket frame</div><div class=3D"gmai=
l_quote">- 3bits: RSV1-3 of ...</div>

<div class=3D"gmail_quote">- 4bits: Opcode of ...</div><div class=3D"gmail_=
quote">- Rest: 1st part of payload of ...</div><div class=3D"gmail_quote"><=
br></div><div class=3D"gmail_quote">Data frame 2</div><div class=3D"gmail_q=
uote">

- 2nd part of payload of ...</div><div class=3D"gmail_quote"><br></div><div=
 class=3D"gmail_quote">...</div><div class=3D"gmail_quote"><br></div><div c=
lass=3D"gmail_quote">Data frame N (with END_SEGMENT set)</div><div class=3D=
"gmail_quote">

- Last part of payload of ...</div><div class=3D"gmail_quote"><br></div><di=
v class=3D"gmail_quote">=3D=3D=3D=3D</div><div class=3D"gmail_quote"><br></=
div><div class=3D"gmail_quote">See also=C2=A0<a href=3D"http://www.ietf.org=
/mail-archive/web/hybi/current/msg10404.html">http://www.ietf.org/mail-arch=
ive/web/hybi/current/msg10404.html</a> for the reason why I think it&#39;s =
fine to exclude frame length.</div>

<div class=3D"gmail_quote"><br></div><div class=3D"gmail_quote"><br></div><=
div class=3D"gmail_quote">On Fri, Feb 7, 2014 at 4:23 AM, Roberto Peon <spa=
n dir=3D"ltr">&lt;<a href=3D"mailto:fenix@google.com" target=3D"_blank">fen=
ix@google.com</a>&gt;</span> wrote:<br>

<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div dir=3D"ltr"><br><div class=3D"gmail_extra"><br><br><d=
iv class=3D"gmail_quote">

<div class=3D"im">On Thu, Feb 6, 2014 at 7:26 AM, Peter Thorson <span dir=
=3D"ltr">&lt;<a href=3D"mailto:webmaster@zaphoyd.com" target=3D"_blank">web=
master@zaphoyd.com</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div style=3D"word-wrap:break-word"><br><div><div><div>I a=
m fine with a =E2=80=9Cflush hint=E2=80=9D flag where the sending endpoint =
*may/should* add it to HTTP/2.0 frames that the application prefers not to =
have buffered/delayed by intermediaries. I am fine with a separate bit name=
d FLUSH_HINT (useful for either Plan D or native messaging) or re-using MSG=
_DONE to imply =E2=80=9Cflush hint=E2=80=9D for Plan D streams if MSG_DONE =
is used elsewhere by HTTP/2.0. I would not want to define specific scenario=
s where a flush must occur.<br>

</div></div></div></div></blockquote></div><div class=3D"im"><div><br></div=
><div><br></div></div><div>HTTP2&#39;s MSG_DONE (under a different name) is=
 in the next implementation draft, coming out shortly.</div><div>For HTTP o=
n HTTP/2 it is use for delimiting chunk boundaries on an HTTP stream, since=
 the lack of this has proven painful for some. Intermediaries are disallowe=
d from coalescing data across frame where this is set.</div>

</div></div></div></blockquote></div></div></div>

--047d7bfd0172d2518504f1cfd655--

From tyoshino@google.com  Fri Feb  7 04:11:26 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id E14A31A0605 for <hybi@ietfa.amsl.com>; Fri,  7 Feb 2014 04:11:25 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.913
X-Spam-Level: 
X-Spam-Status: No, score=-1.913 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.535, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id C3BOhKj1i7RO for <hybi@ietfa.amsl.com>; Fri,  7 Feb 2014 04:11:23 -0800 (PST)
Received: from mail-wi0-x235.google.com (mail-wi0-x235.google.com [IPv6:2a00:1450:400c:c05::235]) by ietfa.amsl.com (Postfix) with ESMTP id 8B8461A1DBC for <hybi@ietf.org>; Fri,  7 Feb 2014 04:11:23 -0800 (PST)
Received: by mail-wi0-f181.google.com with SMTP id hi5so766595wib.2 for <hybi@ietf.org>; Fri, 07 Feb 2014 04:11:23 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc:content-type; bh=Gu1XpdPAFZLnGhX7+pNTnvWmSRGQXhzHeA2w6IajTU4=; b=pFMFOJdLWrYJ/s+AclLRt5dyqqshwiaGPDgtawDwEuNlBXbsBZg9sLlrQnKSqcAEBe f/tbqmgsIUr8QALWbXBIYgCFOhAiN7ynQdoH/WkXUQI9id7mcZEiWM3+BZKGaCaURezD TTaHDvIeZmOlRpZ+6w3smbJ2IPVTShAS8rWBaOEzrsKsdZZHG3ZfR6JEzlR/ZOgnL675 6W9Di+VFpQrQDiMjMPURKu7TNV0h8oP7AvvCfFrger7Uekv/MIt15EdeHupT/KCuAKLV p2fOyjWA8e0tIl0BvfBHZIL9bKOR+ujzMvUmor/IM4XZp/qZi3ONs+UGuk2peuQ29SF1 KOAg==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc:content-type; bh=Gu1XpdPAFZLnGhX7+pNTnvWmSRGQXhzHeA2w6IajTU4=; b=V++4GMaXXvk2elKuKMCvVy0I4bluY3Xac0KmNRnq0L4PmGG2OaTg3kjJhTdJmlKFRR WuUIeT929AKykeQhjAZFd92FUix84TMBkMwdO54mzYbQn30jGgC9/Jok36iGnUsFsVS2 hvPCT0Q1SwWnyZNqg4Fj5yzoZVgF7bOMvjq7xotJNHWr2QK/ZX4CTaZN4VZapYIlHk9z d50H4OOuNlvNuDewe1RoYdIeT0h0mhhk6V+uUU+o51gTcKbLljDhCmUEIh3kUPyVfFOz zvwUiyI/EL/KUIQkTs8WZEcu19q5aTfxEZSd9pDxpNN4X67a9KUVTSuY5Gr6mv7m/dFB jIEQ==
X-Gm-Message-State: ALoCoQlXFMT3rOZSkHJVmLMZLEBCgqFZ/W5SxRwDuDnBjTer0JxgKQheDZJJOva147U37WOMNAE5RZqrU59uVI8+RS7N03iGn7JDZ5qcDgPNGBt/6MxpYNHWqXas33g4nkwsWxAdiXrmwTDFlilwNMum1hJI6wFnMfrRAtli+LWZhEVqrrJp1aWGamoi26vlTJb30Lvk6CWn
X-Received: by 10.180.37.193 with SMTP id a1mr3594781wik.52.1391775083048; Fri, 07 Feb 2014 04:11:23 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Fri, 7 Feb 2014 04:11:02 -0800 (PST)
In-Reply-To: <634914A010D0B943A035D226786325D4446C36C47C@EXVMBX020-12.exch020.serverdata.net>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <740FF040-F1E9-43E6-ACDE-0014FE4D012F@zaphoyd.com> <CABihn6HM_yXSB9RapNEQEG0_9OYfpHG3uH+=wnDwRJAFbh9EJA@mail.gmail.com> <0F978119-88F2-49CE-98C6-2A640527AA64@zaphoyd.com> <CAGzyod7BA_UXjN4r+Qr+P33QsPpeWWotxiScTpzqyhvbtPgNYQ@mail.gmail.com> <2B0B9332-D002-460A-8D50-B90B55999F42@zaphoyd.com> <CABihn6Hykczbu0YmUDuFcjVt2+xyeYzmdOpEkiZmP-2Vzuiuow@mail.gmail.com> <634914A010D0B943A035D226786325D4446C36C47C@EXVMBX020-12.exch020.serverdata.net>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Fri, 7 Feb 2014 21:11:02 +0900
Message-ID: <CAH9hSJaKKP-Mhf=oj-J663EPqQfHN=xktCpP0rWi2MqtfS3wKQ@mail.gmail.com>
To: Tobias Oberstein <tobias.oberstein@tavendo.de>
Content-Type: multipart/alternative; boundary=e89a8f502fa06ee48504f1cfe32d
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 07 Feb 2014 12:11:26 -0000

--e89a8f502fa06ee48504f1cfe32d
Content-Type: text/plain; charset=UTF-8

On Fri, Feb 7, 2014 at 7:16 PM, Tobias Oberstein <
tobias.oberstein@tavendo.de> wrote:

> >Questions
> >As you notice, there are two questions from me.
> >1: Are you OK with deciding NOW if we should preserve the RFC6455
> semantics or not?
> >2: (If the answer for 1 is yes) Should we preserve the RFC6455 semantics?
> >
> >My answer is, yes & yes.
>
> My answer: yes & yes.


Yes & Yes

Takeshi

--e89a8f502fa06ee48504f1cfe32d
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote">On F=
ri, Feb 7, 2014 at 7:16 PM, Tobias Oberstein <span dir=3D"ltr">&lt;<a href=
=3D"mailto:tobias.oberstein@tavendo.de" target=3D"_blank">tobias.oberstein@=
tavendo.de</a>&gt;</span> wrote:<br>

<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div class=3D"im">&gt;Questions<br>
&gt;As you notice, there are two questions from me.<br>
&gt;1: Are you OK with deciding NOW if we should preserve the RFC6455 seman=
tics or not?<br>
&gt;2: (If the answer for 1 is yes) Should we preserve the RFC6455 semantic=
s?<br>
&gt;<br>
&gt;My answer is, yes &amp; yes.<br>
<br>
</div>My answer: yes &amp; yes.=C2=A0</blockquote></div><br></div><div clas=
s=3D"gmail_extra">Yes &amp; Yes</div><div class=3D"gmail_extra"><br></div><=
div class=3D"gmail_extra">Takeshi</div></div>

--e89a8f502fa06ee48504f1cfe32d--

From joakim@intalio.com  Fri Feb  7 05:11:30 2014
Return-Path: <joakim@intalio.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id C3BA51A1EE8 for <hybi@ietfa.amsl.com>; Fri,  7 Feb 2014 05:11:30 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.977
X-Spam-Level: 
X-Spam-Status: No, score=-1.977 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_LOW=-0.7] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id r7Z01TbBr8g8 for <hybi@ietfa.amsl.com>; Fri,  7 Feb 2014 05:11:29 -0800 (PST)
Received: from mail-ee0-f52.google.com (mail-ee0-f52.google.com [74.125.83.52]) by ietfa.amsl.com (Postfix) with ESMTP id 0E8B61A1E10 for <hybi@ietf.org>; Fri,  7 Feb 2014 05:11:27 -0800 (PST)
Received: by mail-ee0-f52.google.com with SMTP id e53so1512062eek.25 for <hybi@ietf.org>; Fri, 07 Feb 2014 05:11:27 -0800 (PST)
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=usze2+bLIAn5ziOzjDqLPa6A7fexwqqTAF8+F9fr8lY=; b=iGSdl747cHFu+rkxrkH0cc7GyrcoBWTLIMyqea2qUPIOjHKrmPfFXPNo/lwlfsD1hH eixHbc3MD4L+u7ANJUjt/uaIzWha9/gsPCfi6dJ6Gb7WDsDFE24I5gMg57Z9Uh4XnSu5 b/2yeko+6y+DVE14tIxJJ9ebxf3v5FO9DFn6G6mi+qJe4tQeEFMjfayNBn4a2LtttrdR Oyz+CtJJRSglGojUth4w9Ie+uNRXNQU7f3fdl21zMZvFHfULR5mb957T6xAPTiGtVlLg /kT4AyTbBTKZbLkb/B32FIYWcU+IaZrIH9cAmSt9pxhx+WzP+kmeb5dtutSSTfroacMh 0f0w==
X-Gm-Message-State: ALoCoQk1/UjL5ONsA/uXU99KMkHiYZ+UEGMCPemUjQcyXIMrI71yalZxC0n0iynXZOoOT8QlKFpH
MIME-Version: 1.0
X-Received: by 10.14.106.193 with SMTP id m41mr4070110eeg.62.1391778687557; Fri, 07 Feb 2014 05:11:27 -0800 (PST)
Received: by 10.14.152.202 with HTTP; Fri, 7 Feb 2014 05:11:27 -0800 (PST)
In-Reply-To: <CAH9hSJaKKP-Mhf=oj-J663EPqQfHN=xktCpP0rWi2MqtfS3wKQ@mail.gmail.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <740FF040-F1E9-43E6-ACDE-0014FE4D012F@zaphoyd.com> <CABihn6HM_yXSB9RapNEQEG0_9OYfpHG3uH+=wnDwRJAFbh9EJA@mail.gmail.com> <0F978119-88F2-49CE-98C6-2A640527AA64@zaphoyd.com> <CAGzyod7BA_UXjN4r+Qr+P33QsPpeWWotxiScTpzqyhvbtPgNYQ@mail.gmail.com> <2B0B9332-D002-460A-8D50-B90B55999F42@zaphoyd.com> <CABihn6Hykczbu0YmUDuFcjVt2+xyeYzmdOpEkiZmP-2Vzuiuow@mail.gmail.com> <634914A010D0B943A035D226786325D4446C36C47C@EXVMBX020-12.exch020.serverdata.net> <CAH9hSJaKKP-Mhf=oj-J663EPqQfHN=xktCpP0rWi2MqtfS3wKQ@mail.gmail.com>
Date: Fri, 7 Feb 2014 06:11:27 -0700
Message-ID: <CAG4zZZAv+RiwA777Q=QUBZsLRnnHnt-hD-Z1iX_EUMBVuFTd7A@mail.gmail.com>
From: Joakim Erdfelt <joakim@intalio.com>
To: Takeshi Yoshino <tyoshino@google.com>
Content-Type: multipart/alternative; boundary=001a11c1b2dc4738a104f1d0bad1
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 07 Feb 2014 13:11:31 -0000

--001a11c1b2dc4738a104f1d0bad1
Content-Type: text/plain; charset=ISO-8859-1

On Fri, Feb 7, 2014 at 5:11 AM, Takeshi Yoshino <tyoshino@google.com> wrote:

> On Fri, Feb 7, 2014 at 7:16 PM, Tobias Oberstein <
> tobias.oberstein@tavendo.de> wrote:
>
>> >Questions
>> >As you notice, there are two questions from me.
>> >1: Are you OK with deciding NOW if we should preserve the RFC6455
>> semantics or not?
>> >2: (If the answer for 1 is yes) Should we preserve the RFC6455 semantics?
>> >
>> >My answer is, yes & yes.
>>
>> My answer: yes & yes.
>
>
>
Eclipse Jetty's answer is yes & yes.
We agree with Roberto and Tobias.
We throw our support to Plan D.
(Oddly, our reading of this thread made is seem that Roberto was in support
for Plan D too)

--001a11c1b2dc4738a104f1d0bad1
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote">On F=
ri, Feb 7, 2014 at 5:11 AM, Takeshi Yoshino <span dir=3D"ltr">&lt;<a href=
=3D"mailto:tyoshino@google.com" target=3D"_blank">tyoshino@google.com</a>&g=
t;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div dir=3D"ltr"><div class=3D""><div class=
=3D"gmail_extra"><div class=3D"gmail_quote">On Fri, Feb 7, 2014 at 7:16 PM,=
 Tobias Oberstein <span dir=3D"ltr">&lt;<a href=3D"mailto:tobias.oberstein@=
tavendo.de" target=3D"_blank">tobias.oberstein@tavendo.de</a>&gt;</span> wr=
ote:<br>


<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div>&gt;Questions<br>
&gt;As you notice, there are two questions from me.<br>
&gt;1: Are you OK with deciding NOW if we should preserve the RFC6455 seman=
tics or not?<br>
&gt;2: (If the answer for 1 is yes) Should we preserve the RFC6455 semantic=
s?<br>
&gt;<br>
&gt;My answer is, yes &amp; yes.<br>
<br>
</div>My answer: yes &amp; yes.=A0</blockquote></div></div></div><div class=
=3D"gmail_extra"><br></div></div></blockquote><div><br></div><div>Eclipse J=
etty&#39;s answer is yes &amp; yes.</div><div>We agree with Roberto and Tob=
ias.</div>
<div>We throw our support to Plan D.</div><div>(Oddly, our reading of this =
thread made is seem that Roberto was in support for Plan D too)=A0</div></d=
iv></div></div>

--001a11c1b2dc4738a104f1d0bad1--

From webmaster@zaphoyd.com  Fri Feb  7 05:56:47 2014
Return-Path: <webmaster@zaphoyd.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 90E741A1F02 for <hybi@ietfa.amsl.com>; Fri,  7 Feb 2014 05:56:47 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.899
X-Spam-Level: 
X-Spam-Status: No, score=-1.899 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HTML_MESSAGE=0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id IVlIuE-XCoky for <hybi@ietfa.amsl.com>; Fri,  7 Feb 2014 05:56:46 -0800 (PST)
Received: from authsmtp00.uchicago.edu (authsmtp00.uchicago.edu [128.135.12.120]) by ietfa.amsl.com (Postfix) with ESMTP id F1C2D1A1EFC for <hybi@ietf.org>; Fri,  7 Feb 2014 05:56:45 -0800 (PST)
Received: from [10.0.1.89] (c-68-51-76-178.hsd1.il.comcast.net [68.51.76.178]) (Authenticated sender: zaphoyd) by authsmtp00.uchicago.edu (Postfix) with ESMTP id 70DCD4980A9; Fri,  7 Feb 2014 07:56:45 -0600 (CST)
Content-Type: multipart/alternative; boundary="Apple-Mail=_306D689A-E7DA-4796-831B-5D66488912D5"
Mime-Version: 1.0 (Mac OS X Mail 7.0 \(1822\))
From: Peter Thorson <webmaster@zaphoyd.com>
In-Reply-To: <CABihn6Hykczbu0YmUDuFcjVt2+xyeYzmdOpEkiZmP-2Vzuiuow@mail.gmail.com>
Date: Fri, 7 Feb 2014 07:56:45 -0600
Message-Id: <49E2DC11-0C77-48E8-A044-8B67AD5C7B5D@zaphoyd.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <740FF 040-F1E9-43E6-ACDE-0014FE4D012F@zaphoyd.com> <CABihn6HM_yXSB9RapNEQEG0_9OYfpHG3uH+=wnDwRJAFbh9EJA@mail.gmail.com> <0F978119-88F2-49CE-98C6-2A640527AA64@zaphoyd.com> <CAGzyod7BA_UXjN4r+Qr+P33QsPpeWWotxiScTpzqyhvbtPgNYQ@mail.gmail.com> <2B0B9332-D002-460A-8D50-B90B55999F42@zaphoyd.com> <CABihn6Hykczbu0YmUDuFcjVt2+xyeYzmdOpEkiZmP-2Vzuiuow@mail.gmail.com>
To: Yutaka Hirano <yhirano@chromium.org>
X-Mailer: Apple Mail (2.1822)
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 07 Feb 2014 13:56:47 -0000

--Apple-Mail=_306D689A-E7DA-4796-831B-5D66488912D5
Content-Transfer-Encoding: quoted-printable
Content-Type: text/plain;
	charset=us-ascii


On Feb 7, 2014, at 1:05, Yutaka Hirano <yhirano@chromium.org> wrote:

> As the only supporter of Plan A, I promise that I will retract the =
plan if a critical problem is found on it (it is not compatible with =
RFC6455, MSG_DONE can't be used for the flushing purpose, FLUSH can't be =
introduced and so on).

With the understanding that under Plan A MSG_DONE is used as a flush =
signal and not a literal message end flag I would support Plan A as =
well. I have no particular use for a flush signal, but I would not =
object to it being present if others found it necessary.

> Questions
> As you notice, there are two questions from me.
> 1: Are you OK with deciding NOW if we should preserve the RFC6455 =
semantics or not?
> 2: (If the answer for 1 is yes) Should we preserve the RFC6455 =
semantics?

1. Yes
2. As HTTP/2.0 is constructed now with a strong focus on backwards =
compatibility for the HTTP side, Yes=20=

--Apple-Mail=_306D689A-E7DA-4796-831B-5D66488912D5
Content-Transfer-Encoding: quoted-printable
Content-Type: text/html;
	charset=us-ascii

<html><head><meta http-equiv=3D"Content-Type" content=3D"text/html =
charset=3Dus-ascii"></head><body style=3D"word-wrap: break-word; =
-webkit-nbsp-mode: space; -webkit-line-break: =
after-white-space;"><br><div><div>On Feb 7, 2014, at 1:05, Yutaka Hirano =
&lt;<a href=3D"mailto:yhirano@chromium.org">yhirano@chromium.org</a>&gt; =
wrote:</div><br class=3D"Apple-interchange-newline"><blockquote =
type=3D"cite"><div dir=3D"ltr"><div>As the only supporter of Plan A, I =
promise that I will retract the plan if a critical problem is found on =
it (it is not compatible with RFC6455, MSG_DONE can't be used for the =
flushing purpose, FLUSH can't be introduced and so =
on).</div></div></blockquote><div><br></div><div>With the understanding =
that under Plan A MSG_DONE is used as a flush signal and not a literal =
message end flag I would support Plan A as well. I have no particular =
use for a flush signal, but I would not object to it being present if =
others found it necessary.</div><br><blockquote type=3D"cite"><div =
dir=3D"ltr"><div><b>Questions</b></div><div>As you notice, there are two =
questions from me.</div>
<div>1: Are you OK with deciding NOW if we should preserve the RFC6455 =
semantics or not?</div>
<div>2: (If the answer for 1 is yes) Should we preserve the RFC6455 =
semantics?</div></div></blockquote><br></div><div>1. Yes</div>2. As =
HTTP/2.0 is constructed now with a strong focus on backwards =
compatibility for the HTTP side, Yes&nbsp;</body></html>=

--Apple-Mail=_306D689A-E7DA-4796-831B-5D66488912D5--

From salvatore.loreto@ericsson.com  Mon Feb 10 01:09:52 2014
Return-Path: <salvatore.loreto@ericsson.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 5F5721A04C5 for <hybi@ietfa.amsl.com>; Mon, 10 Feb 2014 01:09:52 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -3.85
X-Spam-Level: 
X-Spam-Status: No, score=-3.85 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HELO_EQ_SE=0.35, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_MED=-2.3, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id MVIzuqyFh4hO for <hybi@ietfa.amsl.com>; Mon, 10 Feb 2014 01:09:50 -0800 (PST)
Received: from mailgw1.ericsson.se (mailgw1.ericsson.se [193.180.251.45]) by ietfa.amsl.com (Postfix) with ESMTP id 8F51D1A06A9 for <hybi@ietf.org>; Mon, 10 Feb 2014 01:09:49 -0800 (PST)
X-AuditID: c1b4fb2d-b7f5d8e000002a7b-e9-52f8975c5ba3
Received: from ESESSHC009.ericsson.se (Unknown_Domain [153.88.253.124]) by mailgw1.ericsson.se (Symantec Mail Security) with SMTP id 29.20.10875.C5798F25; Mon, 10 Feb 2014 10:09:48 +0100 (CET)
Received: from ESESSMB109.ericsson.se ([169.254.9.236]) by ESESSHC009.ericsson.se ([153.88.183.45]) with mapi id 14.02.0387.000; Mon, 10 Feb 2014 10:09:47 +0100
From: Salvatore Loreto <salvatore.loreto@ericsson.com>
To: Yutaka Hirano <yhirano@chromium.org>
Thread-Topic: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
Thread-Index: AQHPIQRX/FgLHM2uwUe2xaUgwKHEnpqlFvGAgAAtp4CAASpYAIAAURwAgADvowCAAJ0VgIAAQlYAgABBhwCAAIKagIAE2amA
Date: Mon, 10 Feb 2014 09:09:47 +0000
Message-ID: <4A83D478-81A7-4B3F-BFF1-C88F6358F332@ericsson.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <740FF040-F1E9-43E6-ACDE-0014FE4D012F@zaphoyd.com> <CABihn6HM_yXSB9RapNEQEG0_9OYfpHG3uH+=wnDwRJAFbh9EJA@mail.gmail.com> <0F978119-88F2-49CE-98C6-2A640527AA64@zaphoyd.com> <CAGzyod7BA_UXjN4r+Qr+P33QsPpeWWotxiScTpzqyhvbtPgNYQ@mail.gmail.com> <2B0B9332-D002-460A-8D50-B90B55999F42@zaphoyd.com> <CABihn6Hykczbu0YmUDuFcjVt2+xyeYzmdOpEkiZmP-2Vzuiuow@mail.gmail.com>
In-Reply-To: <CABihn6Hykczbu0YmUDuFcjVt2+xyeYzmdOpEkiZmP-2Vzuiuow@mail.gmail.com>
Accept-Language: en-US
Content-Language: en-US
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
x-originating-ip: [153.88.183.150]
Content-Type: multipart/alternative; boundary="_000_4A83D47881A74B3FBFF1C88F6358F332ericssoncom_"
MIME-Version: 1.0
X-Brightmail-Tracker: H4sIAAAAAAAAA+NgFvrFLMWRmVeSWpSXmKPExsUyM+JvjW7M9B9BBncuqli8f7mNyeL90x1s Fpdv/mZ1YPaY3XCRxWPJkp9MHs8b21kDmKO4bFJSczLLUov07RK4Mr48WchU8EKu4tTJ7ywN jC+kuhg5OSQETCQeL33GCGGLSVy4t56ti5GLQ0jgEKPE12snoJwljBJXpk9mBqliEzCTeP5w C5DNwSEioCHRcC0WJMws4CWxfvlbVhBbWCBYYvqhFqiSEIk7d1RBwiICeRJTDt9hA7FZBFQl 5h3Zxg5i8wrYS9zvb2SHWPWUS6LpYw/YHE6BQIlfK/rBbEag476fWsMEsUtc4taT+UwQRwtI LNlznhnCFpV4+fgfK4StJLFi+yVGiPpkiTUd85khlglKnJz5hGUCo+gsJKNmISmbhaQMIq4j sWD3JzYIW1ti2cLXzDD2mQOPoXqtJU6sfsWIrGYBI8cqRvbcxMyc9HLDTYzA+Du45bfuDsZT 50QOMUpzsCiJ83546xwkJJCeWJKanZpakFoUX1Sak1p8iJGJg1OqgVH41Zpbos+esfz3ORnG 2uN6burC6quqLl9ZnO2EzutbG2Zqr3i272q36O6SG9rPvUrdkpNb22UnvcvneCd7f1Pus8Zo 1jiz3L7dhQ0Lmjb0d8/U5zowabHjs1d2t+Vfr5zLzs5osuTMjV6+9/u/nj3/wrcrtk2rv2RT TNdGiW5PFaM17CGbgpRYijMSDbWYi4oTAQrZblSNAgAA
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of	WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 10 Feb 2014 09:09:52 -0000

--_000_4A83D47881A74B3FBFF1C88F6358F332ericssoncom_
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable


On Feb 7, 2014, at 9:05 AM, Yutaka Hirano <yhirano@chromium.org<mailto:yhir=
ano@chromium.org>> wrote:

Questions
As you notice, there are two questions from me.
1: Are you OK with deciding NOW if we should preserve the RFC6455 semantics=
 or not?
2: (If the answer for 1 is yes) Should we preserve the RFC6455 semantics?

My answer is, yes & yes.

yes & yes is also my answer

/Salvatore


--_000_4A83D47881A74B3FBFF1C88F6358F332ericssoncom_
Content-Type: text/html; charset="us-ascii"
Content-ID: <637AF1AA6DA6F941834F1C95D2287117@ericsson.com>
Content-Transfer-Encoding: quoted-printable

<html>
<head>
<meta http-equiv=3D"Content-Type" content=3D"text/html; charset=3Dus-ascii"=
>
</head>
<body style=3D"word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-lin=
e-break: after-white-space; ">
<br>
<div>
<div>On Feb 7, 2014, at 9:05 AM, Yutaka Hirano &lt;<a href=3D"mailto:yhiran=
o@chromium.org">yhirano@chromium.org</a>&gt; wrote:</div>
<br class=3D"Apple-interchange-newline">
<blockquote type=3D"cite">
<div style=3D"font-family: Helvetica; font-size: medium; font-style: normal=
; font-variant: normal; font-weight: normal; letter-spacing: normal; line-h=
eight: normal; orphans: 2; text-align: -webkit-auto; text-indent: 0px; text=
-transform: none; white-space: normal; widows: 2; word-spacing: 0px; -webki=
t-text-size-adjust: auto; -webkit-text-stroke-width: 0px; ">
<b>Questions</b></div>
<div style=3D"font-family: Helvetica; font-size: medium; font-style: normal=
; font-variant: normal; font-weight: normal; letter-spacing: normal; line-h=
eight: normal; orphans: 2; text-align: -webkit-auto; text-indent: 0px; text=
-transform: none; white-space: normal; widows: 2; word-spacing: 0px; -webki=
t-text-size-adjust: auto; -webkit-text-stroke-width: 0px; ">
As you notice, there are two questions from me.</div>
<div style=3D"font-family: Helvetica; font-size: medium; font-style: normal=
; font-variant: normal; font-weight: normal; letter-spacing: normal; line-h=
eight: normal; orphans: 2; text-align: -webkit-auto; text-indent: 0px; text=
-transform: none; white-space: normal; widows: 2; word-spacing: 0px; -webki=
t-text-size-adjust: auto; -webkit-text-stroke-width: 0px; ">
1: Are you OK with deciding NOW if we should preserve the RFC6455 semantics=
 or not?</div>
<div style=3D"font-family: Helvetica; font-size: medium; font-style: normal=
; font-variant: normal; font-weight: normal; letter-spacing: normal; line-h=
eight: normal; orphans: 2; text-align: -webkit-auto; text-indent: 0px; text=
-transform: none; white-space: normal; widows: 2; word-spacing: 0px; -webki=
t-text-size-adjust: auto; -webkit-text-stroke-width: 0px; ">
2: (If the answer for 1 is yes) Should we preserve the RFC6455 semantics?</=
div>
<div style=3D"font-family: Helvetica; font-size: medium; font-style: normal=
; font-variant: normal; font-weight: normal; letter-spacing: normal; line-h=
eight: normal; orphans: 2; text-align: -webkit-auto; text-indent: 0px; text=
-transform: none; white-space: normal; widows: 2; word-spacing: 0px; -webki=
t-text-size-adjust: auto; -webkit-text-stroke-width: 0px; ">
<br>
</div>
<div style=3D"font-family: Helvetica; font-size: medium; font-style: normal=
; font-variant: normal; font-weight: normal; letter-spacing: normal; line-h=
eight: normal; orphans: 2; text-align: -webkit-auto; text-indent: 0px; text=
-transform: none; white-space: normal; widows: 2; word-spacing: 0px; -webki=
t-text-size-adjust: auto; -webkit-text-stroke-width: 0px; ">
My answer is, yes &amp; yes.</div>
</blockquote>
<br>
</div>
<div>yes &amp; yes is also my answer</div>
<div><br>
</div>
<div>/Salvatore</div>
<br>
</body>
</html>

--_000_4A83D47881A74B3FBFF1C88F6358F332ericssoncom_--

From salvatore.loreto@ericsson.com  Mon Feb 10 01:18:03 2014
Return-Path: <salvatore.loreto@ericsson.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 3D8EE1A05B8 for <hybi@ietfa.amsl.com>; Mon, 10 Feb 2014 01:18:03 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.239
X-Spam-Level: 
X-Spam-Status: No, score=-1.239 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HELO_EQ_SE=0.35, HOST_MISMATCH_NET=0.311, HTML_MESSAGE=0.001, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id pe8OgPB-wh51 for <hybi@ietfa.amsl.com>; Mon, 10 Feb 2014 01:18:01 -0800 (PST)
Received: from sessmg20.mgmt.ericsson.se (sessmg20.ericsson.net [193.180.251.50]) by ietfa.amsl.com (Postfix) with ESMTP id 0C0501A0502 for <hybi@ietf.org>; Mon, 10 Feb 2014 01:18:00 -0800 (PST)
X-AuditID: c1b4fb32-b7f4c8e0000012f5-5b-52f899488170
Received: from ESESSHC008.ericsson.se (Unknown_Domain [153.88.253.124]) by sessmg20.mgmt.ericsson.se (Symantec Mail Security) with SMTP id 44.90.04853.84998F25; Mon, 10 Feb 2014 10:18:00 +0100 (CET)
Received: from ESESSMB109.ericsson.se ([169.254.9.236]) by ESESSHC008.ericsson.se ([153.88.183.42]) with mapi id 14.02.0387.000; Mon, 10 Feb 2014 10:18:00 +0100
From: Salvatore Loreto <salvatore.loreto@ericsson.com>
To: Peter Thorson <webmaster@zaphoyd.com>
Thread-Topic: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
Thread-Index: AQHPFggAq4DVnVNfkkOHN4dpGLr0upqSHDAAgAN6vICAGK46AA==
Date: Mon, 10 Feb 2014 09:17:58 +0000
Message-ID: <8A6D73B0-6339-46A7-B3AB-F02880C81CD7@ericsson.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <634914A010D0B943A035D226786325D4446C0826EC@EXVMBX020-12. exch020.serverdata.net> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <CABihn6GSrU8oJa4nbMJiDDo2Z+-c4fT69roVFxcHYM4XYFToRQ@mail.gmail.com> <F1F115D4-29EB-457B-B02C-71E46D3E1C81@zaphoyd.com>
In-Reply-To: <F1F115D4-29EB-457B-B02C-71E46D3E1C81@zaphoyd.com>
Accept-Language: en-US
Content-Language: en-US
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
x-originating-ip: [153.88.183.149]
Content-Type: multipart/alternative; boundary="_000_8A6D73B0633946A7B3ABF02880C81CD7ericssoncom_"
MIME-Version: 1.0
X-Brightmail-Tracker: H4sIAAAAAAAAA+NgFupmkeLIzCtJLcpLzFFi42KZGfG3Rtdj5o8gg317FSzev9zGZPH+6Q42 i8s3f7M6MHvMbrjI4rFkyU8mj+eN7awBzFFcNimpOZllqUX6dglcGYuXn2IvuCFf8fD/GvYG xllSXYycHBICJhJr715ghrDFJC7cW8/WxcjFISRwglHi1v/5zBDOEkaJmR872UCq2ATMJJ4/ 3AKU4OAQEdCUWPlYBcRkFvCUuPDFCaRCWCBYYvqhFqiKEIk7d1RBwiICThJdc/eCrWIRUJVY 82QR2EBeAXuJ3+1LWSE2HeSUWDKlgQkkwQmUmPhkOjuIzQh02/dTa8DizALiEreezGeCuFlA Ysme81D3i0q8fPyPFcJWklh7eDsLRH2yxModi5ghlglKnJz5hGUCo+gsJKNmISmbhaQMIm4g 8f7cfGYIW1ti2cLXULa+xMYvZxkhbGuJPU82MiGrWcDIsYpRsji1uDg33chALzc9t0QvtSgz ubg4P0+vOHUTIzBCD275bbSD8eQe+0OM0hwsSuK811lrgoQE0hNLUrNTUwtSi+KLSnNSiw8x MnFwSjUwmr/p/ibIMtPEa+a95oLv85UX923YOYP7lmz5PL538fJJ6mc+SlTvSDpyV34f9wnl nfOMJfRjFwVxeQYZpxuYd8x3Cymc8IDTM3Pu+iP2MuWcLpxez2SUp107KKL7eZOw0PWjwrnH l0c/TMo9s8Ztv2eG7jkVP5XZeS3pr3YduMSwePWhzR4KSizFGYmGWsxFxYkApest4J4CAAA=
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of	WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 10 Feb 2014 09:18:03 -0000

--_000_8A6D73B0633946A7B3ABF02880C81CD7ericssoncom_
Content-Type: text/plain; charset="Windows-1252"
Content-Transfer-Encoding: quoted-printable

as one of HyBi wg chair, but without having been able to coordinate with my=
 co-chair yet
I do think that we should do this in cooperation with HTTP 2.0
and maybe it would be even better move the work directly in the httpbis wg
as from a HyBi chair prospective I see at moment much more energy in httpbi=
s wg than
in the hybi wg, and it would be great to leverage on that energy to also sp=
eed up this piece
of work

the best way I would suggest is to submit a proposal to the httpbis wg in f=
orm of a draft
(consider that the deadline to submit a draft is Friday 14th, so you have 5=
 days, if you
want have a chance for the proposal to be discussed to the next IETF meetin=
g in London
that will start March 2nd).

I would really encourage you do submit a draft, it is OK if it contains alt=
ernative options,
and so take the opportunity to discuss it in London

if you need any help in editing the draft I can also offer my help

br
Salvatore

On Jan 25, 2014, at 6:24 PM, Peter Thorson <webmaster@zaphoyd.com<mailto:we=
bmaster@zaphoyd.com>> wrote:

Critical first question here: Do we have the cooperation of the HTTP 2.0 wo=
rking group or is the WS over HTTP 2.0 effort something being done unilater=
ally by the WebSocket working group?


--_000_8A6D73B0633946A7B3ABF02880C81CD7ericssoncom_
Content-Type: text/html; charset="Windows-1252"
Content-ID: <BB8DD89DAE1C594C8507E35FD8A0433D@ericsson.com>
Content-Transfer-Encoding: quoted-printable

<html>
<head>
<meta http-equiv=3D"Content-Type" content=3D"text/html; charset=3DWindows-1=
252">
</head>
<body style=3D"word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-lin=
e-break: after-white-space; ">
as one of HyBi wg chair, but without having been able to coordinate with my=
 co-chair yet
<div>I do think that we should do this in cooperation with HTTP 2.0</div>
<div>and maybe it would be even better move the work directly in the httpbi=
s wg</div>
<div>
<div>as from a HyBi chair prospective I see at moment much more energy in h=
ttpbis wg than</div>
<div>in the hybi wg, and it would be great to leverage on that energy to al=
so speed up this piece</div>
</div>
<div>of work</div>
<div><br>
</div>
<div>the best way I would suggest is to submit a proposal to the httpbis wg=
 in form of a draft</div>
<div>(consider that the deadline to submit a draft is Friday 14th, so you h=
ave 5 days, if you</div>
<div>want have a chance for the proposal to be discussed to the next IETF m=
eeting in London</div>
<div>that will start March 2nd).</div>
<div><br>
</div>
<div>I would really encourage you do submit a draft, it is OK if it contain=
s alternative options,</div>
<div>and so take the opportunity to discuss it in London</div>
<div><br>
</div>
<div>if you need any help in editing the draft I can also offer my help</di=
v>
<div><br>
</div>
<div>br</div>
<div>Salvatore</div>
<div>
<div><br>
<div>
<div>On Jan 25, 2014, at 6:24 PM, Peter Thorson &lt;<a href=3D"mailto:webma=
ster@zaphoyd.com">webmaster@zaphoyd.com</a>&gt; wrote:</div>
<br class=3D"Apple-interchange-newline">
<blockquote type=3D"cite"><span style=3D"font-family: Helvetica; font-size:=
 medium; font-style: normal; font-variant: normal; font-weight: normal; let=
ter-spacing: normal; line-height: normal; orphans: 2; text-align: -webkit-a=
uto; text-indent: 0px; text-transform: none; white-space: normal; widows: 2=
; word-spacing: 0px; -webkit-text-size-adjust: auto; -webkit-text-stroke-wi=
dth: 0px; display: inline !important; float: none; ">Critical
 first question here: Do we have the cooperation of the HTTP 2.0 working gr=
oup or is the WS over HTTP 2.0 effort something being done unilaterally by =
the WebSocket working group?</span></blockquote>
</div>
<br>
</div>
</div>
</body>
</html>

--_000_8A6D73B0633946A7B3ABF02880C81CD7ericssoncom_--

From yhirano@google.com  Mon Feb 10 23:25:31 2014
Return-Path: <yhirano@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id D038F1A07B1 for <hybi@ietfa.amsl.com>; Mon, 10 Feb 2014 23:25:31 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.926
X-Spam-Level: 
X-Spam-Status: No, score=-1.926 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id MtL7hpoWLJdm for <hybi@ietfa.amsl.com>; Mon, 10 Feb 2014 23:25:29 -0800 (PST)
Received: from mail-yh0-x22c.google.com (mail-yh0-x22c.google.com [IPv6:2607:f8b0:4002:c01::22c]) by ietfa.amsl.com (Postfix) with ESMTP id 727431A0772 for <hybi@ietf.org>; Mon, 10 Feb 2014 23:25:29 -0800 (PST)
Received: by mail-yh0-f44.google.com with SMTP id f73so6333154yha.17 for <hybi@ietf.org>; Mon, 10 Feb 2014 23:25:29 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=43KKwxrUTw8D7A05bHcS8iMnjZEiIB1TQ9ijNvuG8cg=; b=JGRg9IfUKN/yN9ncStJYplIieWYfI+VcXnj1YojG2MaQ6iNpqtW2U0DXnAJfN5uz9Z bzAE7NSslctURA4tUz4Na4coCshkwcBdJr4awpMSA2HlpAiP9p5ikr20s768l2Z2OF6k A1ZIsDDwhXXaH5wLZVn3t2B/Vzi9qTbUO6NwAUT8y5dwIRgTngRenN52GEfjHNtwTkUx G+Asqj3daWhY2zS1YA/qV1U5w9GWfagC7/hxZivEquKhr2cYXmPv1Idb036wDuWbKKjd QBzKIgy4GjEp+7r7BScujZQgwN88drVk9k2HXRIqpNlEcLz7txTEe/nPasx3SYo9nutN /A+Q==
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=chromium.org; s=google; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=43KKwxrUTw8D7A05bHcS8iMnjZEiIB1TQ9ijNvuG8cg=; b=mjbZSDPmgM+7yhrnEVqzY1ZxmfT72xKIFHV4+kxJ0T+9W9Puc9DpH4vNpO9NbJUdDz GFpZb0ES8u7NF5o74y+sd5j0kClUt/ElxbLjk1E0ij/Z913KUA1LGW2xfVZ5tc7+N/PD Gzp9KS6q1xS5Jp5TZcNhBE6v3k7QqqbX+Jf78=
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:sender:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=43KKwxrUTw8D7A05bHcS8iMnjZEiIB1TQ9ijNvuG8cg=; b=mbEZxkK2yiXLkBn06qUBmfSonrY3NjWheLrRW47S/De1IQ2HUvQj7lXtAunXGENXck 7c2Bt5w7LMMW59vBghphcgoaDx0JnfV+TzHcZFMGWw8K2XOX3H4GB9jIgc8dnDw9Gf5n VFFoqZnp7RgqjLKY8W1+O5b1Ek3cr7JhxijqOnQIAh//cX8LU/LKDqX45T+PVlV7SJP9 BXF9i4nqiFp3rohCe4ZFtFn1Fm+jOXhY//5FpzTEz1pUwhBWPv/F9FKa185tTIq1yroP KPLIvwDjjSXQUmKBXk9LthskK+F5gCiLyWnl4zhUQ/V+86JbTIJX21Cfyd0MWxddmA39 rKfQ==
X-Gm-Message-State: ALoCoQktlNN2TR3OnhQo6laKuGtnevawLJ52qSBGGxU9+FOgEGXb3n9R8tINJVeHwXKfYcT2CTyWWliw4JeKeBYq6ghkXTLnvantJ5dongACu6HdH6HGlXYxXpEV/Ogm8/TwOrPs1mO7uLZpmCS1PDqZviRdi+0eQjzY5ZQCcUz2plSKRAK3bMvt1plpXU0xsa4OTWezzZUL
MIME-Version: 1.0
X-Received: by 10.236.15.102 with SMTP id e66mr697715yhe.69.1392103528927; Mon, 10 Feb 2014 23:25:28 -0800 (PST)
Sender: yhirano@google.com
Received: by 10.170.151.137 with HTTP; Mon, 10 Feb 2014 23:25:28 -0800 (PST)
In-Reply-To: <8A6D73B0-6339-46A7-B3AB-F02880C81CD7@ericsson.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <CABihn6GSrU8oJa4nbMJiDDo2Z+-c4fT69roVFxcHYM4XYFToRQ@mail.gmail.com> <F1F115D4-29EB-457B-B02C-71E46D3E1C81@zaphoyd.com> <8A6D73B0-6339-46A7-B3AB-F02880C81CD7@ericsson.com>
Date: Tue, 11 Feb 2014 16:25:28 +0900
X-Google-Sender-Auth: IC-Z5tNBsOx0f3pr1bAVq_sIhcM
Message-ID: <CABihn6EiVmCx50MPv5=crHymez4bbn8kieto-je5niVK1w8iBA@mail.gmail.com>
From: Yutaka Hirano <yhirano@chromium.org>
To: Salvatore Loreto <salvatore.loreto@ericsson.com>
Content-Type: multipart/alternative; boundary=089e0122a3d455301b04f21c5cb6
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 11 Feb 2014 07:25:32 -0000

--089e0122a3d455301b04f21c5cb6
Content-Type: text/plain; charset=UTF-8

Hi Salvatore,
Thank you for the suggestions.

as one of HyBi wg chair, but without having been able to coordinate with my
> co-chair yet
> I do think that we should do this in cooperation with HTTP 2.0
> and maybe it would be even better move the work directly in the httpbis wg
> as from a HyBi chair prospective I see at moment much more energy in
> httpbis wg than
> in the hybi wg, and it would be great to leverage on that energy to also
> speed up this piece
> of work

I am waiting for Roberto's response. If his response is positive, I will
update the draft and it is OK with me to move this topic to httpbis.

For the RFC draft, I am trying. Though it is unclear if I can make it in
time (I have never written one), I will do my best.

Thanks,



On Mon, Feb 10, 2014 at 6:17 PM, Salvatore Loreto <
salvatore.loreto@ericsson.com> wrote:

>  as one of HyBi wg chair, but without having been able to coordinate with
> my co-chair yet
> I do think that we should do this in cooperation with HTTP 2.0
> and maybe it would be even better move the work directly in the httpbis wg
>  as from a HyBi chair prospective I see at moment much more energy in
> httpbis wg than
> in the hybi wg, and it would be great to leverage on that energy to also
> speed up this piece
>  of work
>
>  the best way I would suggest is to submit a proposal to the httpbis wg
> in form of a draft
> (consider that the deadline to submit a draft is Friday 14th, so you have
> 5 days, if you
> want have a chance for the proposal to be discussed to the next IETF
> meeting in London
> that will start March 2nd).
>
>  I would really encourage you do submit a draft, it is OK if it contains
> alternative options,
> and so take the opportunity to discuss it in London
>
>  if you need any help in editing the draft I can also offer my help
>
>  br
> Salvatore
>
>  On Jan 25, 2014, at 6:24 PM, Peter Thorson <webmaster@zaphoyd.com> wrote:
>
> Critical first question here: Do we have the cooperation of the HTTP 2.0
> working group or is the WS over HTTP 2.0 effort something being done
> unilaterally by the WebSocket working group?
>
>
>

--089e0122a3d455301b04f21c5cb6
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Hi Salvatore,<div>Thank you for the suggestions.</div><div=
><br></div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.=
8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-st=
yle:solid;padding-left:1ex">
<span style=3D"font-family:arial,sans-serif;font-size:13px">as one of HyBi =
wg chair, but without having been able to coordinate with my co-chair yet<b=
r></span>I do think that we should do this in cooperation with HTTP 2.0<br>
and maybe it would be even better move the work directly in the httpbis wg<=
br>as from a HyBi chair prospective I see at moment much more energy in htt=
pbis wg than<br>in the hybi wg, and it would be great to leverage on that e=
nergy to also speed up this piece<br>
of work</blockquote><div>I am waiting for Roberto&#39;s response. If his re=
sponse is positive, I will update the draft and it is OK with me to move th=
is topic to httpbis.</div><div><br></div><div>For the RFC draft, I am tryin=
g. Though it is unclear if I can make it in time (I have never written one)=
, I will do my best.</div>
<div><br></div><div>Thanks,</div><div><br></div></div><div class=3D"gmail_e=
xtra"><br><br><div class=3D"gmail_quote">On Mon, Feb 10, 2014 at 6:17 PM, S=
alvatore Loreto <span dir=3D"ltr">&lt;<a href=3D"mailto:salvatore.loreto@er=
icsson.com" target=3D"_blank">salvatore.loreto@ericsson.com</a>&gt;</span> =
wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">



<div style=3D"word-wrap:break-word">
as one of HyBi wg chair, but without having been able to coordinate with my=
 co-chair yet
<div>I do think that we should do this in cooperation with HTTP 2.0</div>
<div>and maybe it would be even better move the work directly in the httpbi=
s wg</div>
<div>
<div>as from a HyBi chair prospective I see at moment much more energy in h=
ttpbis wg than</div>
<div>in the hybi wg, and it would be great to leverage on that energy to al=
so speed up this piece</div>
</div>
<div>of work</div>
<div><br>
</div>
<div>the best way I would suggest is to submit a proposal to the httpbis wg=
 in form of a draft</div>
<div>(consider that the deadline to submit a draft is Friday 14th, so you h=
ave 5 days, if you</div>
<div>want have a chance for the proposal to be discussed to the next IETF m=
eeting in London</div>
<div>that will start March 2nd).</div>
<div><br>
</div>
<div>I would really encourage you do submit a draft, it is OK if it contain=
s alternative options,</div>
<div>and so take the opportunity to discuss it in London</div>
<div><br>
</div>
<div>if you need any help in editing the draft I can also offer my help</di=
v>
<div><br>
</div>
<div>br</div><span class=3D"HOEnZb"><font color=3D"#888888">
<div>Salvatore</div></font></span><div class=3D"">
<div>
<div><br>
<div>
<div>On Jan 25, 2014, at 6:24 PM, Peter Thorson &lt;<a href=3D"mailto:webma=
ster@zaphoyd.com" target=3D"_blank">webmaster@zaphoyd.com</a>&gt; wrote:</d=
iv>
<br>
<blockquote type=3D"cite"><span style=3D"font-family:Helvetica;font-size:me=
dium;font-style:normal;font-variant:normal;font-weight:normal;letter-spacin=
g:normal;line-height:normal;text-align:-webkit-auto;text-indent:0px;text-tr=
ansform:none;white-space:normal;word-spacing:0px;display:inline!important;f=
loat:none">Critical
 first question here: Do we have the cooperation of the HTTP 2.0 working gr=
oup or is the WS over HTTP 2.0 effort something being done unilaterally by =
the WebSocket working group?</span></blockquote>
</div>
<br>
</div>
</div>
</div></div>

</blockquote></div><br></div>

--089e0122a3d455301b04f21c5cb6--


From salvatore.loreto@ericsson.com  Mon Feb 10 23:30:00 2014
Return-Path: <salvatore.loreto@ericsson.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 77D2D1A07A2 for <hybi@ietfa.amsl.com>; Mon, 10 Feb 2014 23:30:00 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -3.85
X-Spam-Level: 
X-Spam-Status: No, score=-3.85 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HELO_EQ_SE=0.35, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_MED=-2.3, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id tGHrVCnLb0Eh for <hybi@ietfa.amsl.com>; Mon, 10 Feb 2014 23:29:58 -0800 (PST)
Received: from mailgw1.ericsson.se (mailgw1.ericsson.se [193.180.251.45]) by ietfa.amsl.com (Postfix) with ESMTP id 5EF841A0772 for <hybi@ietf.org>; Mon, 10 Feb 2014 23:29:57 -0800 (PST)
X-AuditID: c1b4fb2d-b7f5d8e000002a7b-e5-52f9d173b029
Received: from ESESSHC008.ericsson.se (Unknown_Domain [153.88.253.124]) by mailgw1.ericsson.se (Symantec Mail Security) with SMTP id 26.BB.10875.371D9F25; Tue, 11 Feb 2014 08:29:56 +0100 (CET)
Received: from ESESSMB109.ericsson.se ([169.254.9.236]) by ESESSHC008.ericsson.se ([153.88.183.42]) with mapi id 14.02.0387.000; Tue, 11 Feb 2014 08:29:56 +0100
From: Salvatore Loreto <salvatore.loreto@ericsson.com>
To: Yutaka Hirano <yhirano@chromium.org>
Thread-Topic: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
Thread-Index: AQHPJvpxJpclEbbv1UGaKICk/2DpGZqvl2iA
Date: Tue, 11 Feb 2014 07:29:54 +0000
Message-ID: <454347D9-8E92-496B-A1F2-6CA963DC9845@ericsson.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <CABihn6GSrU8oJa4nbMJiDDo2Z+-c4fT69roVFxcHYM4XYFToRQ@mail.gmail.com> <F1F115D4-29EB-457B-B02C-71E46D3E1C81@zaphoyd.com> <8A6D73B0-6339-46A7-B3AB-F02880C81CD7@ericsson.com> <CABihn6EiVmCx50MPv5=crHymez4bbn8kieto-je5niVK1w8iBA@mail.gmail.com>
In-Reply-To: <CABihn6EiVmCx50MPv5=crHymez4bbn8kieto-je5niVK1w8iBA@mail.gmail.com>
Accept-Language: en-US
Content-Language: en-US
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
x-originating-ip: [153.88.183.148]
Content-Type: multipart/alternative; boundary="_000_454347D98E92496BA1F26CA963DC9845ericssoncom_"
MIME-Version: 1.0
X-Brightmail-Tracker: H4sIAAAAAAAAA+NgFmpikeLIzCtJLcpLzFFi42KZGfG3Rrfk4s8gg2V7+Sz+9xhYvH+5jcni /dMdbBaXb/5mdWDxmN1wkcVjwaZSjyVLfjJ5PG9sZw1gieKySUnNySxLLdK3S+DKmHroPkvB SfuKlYfamRoYN5p0MXJySAiYSHQ1r2eEsMUkLtxbz9bFyMUhJHCIUWJJ5zw2kISQwBIgp0cb xGYTMJN4/nALcxcjB4eIgIZEw7VYkDCzQK7Eml+PWEBsYYFgiR3znoLZIgIhEo/3z2SGsI0k Vl3YDraLRUBV4tnSuWDjeQXsJebO6WSB2PuKU2LNlw6wBKdAoMTaq0vBGhiBjvt+ag0TxDJx iVtP5jNBHC0gsWTPeWYIW1Ti5eN/rBC2kkTjkiesEPXJEkcPL2SCWCYocXLmE5YJjKKzkIya haRsFpIyiLiOxILdn9ggbG2JZQtfM8PYZw48huq1lth09hqKmgWMHKsY2XMTM3PSyw03MQLj 8eCW37o7GE+dEznEKM3BoiTO++Gtc5CQQHpiSWp2ampBalF8UWlOavEhRiYOTqkGRs+NVwvj ln510mI5fa3zrx3HZbsnnF8fLH7qe9Wi8NY1FVevmQc2KERPePmDTYHpcs48vrm6x8+EHzVz fh0UUHqwdLKvx4G3vz9Or8mLjNpy3KV2p+Z5G6l9QRWhNnUxzRV3vZcGGpow/WAO5Mzvj377 0yD6s/jjJa2vf6X+LDdYebJEIuukiBJLcUaioRZzUXEiACF04z2VAgAA
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 11 Feb 2014 07:30:00 -0000

--_000_454347D98E92496BA1F26CA963DC9845ericssoncom_
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable


On Feb 11, 2014, at 9:25 AM, Yutaka Hirano <yhirano@chromium.org<mailto:yhi=
rano@chromium.org>>
 wrote:

Hi Salvatore,
Thank you for the suggestions.

as one of HyBi wg chair, but without having been able to coordinate with my=
 co-chair yet
I do think that we should do this in cooperation with HTTP 2.0
and maybe it would be even better move the work directly in the httpbis wg
as from a HyBi chair prospective I see at moment much more energy in httpbi=
s wg than
in the hybi wg, and it would be great to leverage on that energy to also sp=
eed up this piece
of work
I am waiting for Roberto's response. If his response is positive, I will up=
date the draft and it is OK with me to move this topic to httpbis.

note that I am only suggesting to submit to httpbis to check if the httpbis=
 is willing (and it has the energy) to work on it or not.
In order to discover it, the best way is to submit a draft (with the severa=
l options) to httpbis and have a discussion there
in the httpbis mailing list and preferably also during the f2f meeting in L=
ondon


For the RFC draft, I am trying. Though it is unclear if I can make it in ti=
me (I have never written one), I will do my best.

if you need any help in put the text in an draft format I can help without =
any problem.

br
Salvatore


Thanks,



On Mon, Feb 10, 2014 at 6:17 PM, Salvatore Loreto <salvatore.loreto@ericsso=
n.com<mailto:salvatore.loreto@ericsson.com>> wrote:
as one of HyBi wg chair, but without having been able to coordinate with my=
 co-chair yet
I do think that we should do this in cooperation with HTTP 2.0
and maybe it would be even better move the work directly in the httpbis wg
as from a HyBi chair prospective I see at moment much more energy in httpbi=
s wg than
in the hybi wg, and it would be great to leverage on that energy to also sp=
eed up this piece
of work

the best way I would suggest is to submit a proposal to the httpbis wg in f=
orm of a draft
(consider that the deadline to submit a draft is Friday 14th, so you have 5=
 days, if you
want have a chance for the proposal to be discussed to the next IETF meetin=
g in London
that will start March 2nd).

I would really encourage you do submit a draft, it is OK if it contains alt=
ernative options,
and so take the opportunity to discuss it in London

if you need any help in editing the draft I can also offer my help

br
Salvatore

On Jan 25, 2014, at 6:24 PM, Peter Thorson <webmaster@zaphoyd.com<mailto:we=
bmaster@zaphoyd.com>> wrote:

Critical first question here: Do we have the cooperation of the HTTP 2.0 wo=
rking group or is the WS over HTTP 2.0 effort something being done unilater=
ally by the WebSocket working group?




--_000_454347D98E92496BA1F26CA963DC9845ericssoncom_
Content-Type: text/html; charset="us-ascii"
Content-ID: <B6D3A7D61125CF4D918E932995ADA4D8@ericsson.com>
Content-Transfer-Encoding: quoted-printable

<html>
<head>
<meta http-equiv=3D"Content-Type" content=3D"text/html; charset=3Dus-ascii"=
>
</head>
<body style=3D"word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-lin=
e-break: after-white-space; ">
<br>
<div>
<div>On Feb 11, 2014, at 9:25 AM, Yutaka Hirano &lt;<a href=3D"mailto:yhira=
no@chromium.org">yhirano@chromium.org</a>&gt;</div>
<div>&nbsp;wrote:</div>
<br class=3D"Apple-interchange-newline">
<blockquote type=3D"cite">
<div dir=3D"ltr">Hi Salvatore,
<div>Thank you for the suggestions.</div>
<div><br>
</div>
<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex">
<span style=3D"font-family:arial,sans-serif;font-size:13px">as one of HyBi =
wg chair, but without having been able to coordinate with my co-chair yet<b=
r>
</span>I do think that we should do this in cooperation with HTTP 2.0<br>
and maybe it would be even better move the work directly in the httpbis wg<=
br>
as from a HyBi chair prospective I see at moment much more energy in httpbi=
s wg than<br>
in the hybi wg, and it would be great to leverage on that energy to also sp=
eed up this piece<br>
of work</blockquote>
<div>I am waiting for Roberto's response. If his response is positive, I wi=
ll update the draft and it is OK with me to move this topic to httpbis.</di=
v>
</div>
</blockquote>
<div><br>
</div>
<div>note that I am only suggesting to submit to httpbis to check if the ht=
tpbis is willing (and it has the energy) to work on it or not.</div>
<div>In order to discover it, the best way is to submit a draft (with the s=
everal options) to httpbis and have a discussion there</div>
<div>in the httpbis mailing list and preferably also during the f2f meeting=
 in London</div>
<br>
<blockquote type=3D"cite">
<div dir=3D"ltr">
<div><br>
</div>
<div>For the RFC draft, I am trying. Though it is unclear if I can make it =
in time (I have never written one), I will do my best.</div>
</div>
</blockquote>
<div><br>
</div>
<div>if you need any help in put the text in an draft format I can help wit=
hout any problem.</div>
<div><br>
</div>
<div>br</div>
<div>Salvatore</div>
<br>
<blockquote type=3D"cite">
<div dir=3D"ltr">
<div><br>
</div>
<div>Thanks,</div>
<div><br>
</div>
</div>
<div class=3D"gmail_extra"><br>
<br>
<div class=3D"gmail_quote">On Mon, Feb 10, 2014 at 6:17 PM, Salvatore Loret=
o <span dir=3D"ltr">
&lt;<a href=3D"mailto:salvatore.loreto@ericsson.com" target=3D"_blank">salv=
atore.loreto@ericsson.com</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
<div style=3D"word-wrap:break-word">as one of HyBi wg chair, but without ha=
ving been able to coordinate with my co-chair yet
<div>I do think that we should do this in cooperation with HTTP 2.0</div>
<div>and maybe it would be even better move the work directly in the httpbi=
s wg</div>
<div>
<div>as from a HyBi chair prospective I see at moment much more energy in h=
ttpbis wg than</div>
<div>in the hybi wg, and it would be great to leverage on that energy to al=
so speed up this piece</div>
</div>
<div>of work</div>
<div><br>
</div>
<div>the best way I would suggest is to submit a proposal to the httpbis wg=
 in form of a draft</div>
<div>(consider that the deadline to submit a draft is Friday 14th, so you h=
ave 5 days, if you</div>
<div>want have a chance for the proposal to be discussed to the next IETF m=
eeting in London</div>
<div>that will start March 2nd).</div>
<div><br>
</div>
<div>I would really encourage you do submit a draft, it is OK if it contain=
s alternative options,</div>
<div>and so take the opportunity to discuss it in London</div>
<div><br>
</div>
<div>if you need any help in editing the draft I can also offer my help</di=
v>
<div><br>
</div>
<div>br</div>
<span class=3D"HOEnZb"><font color=3D"#888888">
<div>Salvatore</div>
</font></span>
<div class=3D"">
<div>
<div><br>
<div>
<div>On Jan 25, 2014, at 6:24 PM, Peter Thorson &lt;<a href=3D"mailto:webma=
ster@zaphoyd.com" target=3D"_blank">webmaster@zaphoyd.com</a>&gt; wrote:</d=
iv>
<br>
<blockquote type=3D"cite"><span style=3D"font-family:Helvetica;font-size:me=
dium;font-style:normal;font-variant:normal;font-weight:normal;letter-spacin=
g:normal;line-height:normal;text-align:-webkit-auto;text-indent:0px;text-tr=
ansform:none;white-space:normal;word-spacing:0px;display:inline!important;f=
loat:none">Critical
 first question here: Do we have the cooperation of the HTTP 2.0 working gr=
oup or is the WS over HTTP 2.0 effort something being done unilaterally by =
the WebSocket working group?</span></blockquote>
</div>
<br>
</div>
</div>
</div>
</div>
</blockquote>
</div>
<br>
</div>
</blockquote>
</div>
<br>
</body>
</html>

--_000_454347D98E92496BA1F26CA963DC9845ericssoncom_--


From tyoshino@google.com  Tue Feb 11 19:50:36 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 99C1A1A07A9 for <hybi@ietfa.amsl.com>; Tue, 11 Feb 2014 19:50:36 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.326
X-Spam-Level: 
X-Spam-Status: No, score=-1.326 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, J_CHICKENPOX_27=0.6, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Llws0sLbwwkM for <hybi@ietfa.amsl.com>; Tue, 11 Feb 2014 19:50:33 -0800 (PST)
Received: from mail-wi0-x22f.google.com (mail-wi0-x22f.google.com [IPv6:2a00:1450:400c:c05::22f]) by ietfa.amsl.com (Postfix) with ESMTP id CD1B41A033A for <hybi@ietf.org>; Tue, 11 Feb 2014 19:50:32 -0800 (PST)
Received: by mail-wi0-f175.google.com with SMTP id hm4so6027427wib.8 for <hybi@ietf.org>; Tue, 11 Feb 2014 19:50:31 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc:content-type; bh=IG1az5kh75gVAp5c59lObBhhxjdmSsWLGrCL2hKPFAc=; b=gL2SARcSGptYi677LY7abd8laNaFM5OfupULeZoKOI5Th9kvnYdjk5EEv/Cd4O9Hf8 ghSIFllz/gJKfKmasdGN2Orkx7NSSYRUJKZpQ+n2socFcxcXnhZp8hjJEgSyYwcl5/iA TvY0YN4Isfmc6dhJ4fcALtB0t6WtH5bZW2P28d4CI07E+5gEDkCQ87gv5HtK7S7F2jn8 u3vD/a6tDkkVUODCV3hvhrTPlRR5ClVUEJnIy62z9cCflG7wY+k98fDSHgc4Ii1nhodj kP5Sv3vf7RGyH77tFn/2ATYlFOvWsZ4OntGyVMsjaiNI+2P2rI7f3SK9eK1a6M7W/EAb bU1g==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc:content-type; bh=IG1az5kh75gVAp5c59lObBhhxjdmSsWLGrCL2hKPFAc=; b=J970ePNDWIKNKL9mtmYT3s84xp5lZKDeSSGp+JhBB2OZ02uTBQOHkk4HB09iKZt5kh 4cvJYiG8TWLmGHm1X5re/iNV3yo23Xrxo2WsePlSjO6w5pEpz5+afq95Kd9ewbrszbzu Obafq36aLO6/7DW4AnXxlzPKASyZthGkbxd8MvYZ09gmztEKTIH9SOmM0NUN6PqKBivs pTCk7XaQRdovPNFITylj5gU+irZRuCaTslpj9e55W8l/9hBrnDgKWM+jFJsm3FmdaAa1 B6MPrSkuNg7TtWbePD7nuX38f/bCoWRMnTRG8stDkkwERI80Qk/TiG+k63ouwkkMPP0Q JkIw==
X-Gm-Message-State: ALoCoQnjl4hhP6GafGkANhK7OYqCarvO3WRbX1vBNnEIHbCQXfOoK1giINtkqBjXzHRZxOmIKt3ZdpIZNPQfnCcWjkGalELr3JKNpp0mFOpa0TZhOjG5dalQ5W/BFM4bjm71bg0NaAubSQAoh3Srv46sz1yA6MYhbD8Tv5PafI/pmz8UorOYVqIQAcmoQGUYRwJSSxylGtxe
X-Received: by 10.180.10.105 with SMTP id h9mr634952wib.11.1392177031443; Tue, 11 Feb 2014 19:50:31 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Tue, 11 Feb 2014 19:50:11 -0800 (PST)
In-Reply-To: <634914A010D0B943A035D226786325D4446BE403EB@EXVMBX020-12.exch020.serverdata.net>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E0@EXVMBX020-12.exch020.serverdata.net> <52CE8D59.3070802@it.aoyama.ac.jp> <634914A010D0B943A035D226786325D4446BE403EB@EXVMBX020-12.exch020.serverdata.net>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Wed, 12 Feb 2014 12:50:11 +0900
Message-ID: <CAH9hSJZfEoYv9uxH31R-s1zbXy7PqQW7MsA475VJJdFrjDhj_w@mail.gmail.com>
To: Tobias Oberstein <tobias.oberstein@tavendo.de>
Content-Type: multipart/alternative; boundary=001a11c263666cbea204f22d794b
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Detection of broken WS connections
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 12 Feb 2014 03:50:36 -0000

--001a11c263666cbea204f22d794b
Content-Type: text/plain; charset=UTF-8

On Thu, Jan 9, 2014 at 9:13 PM, Tobias Oberstein <
tobias.oberstein@tavendo.de> wrote:

> > The WS API is explicit about the fact that there is no way from
> JavaScript to
> > generate PINGs or check for PONGs. So do I just create my own,
> application
> > level equivalent, and send a ping-like message every so often and assume
>
> You can have a WS server that does automatic ping-pong for connection
> keep-alive/loss-detection.
>
>
This works for server. For client, app level work is needed.


> Or you can do it on app level from JS.
>
> > the connection is dead when I don't get a reply "soonish"? Are there
> libraries
> > that support such a mechanism? Or is every browser supported to regularly
> > PINGPONG with the server, and call the onclose function when there's no
>
> As far as I know, there is no JS API where you could request the browser
> to do automatic ping-pong upon opening a WS connection.
>
> > answer?
> >
>

Right. We've been saying that we should start discussing such a change on
the API, but haven't.

// Send a ping (w/o body) every 5 sec
ws.pingInterval = 5;
// _Close the WebSocket connection_ when it passed 10 sec without receiving
any byte (not limited to PONG)
ws.timeout = 10;

--001a11c263666cbea204f22d794b
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote">On T=
hu, Jan 9, 2014 at 9:13 PM, Tobias Oberstein <span dir=3D"ltr">&lt;<a href=
=3D"mailto:tobias.oberstein@tavendo.de" target=3D"_blank">tobias.oberstein@=
tavendo.de</a>&gt;</span> wrote:<br>

<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div class=3D"im">&gt; The WS API is explici=
t about the fact that there is no way from JavaScript to<br>
&gt; generate PINGs or check for PONGs. So do I just create my own, applica=
tion<br>
&gt; level equivalent, and send a ping-like message every so often and assu=
me<br>
<br>
</div>You can have a WS server that does automatic ping-pong for connection=
 keep-alive/loss-detection.<br>
<br></blockquote><div><br></div><div>This works for server. For client, app=
 level work is needed.</div><div>=C2=A0</div><blockquote class=3D"gmail_quo=
te" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"=
>
Or you can do it on app level from JS.<br>
<div class=3D"im"><br>
&gt; the connection is dead when I don&#39;t get a reply &quot;soonish&quot=
;? Are there libraries<br>
&gt; that support such a mechanism? Or is every browser supported to regula=
rly<br>
&gt; PINGPONG with the server, and call the onclose function when there&#39=
;s no<br>
<br>
</div>As far as I know, there is no JS API where you could request the brow=
ser to do automatic ping-pong upon opening a WS connection.<br>
<div class=3D"HOEnZb"><div class=3D"h5"><br>
&gt; answer?<br>
&gt;<br></div></div></blockquote><div><br></div><div>Right. We&#39;ve been =
saying that we should start discussing such a change on the API, but haven&=
#39;t.</div><div><br></div><div>// Send a ping (w/o body) every 5 sec</div>

<div>ws.pingInterval =3D 5;</div><div>// _Close the WebSocket connection_ w=
hen it passed 10 sec without receiving any byte (not limited to PONG)</div>=
<div>ws.timeout =3D 10;</div><div>=C2=A0</div></div></div></div>

--001a11c263666cbea204f22d794b--


From tyoshino@google.com  Tue Feb 11 20:37:19 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id E23111A0821 for <hybi@ietfa.amsl.com>; Tue, 11 Feb 2014 20:37:19 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.926
X-Spam-Level: 
X-Spam-Status: No, score=-1.926 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id JbUEWnbgMRs4 for <hybi@ietfa.amsl.com>; Tue, 11 Feb 2014 20:37:17 -0800 (PST)
Received: from mail-wi0-x22e.google.com (mail-wi0-x22e.google.com [IPv6:2a00:1450:400c:c05::22e]) by ietfa.amsl.com (Postfix) with ESMTP id BCEF31A033A for <hybi@ietf.org>; Tue, 11 Feb 2014 20:37:16 -0800 (PST)
Received: by mail-wi0-f174.google.com with SMTP id f8so6067328wiw.7 for <hybi@ietf.org>; Tue, 11 Feb 2014 20:37:15 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc:content-type; bh=DpRNmT6DPhZabkKVlLw3/CgKl3Wcg1ksv6vSkKKAofc=; b=UUv+JrtdNCRnplkDQjtgc3DuepnlVRzYrZuoLJOcsRs/CC7kPw7grvSQUnbQLtm3e3 KuHFuxdn1nmeTozpK0Z5Z5ua3ckcYbNrr8pftxtPz8XirloFqWXYmLyt4yN+YZA3Xes8 QJsK3sa7lHPEnamodTHUGQsFkCDPopNmFN5PA5cfMZg2uwjLeJymCpn9jp1EiOAHzJfI TacLHeCsu1awgL+xI+SQSK4n30/u021bEjHcF1JrNN/x9srm5oJ6cBU7xMaAVPGeGkIq pfcRqByqTY2RmzCj9hFzpdluSqStmlYJWkQykWITQYaroIZirHhejSkFQ/TiGlnS3AVD Jiiw==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc:content-type; bh=DpRNmT6DPhZabkKVlLw3/CgKl3Wcg1ksv6vSkKKAofc=; b=F4WfaafG+uXxORRW7ahkqMrAw20jtHqyTd564NZvOaXX6mNFC2LgNw0AT8J6Dx/W4V MRyAgtIBip7Ev9Sqa9rLsXXZdiESV8FgRnfKgIauE9vqmIpXybjun+HZn+BsG4JrEHTU yPHIpDjlB5hlAieP0MSTQkjRhitGm3WWC7KtEKTykdUF8Y8xxwC7XNnXqF+406IWsh5n EAFgxCHoyOQjzWqYdmpQ0D6oar7B3xsLAeRHEOjx9jLcbq1D2KARf3t6Lr3NMaeLgl72 Aq/LZeYGHBbiAW3JT2+ObQTVjL+tuvbBxTpKXgpkpjwGGNwzY0X+2R+h9LHXGe47+d5/ la5g==
X-Gm-Message-State: ALoCoQleALKyzpTQdtpdZNN16u5R7/1n82ntVYnYp118IlMdLAH0fDec4AmbS6Nl73dcwTum4dn8U+kVfnK0+TC0rOj/Wo5oi3pfdzhS/JJhUCu/ft+2bJMEEskVG6ZxRBW3OGgGyU8vtU+3NoC2M9NSjJ3i6gqyHRfoh3l8v+k2TcWGtcpMgieJ1RebWJtawRZqn5wbI9/F
X-Received: by 10.180.107.136 with SMTP id hc8mr681509wib.11.1392179835566; Tue, 11 Feb 2014 20:37:15 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Tue, 11 Feb 2014 20:36:55 -0800 (PST)
In-Reply-To: <CAH9hSJaKKP-Mhf=oj-J663EPqQfHN=xktCpP0rWi2MqtfS3wKQ@mail.gmail.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <740FF040-F1E9-43E6-ACDE-0014FE4D012F@zaphoyd.com> <CABihn6HM_yXSB9RapNEQEG0_9OYfpHG3uH+=wnDwRJAFbh9EJA@mail.gmail.com> <0F978119-88F2-49CE-98C6-2A640527AA64@zaphoyd.com> <CAGzyod7BA_UXjN4r+Qr+P33QsPpeWWotxiScTpzqyhvbtPgNYQ@mail.gmail.com> <2B0B9332-D002-460A-8D50-B90B55999F42@zaphoyd.com> <CABihn6Hykczbu0YmUDuFcjVt2+xyeYzmdOpEkiZmP-2Vzuiuow@mail.gmail.com> <634914A010D0B943A035D226786325D4446C36C47C@EXVMBX020-12.exch020.serverdata.net> <CAH9hSJaKKP-Mhf=oj-J663EPqQfHN=xktCpP0rWi2MqtfS3wKQ@mail.gmail.com>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Wed, 12 Feb 2014 13:36:55 +0900
Message-ID: <CAH9hSJaAUkvxb2v4pKS7aaEgUUYnx1A0hau4M22OGidwMfD+Zg@mail.gmail.com>
To: Tobias Oberstein <tobias.oberstein@tavendo.de>
Content-Type: multipart/alternative; boundary=e89a8f3bad45903f6c04f22e2041
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 12 Feb 2014 04:37:20 -0000

--e89a8f3bad45903f6c04f22e2041
Content-Type: text/plain; charset=UTF-8

On Fri, Feb 7, 2014 at 9:11 PM, Takeshi Yoshino <tyoshino@google.com> wrote:

> On Fri, Feb 7, 2014 at 7:16 PM, Tobias Oberstein <
> tobias.oberstein@tavendo.de> wrote:
>
>> >Questions
>> >As you notice, there are two questions from me.
>> >1: Are you OK with deciding NOW if we should preserve the RFC6455
>> semantics or not?
>> >2: (If the answer for 1 is yes) Should we preserve the RFC6455 semantics?
>> >
>> >My answer is, yes & yes.
>>
>> My answer: yes & yes.
>
>
> Yes & Yes
>

Complement:

I agree per-frame attribute is evil and am not interested in it. Writing
the compression and mux spec motivated me much not to use per-frame
attribute.

I want to keep control frame injection feature (and as a side effect, frame
boundary will be preserved) because I'm not yet sure throwing it away and
re-wire only ping/pong/close with what's available in HTTP/2.0 framing is
also fine as well as killing per-frame attribute. So, I'm inclined to keep
RFC6455 frame boundary, or provide an opaque box at HTTP/2.0 framing level
that is distinguished from normal DATA traffic. That's why I answered "yes"
to the second question.

--e89a8f3bad45903f6c04f22e2041
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_extra"><div>On Fri, Feb 7, 2014 at 9:1=
1 PM, Takeshi Yoshino <span dir=3D"ltr">&lt;<a href=3D"mailto:tyoshino@goog=
le.com" target=3D"_blank">tyoshino@google.com</a>&gt;</span> wrote:<br></di=
v><div class=3D"gmail_quote">

<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div dir=3D"ltr"><div class=3D"im"><div class=3D"gmail_ext=
ra"><div class=3D"gmail_quote">

On Fri, Feb 7, 2014 at 7:16 PM, Tobias Oberstein <span dir=3D"ltr">&lt;<a h=
ref=3D"mailto:tobias.oberstein@tavendo.de" target=3D"_blank">tobias.oberste=
in@tavendo.de</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div>&gt;Questions<br>
&gt;As you notice, there are two questions from me.<br>
&gt;1: Are you OK with deciding NOW if we should preserve the RFC6455 seman=
tics or not?<br>
&gt;2: (If the answer for 1 is yes) Should we preserve the RFC6455 semantic=
s?<br>
&gt;<br>
&gt;My answer is, yes &amp; yes.<br>
<br>
</div>My answer: yes &amp; yes.=C2=A0</blockquote></div><br></div></div><di=
v class=3D"gmail_extra">Yes &amp; Yes</div></div></blockquote><div><br></di=
v>Complement:<div><br><div>I agree per-frame attribute is evil and am not i=
nterested in it. Writing the compression and mux spec motivated me much not=
 to use per-frame attribute.<br>

</div><div><br></div><div>I want to keep control frame injection feature (a=
nd as a side effect, frame boundary will be preserved) because I&#39;m not =
yet sure throwing it away and re-wire only ping/pong/close with what&#39;s =
available in HTTP/2.0 framing is also fine as well as killing per-frame att=
ribute. So, I&#39;m inclined to keep RFC6455 frame boundary, or provide an =
opaque box at HTTP/2.0 framing level that is distinguished from normal DATA=
 traffic. That&#39;s why I answered &quot;yes&quot; to the second question.=
<br>

</div></div><div>=C2=A0</div></div></div></div>

--e89a8f3bad45903f6c04f22e2041--


From micheil@brandedcode.com  Wed Feb 12 05:34:26 2014
Return-Path: <micheil@brandedcode.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id A15991A0979 for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 05:34:26 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.6
X-Spam-Level: 
X-Spam-Status: No, score=-2.6 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, RCVD_IN_DNSWL_LOW=-0.7] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id vDQSo9LiNoA7 for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 05:34:23 -0800 (PST)
Received: from mail-we0-f175.google.com (mail-we0-f175.google.com [74.125.82.175]) by ietfa.amsl.com (Postfix) with ESMTP id 0B8D21A098F for <hybi@ietf.org>; Wed, 12 Feb 2014 05:34:19 -0800 (PST)
Received: by mail-we0-f175.google.com with SMTP id q59so5839693wes.6 for <hybi@ietf.org>; Wed, 12 Feb 2014 05:34:18 -0800 (PST)
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:content-type:mime-version:subject:from :in-reply-to:date:content-transfer-encoding:message-id:references:to; bh=pYyQYhQ23wgNSViK3MioDrHAV1tlBwyxajiCoVApPvI=; b=KKZ7tc+slEO3F4oxXeiFl9QCGSm4zEEQHVCQRaToxFgTdf0WvqJO9bHizDiRHDjyto QLX9xuIsBYMLpmmAj71ha05v6q77aD8sK2eFRWSTjPnybtjOlBD+JjAgxOOiVngWZ9aw a5ok760tX/KQDm1zYbHobjeJGvGyvJpE4bmQvCB9udfZouojBaHNZpr5B2JbLdyDNFkm kq3HYPJsjS8bHHTWRS9PKx+PPL2GXKVm3LMgSCbdVNxxJ7BAEztF49ja63Krr7HiwELl n4TtfB6PHkWXinu7GWhwGbY9YQc4U3pCvG1BDWBuWADXJt4Y2v36DUkTFFPS3BhTMT7t lhlA==
X-Gm-Message-State: ALoCoQnTtgGqPCVIUqyMWrGMuPqzH+M31oeciIJuLDlnvx1CGJCr8uEk+WTDiUhOe6RsFgw0NSL1
X-Received: by 10.180.211.208 with SMTP id ne16mr1937560wic.21.1392212058595;  Wed, 12 Feb 2014 05:34:18 -0800 (PST)
Received: from [10.0.1.75] (host-92-24-111-81.ppp.as43234.net. [92.24.111.81]) by mx.google.com with ESMTPSA id ju6sm52161512wjc.1.2014.02.12.05.34.17 for <multiple recipients> (version=TLSv1 cipher=ECDHE-RSA-RC4-SHA bits=128/128); Wed, 12 Feb 2014 05:34:17 -0800 (PST)
Content-Type: text/plain; charset=windows-1252
Mime-Version: 1.0 (Mac OS X Mail 7.0 \(1822\))
From: Micheil Smith <micheil@brandedcode.com>
In-Reply-To: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com>
Date: Wed, 12 Feb 2014 13:34:15 +0000
Content-Transfer-Encoding: quoted-printable
Message-Id: <8984EA8D-C6ED-49BF-92C4-113D69121B30@brandedcode.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com>
To: "hybi@ietf.org" <hybi@ietf.org>
X-Mailer: Apple Mail (2.1822)
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 12 Feb 2014 13:34:26 -0000

As far as I recall, this was also to enable more efficient reading and =
error checking.=20
Prefixing with the frame size means you don=92t need to scan every byte =
being=20
received in order to find an end of frame delimiter.

This is a particularly useful thing when doing something like =
upstreaming on a
websocket server, where all you care about is the very start of the =
message,=20
and the length, such that you can then just read() N bytes which is the =
size of=20
your frame and just write() directly.

In other words, I agree on your reply Takeshi.

=96 Micheil

On 7 Feb 2014, at 7:30 am, Takeshi Yoshino <tyoshino@google.com> wrote:

> I remember that in early days of the standardization, somebody was =
proposing (*) that we should have a message length field in the first =
fragment to allow for allocating buffer for full message at the =
beginning.
>=20
> On Thu, Jan 16, 2014 at 3:27 PM, "Martin J. D=FCrst" =
<duerst@it.aoyama.ac.jp> wrote:
> On 2014/01/16 15:05, Yutaka Hirano wrote:
> Hi Joakim,
>=20
> Thanks for the comment.
>=20
> Knowing the size of a frame at the start of the frame, traditional
> websocket behavior, is desired, as it allows us to fast fail for the =
1009
> (message too big) use case.
>=20
> Do you know how effective it is?
>=20
> Having no prior knowledge of frame size is undesirable to us.
>=20
> On the other hand, the frame sender have more freedom because they =
don't
> have to know the size of the frame in front.
>=20
> My understanding of why messages are split into frames was that this =
would allow to start a message without knowing its length. But if you =
don't know the size of a frame before sending it, then the right =
solution is to make it smaller. Frames are just arbitrary pieces of a =
message.
>=20
> We should be clear if we want to make it able to
> (a) know the length of WebSocket frame on receiving the header of each =
fragment
> (b) know the length of WebSocket message on receiving the first =
fragment
> =20
> As the proposal (*) has been rejected and not in the RFC6455 header, =
(b) is not true even in RFC6455. What Joakim said should be to keep (a) =
which is true now in RFC6455.
>=20
> But if a message is fragmented into smaller frames, we cannot reject =
(and send 1009) the message on receiving the first frame. "Whether we =
can fast-reject a big message on receiving some non-final frame" depends =
on how the peer fragments a message.
>=20
> In some plans discussed in the main thread, it's planned to drop frame =
length info for efficiency. But something else such as the length field =
of HTTP/2.0 DATA frame tells how much data the fragment (when multiple =
DATAs convey a WebSocket frame, each DATA would be called fragments of a =
frame) contains. So, we won't be able to know the length of fragments =
(original/encapsulated WebSocket frames) on receiving each of them, but =
still be able to know the length of sub-fragments (e.g. HTTP/2.0 data =
frames).
>=20
> I think this change is acceptable and shouldn't be considered as =
breaking of RFC6455 semantics. Even in the mux extension I was editing =
(http://tools.ietf.org/html/draft-ietf-hybi-websocket-multiplexing-11), =
I made the same change. The length of the encapsulated frame is unknown =
until receiving all the fragments of the encapsulating message. No one =
was objecting to that when we were discussing the spec, IIRC.
>=20
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi


From joakim@intalio.com  Wed Feb 12 05:49:43 2014
Return-Path: <joakim@intalio.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 0AA1F1A097A for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 05:49:43 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.977
X-Spam-Level: 
X-Spam-Status: No, score=-1.977 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_LOW=-0.7] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id K45BRJ_S0r_q for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 05:49:38 -0800 (PST)
Received: from mail-ea0-f173.google.com (mail-ea0-f173.google.com [209.85.215.173]) by ietfa.amsl.com (Postfix) with ESMTP id A27651A02C6 for <hybi@ietf.org>; Wed, 12 Feb 2014 05:49:38 -0800 (PST)
Received: by mail-ea0-f173.google.com with SMTP id d10so4450187eaj.32 for <hybi@ietf.org>; Wed, 12 Feb 2014 05:49:37 -0800 (PST)
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=NudZmZQyLL+nRT01Cc1M0k/oO4qoJELoThxIjQvT+N8=; b=iwE77A89nXGBx/EeI9rovBeWg9ly47vX0NLtPUQg58PnS9uc2lqgLTGeRjOe8h9zBm 37CrFbGryNibmUaFlJ7IHOFmbMHQrYrYRXyDWi8ku1vaaHyT+miuK4V1YwSlOb8sGQfh bKx/nWxV1xfYeEv358q5/KuQ2oDK/8aIabV8QesL7kCAGO1h4nW71NYLvMvzoKkfASg8 OuLbiIAJuMowNW0stuEHn4s4m/Up2r6ElZsW6N/7G6uITYEGESTeiHnqxZjjDw5cn8ZB l7jzbQMCH0hOOqAsSEwnJcki5BOsXSWCNanwg6jeEv76yMWMSbGnVZ2R4V7E+vlBa8Lt zC+Q==
X-Gm-Message-State: ALoCoQlC1guu6qL8VjU7Znir2mI+9xIPTEw53/DBZQG0Tk2YPxp6OKVBUoZy2xowIcsMnhxznPpr
MIME-Version: 1.0
X-Received: by 10.15.102.78 with SMTP id bq54mr2327938eeb.94.1392212977309; Wed, 12 Feb 2014 05:49:37 -0800 (PST)
Received: by 10.14.152.202 with HTTP; Wed, 12 Feb 2014 05:49:37 -0800 (PST)
In-Reply-To: <8984EA8D-C6ED-49BF-92C4-113D69121B30@brandedcode.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <8984EA8D-C6ED-49BF-92C4-113D69121B30@brandedcode.com>
Date: Wed, 12 Feb 2014 06:49:37 -0700
Message-ID: <CAG4zZZAfB_phXSVku+W+VbGRNMJ3+=9YPw8_dD6zSJr5041iEw@mail.gmail.com>
From: Joakim Erdfelt <joakim@intalio.com>
To: Micheil Smith <micheil@brandedcode.com>
Content-Type: multipart/alternative; boundary=089e016812d4f6f49d04f235d7e6
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 12 Feb 2014 13:49:43 -0000

--089e016812d4f6f49d04f235d7e6
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

For various application level APIs that depend on delivery of complete
messages, the need to know the frame level size at the start of the frame
becomes important for memory management.  Eclipse Jetty uses this
information to implement Java's javax.websocket API for limiting memory
use, along with lower level (WS+HTTP/1.1) protocol validation (close code
1009).

Since this group has a history (now) of making protocol decisions based on
existing public APIs (re Javascript WebSocket), I would entreat the members
here to think about the other public, industry standard, non-javascript,
WebSocket APIs already widely deployed.


--
Joakim Erdfelt <joakim@intalio.com>
webtide.com <http://www.webtide.com/> - intalio.com/jetty
Expert advice, services and support from from the Jetty & CometD experts
eclipse.org/jetty - cometd.org


On Wed, Feb 12, 2014 at 6:34 AM, Micheil Smith <micheil@brandedcode.com>wro=
te:

> As far as I recall, this was also to enable more efficient reading and
> error checking.
> Prefixing with the frame size means you don't need to scan every byte bei=
ng
> received in order to find an end of frame delimiter.
>
> This is a particularly useful thing when doing something like upstreaming
> on a
> websocket server, where all you care about is the very start of the
> message,
> and the length, such that you can then just read() N bytes which is the
> size of
> your frame and just write() directly.
>
> In other words, I agree on your reply Takeshi.
>
> - Micheil
>
> On 7 Feb 2014, at 7:30 am, Takeshi Yoshino <tyoshino@google.com> wrote:
>
> > I remember that in early days of the standardization, somebody was
> proposing (*) that we should have a message length field in the first
> fragment to allow for allocating buffer for full message at the beginning=
.
> >
> > On Thu, Jan 16, 2014 at 3:27 PM, "Martin J. D=FCrst" <
> duerst@it.aoyama.ac.jp> wrote:
> > On 2014/01/16 15:05, Yutaka Hirano wrote:
> > Hi Joakim,
> >
> > Thanks for the comment.
> >
> > Knowing the size of a frame at the start of the frame, traditional
> > websocket behavior, is desired, as it allows us to fast fail for the 10=
09
> > (message too big) use case.
> >
> > Do you know how effective it is?
> >
> > Having no prior knowledge of frame size is undesirable to us.
> >
> > On the other hand, the frame sender have more freedom because they don'=
t
> > have to know the size of the frame in front.
> >
> > My understanding of why messages are split into frames was that this
> would allow to start a message without knowing its length. But if you don=
't
> know the size of a frame before sending it, then the right solution is to
> make it smaller. Frames are just arbitrary pieces of a message.
> >
> > We should be clear if we want to make it able to
> > (a) know the length of WebSocket frame on receiving the header of each
> fragment
> > (b) know the length of WebSocket message on receiving the first fragmen=
t
> >
> > As the proposal (*) has been rejected and not in the RFC6455 header, (b=
)
> is not true even in RFC6455. What Joakim said should be to keep (a) which
> is true now in RFC6455.
> >
> > But if a message is fragmented into smaller frames, we cannot reject
> (and send 1009) the message on receiving the first frame. "Whether we can
> fast-reject a big message on receiving some non-final frame" depends on h=
ow
> the peer fragments a message.
> >
> > In some plans discussed in the main thread, it's planned to drop frame
> length info for efficiency. But something else such as the length field o=
f
> HTTP/2.0 DATA frame tells how much data the fragment (when multiple DATAs
> convey a WebSocket frame, each DATA would be called fragments of a frame)
> contains. So, we won't be able to know the length of fragments
> (original/encapsulated WebSocket frames) on receiving each of them, but
> still be able to know the length of sub-fragments (e.g. HTTP/2.0 data
> frames).
> >
> > I think this change is acceptable and shouldn't be considered as
> breaking of RFC6455 semantics. Even in the mux extension I was editing (
> http://tools.ietf.org/html/draft-ietf-hybi-websocket-multiplexing-11), I
> made the same change. The length of the encapsulated frame is unknown unt=
il
> receiving all the fragments of the encapsulating message. No one was
> objecting to that when we were discussing the spec, IIRC.
> >
> > _______________________________________________
> > hybi mailing list
> > hybi@ietf.org
> > https://www.ietf.org/mailman/listinfo/hybi
>
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi
>

--089e016812d4f6f49d04f235d7e6
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">For various application level APIs that depend on delivery=
 of complete messages, the need to know the frame level size at the start o=
f the frame becomes important for memory management. &nbsp;Eclipse Jetty us=
es this information to implement Java&#39;s javax.websocket API for limitin=
g memory use, along with lower level (WS+HTTP/1.1) protocol validation (clo=
se code 1009).<div>
<br></div><div>Since this group has a history (now) of making protocol deci=
sions based on existing public APIs (re Javascript WebSocket), I would entr=
eat the members here to think about the other public, industry standard, no=
n-javascript, WebSocket APIs already widely deployed.</div>
<div><div><br></div></div></div><div class=3D"gmail_extra"><br clear=3D"all=
"><div><div dir=3D"ltr"><div>--</div><div>Joakim Erdfelt &lt;<a href=3D"mai=
lto:joakim@intalio.com" target=3D"_blank">joakim@intalio.com</a>&gt;</div><=
div><a href=3D"http://www.webtide.com/" target=3D"_blank">webtide.com</a>&n=
bsp;- <a href=3D"http://intalio.com/jetty" target=3D"_blank">intalio.com/je=
tty</a></div>
<div><span>Expert advice, services and support from&nbsp;</span><span>from =
the Jetty &amp; CometD experts</span></div><div><a href=3D"http://eclipse.o=
rg/jetty/" target=3D"_blank">eclipse.org/jetty</a>&nbsp;-&nbsp;<a href=3D"h=
ttp://cometd.org/" target=3D"_blank">cometd.org</a></div>
</div></div>
<br><br><div class=3D"gmail_quote">On Wed, Feb 12, 2014 at 6:34 AM, Micheil=
 Smith <span dir=3D"ltr">&lt;<a href=3D"mailto:micheil@brandedcode.com" tar=
get=3D"_blank">micheil@brandedcode.com</a>&gt;</span> wrote:<br><blockquote=
 class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc soli=
d;padding-left:1ex">
As far as I recall, this was also to enable more efficient reading and erro=
r checking.<br>
Prefixing with the frame size means you don&rsquo;t need to scan every byte=
 being<br>
received in order to find an end of frame delimiter.<br>
<br>
This is a particularly useful thing when doing something like upstreaming o=
n a<br>
websocket server, where all you care about is the very start of the message=
,<br>
and the length, such that you can then just read() N bytes which is the siz=
e of<br>
your frame and just write() directly.<br>
<br>
In other words, I agree on your reply Takeshi.<br>
<br>
&ndash; Micheil<br>
<div><div class=3D"h5"><br>
On 7 Feb 2014, at 7:30 am, Takeshi Yoshino &lt;<a href=3D"mailto:tyoshino@g=
oogle.com">tyoshino@google.com</a>&gt; wrote:<br>
<br>
&gt; I remember that in early days of the standardization, somebody was pro=
posing (*) that we should have a message length field in the first fragment=
 to allow for allocating buffer for full message at the beginning.<br>

&gt;<br>
&gt; On Thu, Jan 16, 2014 at 3:27 PM, &quot;Martin J. D=FCrst&quot; &lt;<a =
href=3D"mailto:duerst@it.aoyama.ac.jp">duerst@it.aoyama.ac.jp</a>&gt; wrote=
:<br>
&gt; On 2014/01/16 15:05, Yutaka Hirano wrote:<br>
&gt; Hi Joakim,<br>
&gt;<br>
&gt; Thanks for the comment.<br>
&gt;<br>
&gt; Knowing the size of a frame at the start of the frame, traditional<br>
&gt; websocket behavior, is desired, as it allows us to fast fail for the 1=
009<br>
&gt; (message too big) use case.<br>
&gt;<br>
&gt; Do you know how effective it is?<br>
&gt;<br>
&gt; Having no prior knowledge of frame size is undesirable to us.<br>
&gt;<br>
&gt; On the other hand, the frame sender have more freedom because they don=
&#39;t<br>
&gt; have to know the size of the frame in front.<br>
&gt;<br>
&gt; My understanding of why messages are split into frames was that this w=
ould allow to start a message without knowing its length. But if you don&#3=
9;t know the size of a frame before sending it, then the right solution is =
to make it smaller. Frames are just arbitrary pieces of a message.<br>

&gt;<br>
&gt; We should be clear if we want to make it able to<br>
&gt; (a) know the length of WebSocket frame on receiving the header of each=
 fragment<br>
&gt; (b) know the length of WebSocket message on receiving the first fragme=
nt<br>
&gt;<br>
&gt; As the proposal (*) has been rejected and not in the RFC6455 header, (=
b) is not true even in RFC6455. What Joakim said should be to keep (a) whic=
h is true now in RFC6455.<br>
&gt;<br>
&gt; But if a message is fragmented into smaller frames, we cannot reject (=
and send 1009) the message on receiving the first frame. &quot;Whether we c=
an fast-reject a big message on receiving some non-final frame&quot; depend=
s on how the peer fragments a message.<br>

&gt;<br>
&gt; In some plans discussed in the main thread, it&#39;s planned to drop f=
rame length info for efficiency. But something else such as the length fiel=
d of HTTP/2.0 DATA frame tells how much data the fragment (when multiple DA=
TAs convey a WebSocket frame, each DATA would be called fragments of a fram=
e) contains. So, we won&#39;t be able to know the length of fragments (orig=
inal/encapsulated WebSocket frames) on receiving each of them, but still be=
 able to know the length of sub-fragments (e.g. HTTP/2.0 data frames).<br>

&gt;<br>
&gt; I think this change is acceptable and shouldn&#39;t be considered as b=
reaking of RFC6455 semantics. Even in the mux extension I was editing (<a h=
ref=3D"http://tools.ietf.org/html/draft-ietf-hybi-websocket-multiplexing-11=
" target=3D"_blank">http://tools.ietf.org/html/draft-ietf-hybi-websocket-mu=
ltiplexing-11</a>), I made the same change. The length of the encapsulated =
frame is unknown until receiving all the fragments of the encapsulating mes=
sage. No one was objecting to that when we were discussing the spec, IIRC.<=
br>

&gt;<br>
</div></div>&gt; _______________________________________________<br>
&gt; hybi mailing list<br>
&gt; <a href=3D"mailto:hybi@ietf.org">hybi@ietf.org</a><br>
&gt; <a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blan=
k">https://www.ietf.org/mailman/listinfo/hybi</a><br>
<br>
_______________________________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
</blockquote></div><br></div>

--089e016812d4f6f49d04f235d7e6--


From zhong.j.yu@gmail.com  Wed Feb 12 08:00:48 2014
Return-Path: <zhong.j.yu@gmail.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id B41D21A0458 for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 08:00:48 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2
X-Spam-Level: 
X-Spam-Status: No, score=-2 tagged_above=-999 required=5 tests=[BAYES_00=-1.9,  DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FREEMAIL_FROM=0.001, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id O-0QspRH4Ene for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 08:00:46 -0800 (PST)
Received: from mail-ob0-x229.google.com (mail-ob0-x229.google.com [IPv6:2607:f8b0:4003:c01::229]) by ietfa.amsl.com (Postfix) with ESMTP id 267531A09AC for <hybi@ietf.org>; Wed, 12 Feb 2014 08:00:46 -0800 (PST)
Received: by mail-ob0-f169.google.com with SMTP id wo20so10785917obc.0 for <hybi@ietf.org>; Wed, 12 Feb 2014 08:00:45 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type:content-transfer-encoding; bh=joqDS40TH4dHsPbzugDTrcKD1QqMlIBKt6L54Y1DE1c=; b=IqUwz03kfYFaSsB4/TeEIIXL/yC0r7kBhoU3rKyVKZ5lVSwEK6dXfXKijtSVW3iVqp kUr3YArodO6bC36J8UDTqxoRk/gbo17uUL7TP9p7zK38lW6pcRGns5yBMq68Km4kwkM7 WDpQmnR7hS+W/drIwMBCvP9TcZiKqFsrrnapCuLrl9jlrXTyoN4alFzmE7qsyBmqtLqd WyXeb24IHyMb/HENTb0t6J/Lw9CM73FvAW+kiqRCBQXD/oDgUvMHY2WGXTnla43I/TUN 6s9OZ7J3VcFh7m9U+cPXbxRPZcBGmPcGuwu0r4nX7M02RBRGaiVpA/3G74udj9oPs5ZV 9OhA==
MIME-Version: 1.0
X-Received: by 10.182.28.134 with SMTP id b6mr37762054obh.27.1392220845019; Wed, 12 Feb 2014 08:00:45 -0800 (PST)
Received: by 10.76.21.132 with HTTP; Wed, 12 Feb 2014 08:00:44 -0800 (PST)
In-Reply-To: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com>
Date: Wed, 12 Feb 2014 10:00:44 -0600
Message-ID: <CACuKZqEcA1Pv8RpWfmThMjTzi2BbVMMKXqujs6BxVfxRPZJ9NQ@mail.gmail.com>
From: Zhong Yu <zhong.j.yu@gmail.com>
To: Takeshi Yoshino <tyoshino@google.com>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 12 Feb 2014 16:00:48 -0000

Why not send the message in one frame? The frame length is then the
message length.

One argument for fragmentation is to insert control frames while
transmitting a long message. However that reasoning is not very
convincing:

PONG - as long as we are writing data, any data, the peer should see
that the connection is alive. It's unreasonable for the peer to think
that our end is unresponsive while it is receiving data from us.

PING - not necessary while we are writing data. If peer is
unresponsive, we'll learn that from TCP layer.

CLOSE - closing the connection in the middle of a message is abnormal
anyway, it's probably not important to do the close handshake.

Zhong Yu


On Fri, Feb 7, 2014 at 1:30 AM, Takeshi Yoshino <tyoshino@google.com> wrote=
:
> I remember that in early days of the standardization, somebody was propos=
ing
> (*) that we should have a message length field in the first fragment to
> allow for allocating buffer for full message at the beginning.
>
> On Thu, Jan 16, 2014 at 3:27 PM, "Martin J. D=FCrst" <duerst@it.aoyama.ac=
.jp>
> wrote:
>>
>> On 2014/01/16 15:05, Yutaka Hirano wrote:
>>>
>>> Hi Joakim,
>>>
>>> Thanks for the comment.
>>>
>>> Knowing the size of a frame at the start of the frame, traditional
>>>>
>>>> websocket behavior, is desired, as it allows us to fast fail for the
>>>> 1009
>>>> (message too big) use case.
>>>
>>>
>>> Do you know how effective it is?
>>>
>>> Having no prior knowledge of frame size is undesirable to us.
>>>
>>> On the other hand, the frame sender have more freedom because they don'=
t
>>> have to know the size of the frame in front.
>>
>>
>> My understanding of why messages are split into frames was that this wou=
ld
>> allow to start a message without knowing its length. But if you don't kn=
ow
>> the size of a frame before sending it, then the right solution is to mak=
e it
>> smaller. Frames are just arbitrary pieces of a message.
>
>
> We should be clear if we want to make it able to
> (a) know the length of WebSocket frame on receiving the header of each
> fragment
> (b) know the length of WebSocket message on receiving the first fragment
>
> As the proposal (*) has been rejected and not in the RFC6455 header, (b) =
is
> not true even in RFC6455. What Joakim said should be to keep (a) which is
> true now in RFC6455.
>
> But if a message is fragmented into smaller frames, we cannot reject (and
> send 1009) the message on receiving the first frame. "Whether we can
> fast-reject a big message on receiving some non-final frame" depends on h=
ow
> the peer fragments a message.
>
> In some plans discussed in the main thread, it's planned to drop frame
> length info for efficiency. But something else such as the length field o=
f
> HTTP/2.0 DATA frame tells how much data the fragment (when multiple DATAs
> convey a WebSocket frame, each DATA would be called fragments of a frame)
> contains. So, we won't be able to know the length of fragments
> (original/encapsulated WebSocket frames) on receiving each of them, but
> still be able to know the length of sub-fragments (e.g. HTTP/2.0 data
> frames).
>
> I think this change is acceptable and shouldn't be considered as breaking=
 of
> RFC6455 semantics. Even in the mux extension I was editing
> (http://tools.ietf.org/html/draft-ietf-hybi-websocket-multiplexing-11), I
> made the same change. The length of the encapsulated frame is unknown unt=
il
> receiving all the fragments of the encapsulating message. No one was
> objecting to that when we were discussing the spec, IIRC.
>
>
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi
>


From zhong.j.yu@gmail.com  Wed Feb 12 08:09:42 2014
Return-Path: <zhong.j.yu@gmail.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 1BA3D1A0338 for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 08:09:42 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2
X-Spam-Level: 
X-Spam-Status: No, score=-2 tagged_above=-999 required=5 tests=[BAYES_00=-1.9,  DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FREEMAIL_FROM=0.001, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id aGu2fmsNKBxr for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 08:09:37 -0800 (PST)
Received: from mail-oa0-x230.google.com (mail-oa0-x230.google.com [IPv6:2607:f8b0:4003:c02::230]) by ietfa.amsl.com (Postfix) with ESMTP id 16B681A08D1 for <hybi@ietf.org>; Wed, 12 Feb 2014 08:09:37 -0800 (PST)
Received: by mail-oa0-f48.google.com with SMTP id l6so11027374oag.7 for <hybi@ietf.org>; Wed, 12 Feb 2014 08:09:36 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type:content-transfer-encoding; bh=ixW4tt3NS9X5NBhI0caL7mMVzpo9OFwuVXkUcuDBlsY=; b=DVKhDZcmb7wxwlNDTl6v11pOg6D1PC9Zv/ZvwWbCrke3ujSd2nO0GQqCsM0FHZ93lO MarL/Ke8711G1RdHHBIdKpMhdHxsspCkWXzE975T9KWxtDwHlo7akWVYEXDINwlJz8MY YYAUi8jbs/cI0Fy5JbSTgWyeLE39nFqF9vDDzykFZeBkRm2M6bp7RVTml4VVKdFEiLwl LQgIqdNyquwbgJqFUHQTMGcCYms4wnDxs62AKO++OBi5Yq2qgKXcZO0vBFzH85Ka7KWN I7GM77A7CjFShGclOaajLuhTiNFZ6E0yGSBupY6Df9LE5y2MSPPYYDiR5suosvwtT05g Babg==
MIME-Version: 1.0
X-Received: by 10.60.37.99 with SMTP id x3mr1744003oej.61.1392221376091; Wed, 12 Feb 2014 08:09:36 -0800 (PST)
Received: by 10.76.21.132 with HTTP; Wed, 12 Feb 2014 08:09:36 -0800 (PST)
In-Reply-To: <CAG4zZZAfB_phXSVku+W+VbGRNMJ3+=9YPw8_dD6zSJr5041iEw@mail.gmail.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <8984EA8D-C6ED-49BF-92C4-113D69121B30@brandedcode.com> <CAG4zZZAfB_phXSVku+W+VbGRNMJ3+=9YPw8_dD6zSJr5041iEw@mail.gmail.com>
Date: Wed, 12 Feb 2014 10:09:36 -0600
Message-ID: <CACuKZqF+XZs0y04Svxb=rKhFEydwxp=8pwTiaL+3sm6cXF7_5Q@mail.gmail.com>
From: Zhong Yu <zhong.j.yu@gmail.com>
To: Joakim Erdfelt <joakim@intalio.com>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 12 Feb 2014 16:09:42 -0000

Why does it matter if a frame (of unknown size) is composed of
fragments (of known sizes), w.r.t. memory management?

On Wed, Feb 12, 2014 at 7:49 AM, Joakim Erdfelt <joakim@intalio.com> wrote:
> For various application level APIs that depend on delivery of complete
> messages, the need to know the frame level size at the start of the frame
> becomes important for memory management.  Eclipse Jetty uses this
> information to implement Java's javax.websocket API for limiting memory u=
se,
> along with lower level (WS+HTTP/1.1) protocol validation (close code 1009=
).
>
> Since this group has a history (now) of making protocol decisions based o=
n
> existing public APIs (re Javascript WebSocket), I would entreat the membe=
rs
> here to think about the other public, industry standard, non-javascript,
> WebSocket APIs already widely deployed.
>
>
> --
> Joakim Erdfelt <joakim@intalio.com>
> webtide.com - intalio.com/jetty
> Expert advice, services and support from from the Jetty & CometD experts
> eclipse.org/jetty - cometd.org
>
>
> On Wed, Feb 12, 2014 at 6:34 AM, Micheil Smith <micheil@brandedcode.com>
> wrote:
>>
>> As far as I recall, this was also to enable more efficient reading and
>> error checking.
>> Prefixing with the frame size means you don't need to scan every byte
>> being
>> received in order to find an end of frame delimiter.
>>
>> This is a particularly useful thing when doing something like upstreamin=
g
>> on a
>> websocket server, where all you care about is the very start of the
>> message,
>> and the length, such that you can then just read() N bytes which is the
>> size of
>> your frame and just write() directly.
>>
>> In other words, I agree on your reply Takeshi.
>>
>> - Micheil
>>
>> On 7 Feb 2014, at 7:30 am, Takeshi Yoshino <tyoshino@google.com> wrote:
>>
>> > I remember that in early days of the standardization, somebody was
>> > proposing (*) that we should have a message length field in the first
>> > fragment to allow for allocating buffer for full message at the beginn=
ing.
>> >
>> > On Thu, Jan 16, 2014 at 3:27 PM, "Martin J. D=FCrst"
>> > <duerst@it.aoyama.ac.jp> wrote:
>> > On 2014/01/16 15:05, Yutaka Hirano wrote:
>> > Hi Joakim,
>> >
>> > Thanks for the comment.
>> >
>> > Knowing the size of a frame at the start of the frame, traditional
>> > websocket behavior, is desired, as it allows us to fast fail for the
>> > 1009
>> > (message too big) use case.
>> >
>> > Do you know how effective it is?
>> >
>> > Having no prior knowledge of frame size is undesirable to us.
>> >
>> > On the other hand, the frame sender have more freedom because they don=
't
>> > have to know the size of the frame in front.
>> >
>> > My understanding of why messages are split into frames was that this
>> > would allow to start a message without knowing its length. But if you =
don't
>> > know the size of a frame before sending it, then the right solution is=
 to
>> > make it smaller. Frames are just arbitrary pieces of a message.
>> >
>> > We should be clear if we want to make it able to
>> > (a) know the length of WebSocket frame on receiving the header of each
>> > fragment
>> > (b) know the length of WebSocket message on receiving the first fragme=
nt
>> >
>> > As the proposal (*) has been rejected and not in the RFC6455 header, (=
b)
>> > is not true even in RFC6455. What Joakim said should be to keep (a) wh=
ich is
>> > true now in RFC6455.
>> >
>> > But if a message is fragmented into smaller frames, we cannot reject
>> > (and send 1009) the message on receiving the first frame. "Whether we =
can
>> > fast-reject a big message on receiving some non-final frame" depends o=
n how
>> > the peer fragments a message.
>> >
>> > In some plans discussed in the main thread, it's planned to drop frame
>> > length info for efficiency. But something else such as the length fiel=
d of
>> > HTTP/2.0 DATA frame tells how much data the fragment (when multiple DA=
TAs
>> > convey a WebSocket frame, each DATA would be called fragments of a fra=
me)
>> > contains. So, we won't be able to know the length of fragments
>> > (original/encapsulated WebSocket frames) on receiving each of them, bu=
t
>> > still be able to know the length of sub-fragments (e.g. HTTP/2.0 data
>> > frames).
>> >
>> > I think this change is acceptable and shouldn't be considered as
>> > breaking of RFC6455 semantics. Even in the mux extension I was editing
>> > (http://tools.ietf.org/html/draft-ietf-hybi-websocket-multiplexing-11)=
, I
>> > made the same change. The length of the encapsulated frame is unknown =
until
>> > receiving all the fragments of the encapsulating message. No one was
>> > objecting to that when we were discussing the spec, IIRC.
>> >
>> > _______________________________________________
>> > hybi mailing list
>> > hybi@ietf.org
>> > https://www.ietf.org/mailman/listinfo/hybi
>>
>> _______________________________________________
>> hybi mailing list
>> hybi@ietf.org
>> https://www.ietf.org/mailman/listinfo/hybi
>
>
>
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi
>


From joakim@intalio.com  Wed Feb 12 08:13:25 2014
Return-Path: <joakim@intalio.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 0ED5F1A08D1 for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 08:13:25 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.977
X-Spam-Level: 
X-Spam-Status: No, score=-1.977 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_LOW=-0.7] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id VA5mL7Z8OO6q for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 08:13:20 -0800 (PST)
Received: from mail-ee0-f42.google.com (mail-ee0-f42.google.com [74.125.83.42]) by ietfa.amsl.com (Postfix) with ESMTP id 18F151A0335 for <hybi@ietf.org>; Wed, 12 Feb 2014 08:13:19 -0800 (PST)
Received: by mail-ee0-f42.google.com with SMTP id b15so4470372eek.29 for <hybi@ietf.org>; Wed, 12 Feb 2014 08:13:18 -0800 (PST)
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=2NYckJ4zbkIAEzS/hDWV9kMF9yBUcD6cqjtrjsdgnUI=; b=XsYrroAn5R37Jg2zG0t6VsF3/T0N07tfYlNYPqK5g7jcLL5dXRiFtyQONIKN34Jh2w YG7tZfWSOqtv/RIZfl9ouqJCUiPWWIlrw7bkqpVr79EwYkb40YGM7xK1RXeS7RhNtVSK XBbg/6os/GztlNVNaG5kIqSBcAl1L8IDTvMd3ndDP9q3hdfH6RI23Zf9aDUcbWZ72Q5k XFpYaE9hWDS1n2RhI9WWl88UKgc1uXTTIRwvstkV+NLXmrymkHMX9DXLY6iZet4FE/Wg uMX5ssPrcZiiuKiwu07/51K/4fV0YgP/d5942L5Kvcpqjs8caCXVU54+wa5pZC715VTo ty0Q==
X-Gm-Message-State: ALoCoQlyO5QRmB7iXdO1C3NKZq9zq0wen9bLuqQe1IdDFmwOE9NtSHI835Ywilhs50uOfds9xm9I
MIME-Version: 1.0
X-Received: by 10.14.9.134 with SMTP id 6mr4937761eet.70.1392221598762; Wed, 12 Feb 2014 08:13:18 -0800 (PST)
Received: by 10.14.152.202 with HTTP; Wed, 12 Feb 2014 08:13:18 -0800 (PST)
In-Reply-To: <CACuKZqEcA1Pv8RpWfmThMjTzi2BbVMMKXqujs6BxVfxRPZJ9NQ@mail.gmail.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <CACuKZqEcA1Pv8RpWfmThMjTzi2BbVMMKXqujs6BxVfxRPZJ9NQ@mail.gmail.com>
Date: Wed, 12 Feb 2014 09:13:18 -0700
Message-ID: <CAG4zZZCr4aTfVpw2coX2g0qw++4kdgNCFVze6tHKZ+fJNqb0aQ@mail.gmail.com>
From: Joakim Erdfelt <joakim@intalio.com>
To: Zhong Yu <zhong.j.yu@gmail.com>
Content-Type: multipart/alternative; boundary=001a11c2951cd7de1904f237d92f
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 12 Feb 2014 16:13:25 -0000

--001a11c2951cd7de1904f237d92f
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

Scenario:
  Sending side uses a message stream based API to write out 1 websocket
message.
  The sending side's API is configured/implemented to buffer only 16k at a
time, writing out a frame of 16k for each buffer filled.
  The sending side writes out a 600k message.
  The receiving side is using a different API, one that is based on
complete messages.
  The receiving side is configured to not accept messages over 100k.
  The receiving side's API/implementation tosses a close code 1009 to the
sending side, per API.
  The receiving side's endpoint receives an error and close notification
indicating a protocol closure due to excessive message size from sending
side.



--
Joakim Erdfelt <joakim@intalio.com>
webtide.com <http://www.webtide.com/> - intalio.com/jetty
Expert advice, services and support from from the Jetty & CometD experts
eclipse.org/jetty - cometd.org


On Wed, Feb 12, 2014 at 9:00 AM, Zhong Yu <zhong.j.yu@gmail.com> wrote:

> Why not send the message in one frame? The frame length is then the
> message length.
>
> One argument for fragmentation is to insert control frames while
> transmitting a long message. However that reasoning is not very
> convincing:
>
> PONG - as long as we are writing data, any data, the peer should see
> that the connection is alive. It's unreasonable for the peer to think
> that our end is unresponsive while it is receiving data from us.
>
> PING - not necessary while we are writing data. If peer is
> unresponsive, we'll learn that from TCP layer.
>
> CLOSE - closing the connection in the middle of a message is abnormal
> anyway, it's probably not important to do the close handshake.
>
> Zhong Yu
>
>
> On Fri, Feb 7, 2014 at 1:30 AM, Takeshi Yoshino <tyoshino@google.com>
> wrote:
> > I remember that in early days of the standardization, somebody was
> proposing
> > (*) that we should have a message length field in the first fragment to
> > allow for allocating buffer for full message at the beginning.
> >
> > On Thu, Jan 16, 2014 at 3:27 PM, "Martin J. D=FCrst" <
> duerst@it.aoyama.ac.jp>
> > wrote:
> >>
> >> On 2014/01/16 15:05, Yutaka Hirano wrote:
> >>>
> >>> Hi Joakim,
> >>>
> >>> Thanks for the comment.
> >>>
> >>> Knowing the size of a frame at the start of the frame, traditional
> >>>>
> >>>> websocket behavior, is desired, as it allows us to fast fail for the
> >>>> 1009
> >>>> (message too big) use case.
> >>>
> >>>
> >>> Do you know how effective it is?
> >>>
> >>> Having no prior knowledge of frame size is undesirable to us.
> >>>
> >>> On the other hand, the frame sender have more freedom because they
> don't
> >>> have to know the size of the frame in front.
> >>
> >>
> >> My understanding of why messages are split into frames was that this
> would
> >> allow to start a message without knowing its length. But if you don't
> know
> >> the size of a frame before sending it, then the right solution is to
> make it
> >> smaller. Frames are just arbitrary pieces of a message.
> >
> >
> > We should be clear if we want to make it able to
> > (a) know the length of WebSocket frame on receiving the header of each
> > fragment
> > (b) know the length of WebSocket message on receiving the first fragmen=
t
> >
> > As the proposal (*) has been rejected and not in the RFC6455 header, (b=
)
> is
> > not true even in RFC6455. What Joakim said should be to keep (a) which =
is
> > true now in RFC6455.
> >
> > But if a message is fragmented into smaller frames, we cannot reject (a=
nd
> > send 1009) the message on receiving the first frame. "Whether we can
> > fast-reject a big message on receiving some non-final frame" depends on
> how
> > the peer fragments a message.
> >
> > In some plans discussed in the main thread, it's planned to drop frame
> > length info for efficiency. But something else such as the length field
> of
> > HTTP/2.0 DATA frame tells how much data the fragment (when multiple DAT=
As
> > convey a WebSocket frame, each DATA would be called fragments of a fram=
e)
> > contains. So, we won't be able to know the length of fragments
> > (original/encapsulated WebSocket frames) on receiving each of them, but
> > still be able to know the length of sub-fragments (e.g. HTTP/2.0 data
> > frames).
> >
> > I think this change is acceptable and shouldn't be considered as
> breaking of
> > RFC6455 semantics. Even in the mux extension I was editing
> > (http://tools.ietf.org/html/draft-ietf-hybi-websocket-multiplexing-11),
> I
> > made the same change. The length of the encapsulated frame is unknown
> until
> > receiving all the fragments of the encapsulating message. No one was
> > objecting to that when we were discussing the spec, IIRC.
> >
> >
> > _______________________________________________
> > hybi mailing list
> > hybi@ietf.org
> > https://www.ietf.org/mailman/listinfo/hybi
> >
>
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi
>

--001a11c2951cd7de1904f237d92f
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div>Scenario:</div><div>=A0 Sending side uses a message s=
tream based API to write out 1 websocket message.</div><div>=A0 The sending=
 side&#39;s API is configured/implemented to buffer only 16k at a time, wri=
ting out a frame of 16k for each buffer filled.</div>
<div>=A0 The sending side writes out a 600k message.</div><div>=A0 The rece=
iving side is using a different API, one that is based on complete messages=
.</div><div>=A0 The receiving side is configured to not accept messages ove=
r 100k.</div>
<div>=A0 The receiving side&#39;s API/implementation tosses a close code 10=
09 to the sending side, per API.</div><div>=A0 The receiving side&#39;s end=
point receives an error and close notification indicating a protocol closur=
e due to excessive message size from sending side.</div>
<div><br></div><div><br></div></div><div class=3D"gmail_extra"><br clear=3D=
"all"><div><div dir=3D"ltr"><div>--</div><div>Joakim Erdfelt &lt;<a href=3D=
"mailto:joakim@intalio.com" target=3D"_blank">joakim@intalio.com</a>&gt;</d=
iv><div>
<a href=3D"http://www.webtide.com/" target=3D"_blank">webtide.com</a>=A0- <=
a href=3D"http://intalio.com/jetty" target=3D"_blank">intalio.com/jetty</a>=
</div><div><span>Expert advice, services and support from=A0</span><span>fr=
om the Jetty &amp; CometD experts</span></div>
<div><a href=3D"http://eclipse.org/jetty/" target=3D"_blank">eclipse.org/je=
tty</a>=A0-=A0<a href=3D"http://cometd.org/" target=3D"_blank">cometd.org</=
a></div></div></div>
<br><br><div class=3D"gmail_quote">On Wed, Feb 12, 2014 at 9:00 AM, Zhong Y=
u <span dir=3D"ltr">&lt;<a href=3D"mailto:zhong.j.yu@gmail.com" target=3D"_=
blank">zhong.j.yu@gmail.com</a>&gt;</span> wrote:<br><blockquote class=3D"g=
mail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-l=
eft:1ex">
Why not send the message in one frame? The frame length is then the<br>
message length.<br>
<br>
One argument for fragmentation is to insert control frames while<br>
transmitting a long message. However that reasoning is not very<br>
convincing:<br>
<br>
PONG - as long as we are writing data, any data, the peer should see<br>
that the connection is alive. It&#39;s unreasonable for the peer to think<b=
r>
that our end is unresponsive while it is receiving data from us.<br>
<br>
PING - not necessary while we are writing data. If peer is<br>
unresponsive, we&#39;ll learn that from TCP layer.<br>
<br>
CLOSE - closing the connection in the middle of a message is abnormal<br>
anyway, it&#39;s probably not important to do the close handshake.<br>
<span class=3D"HOEnZb"><font color=3D"#888888"><br>
Zhong Yu<br>
</font></span><div class=3D"HOEnZb"><div class=3D"h5"><br>
<br>
On Fri, Feb 7, 2014 at 1:30 AM, Takeshi Yoshino &lt;<a href=3D"mailto:tyosh=
ino@google.com">tyoshino@google.com</a>&gt; wrote:<br>
&gt; I remember that in early days of the standardization, somebody was pro=
posing<br>
&gt; (*) that we should have a message length field in the first fragment t=
o<br>
&gt; allow for allocating buffer for full message at the beginning.<br>
&gt;<br>
&gt; On Thu, Jan 16, 2014 at 3:27 PM, &quot;Martin J. D=FCrst&quot; &lt;<a =
href=3D"mailto:duerst@it.aoyama.ac.jp">duerst@it.aoyama.ac.jp</a>&gt;<br>
&gt; wrote:<br>
&gt;&gt;<br>
&gt;&gt; On 2014/01/16 15:05, Yutaka Hirano wrote:<br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt; Hi Joakim,<br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt; Thanks for the comment.<br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt; Knowing the size of a frame at the start of the frame, traditi=
onal<br>
&gt;&gt;&gt;&gt;<br>
&gt;&gt;&gt;&gt; websocket behavior, is desired, as it allows us to fast fa=
il for the<br>
&gt;&gt;&gt;&gt; 1009<br>
&gt;&gt;&gt;&gt; (message too big) use case.<br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt; Do you know how effective it is?<br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt; Having no prior knowledge of frame size is undesirable to us.<=
br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt; On the other hand, the frame sender have more freedom because =
they don&#39;t<br>
&gt;&gt;&gt; have to know the size of the frame in front.<br>
&gt;&gt;<br>
&gt;&gt;<br>
&gt;&gt; My understanding of why messages are split into frames was that th=
is would<br>
&gt;&gt; allow to start a message without knowing its length. But if you do=
n&#39;t know<br>
&gt;&gt; the size of a frame before sending it, then the right solution is =
to make it<br>
&gt;&gt; smaller. Frames are just arbitrary pieces of a message.<br>
&gt;<br>
&gt;<br>
&gt; We should be clear if we want to make it able to<br>
&gt; (a) know the length of WebSocket frame on receiving the header of each=
<br>
&gt; fragment<br>
&gt; (b) know the length of WebSocket message on receiving the first fragme=
nt<br>
&gt;<br>
&gt; As the proposal (*) has been rejected and not in the RFC6455 header, (=
b) is<br>
&gt; not true even in RFC6455. What Joakim said should be to keep (a) which=
 is<br>
&gt; true now in RFC6455.<br>
&gt;<br>
&gt; But if a message is fragmented into smaller frames, we cannot reject (=
and<br>
&gt; send 1009) the message on receiving the first frame. &quot;Whether we =
can<br>
&gt; fast-reject a big message on receiving some non-final frame&quot; depe=
nds on how<br>
&gt; the peer fragments a message.<br>
&gt;<br>
&gt; In some plans discussed in the main thread, it&#39;s planned to drop f=
rame<br>
&gt; length info for efficiency. But something else such as the length fiel=
d of<br>
&gt; HTTP/2.0 DATA frame tells how much data the fragment (when multiple DA=
TAs<br>
&gt; convey a WebSocket frame, each DATA would be called fragments of a fra=
me)<br>
&gt; contains. So, we won&#39;t be able to know the length of fragments<br>
&gt; (original/encapsulated WebSocket frames) on receiving each of them, bu=
t<br>
&gt; still be able to know the length of sub-fragments (e.g. HTTP/2.0 data<=
br>
&gt; frames).<br>
&gt;<br>
&gt; I think this change is acceptable and shouldn&#39;t be considered as b=
reaking of<br>
&gt; RFC6455 semantics. Even in the mux extension I was editing<br>
&gt; (<a href=3D"http://tools.ietf.org/html/draft-ietf-hybi-websocket-multi=
plexing-11" target=3D"_blank">http://tools.ietf.org/html/draft-ietf-hybi-we=
bsocket-multiplexing-11</a>), I<br>
&gt; made the same change. The length of the encapsulated frame is unknown =
until<br>
&gt; receiving all the fragments of the encapsulating message. No one was<b=
r>
&gt; objecting to that when we were discussing the spec, IIRC.<br>
&gt;<br>
&gt;<br>
</div></div><div class=3D"HOEnZb"><div class=3D"h5">&gt; __________________=
_____________________________<br>
&gt; hybi mailing list<br>
&gt; <a href=3D"mailto:hybi@ietf.org">hybi@ietf.org</a><br>
&gt; <a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blan=
k">https://www.ietf.org/mailman/listinfo/hybi</a><br>
&gt;<br>
<br>
_______________________________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
</div></div></blockquote></div><br></div>

--001a11c2951cd7de1904f237d92f--


From webmaster@zaphoyd.com  Wed Feb 12 08:19:11 2014
Return-Path: <webmaster@zaphoyd.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id C11511A040E for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 08:19:11 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.9
X-Spam-Level: 
X-Spam-Status: No, score=-1.9 tagged_above=-999 required=5 tests=[BAYES_00=-1.9] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id loMT9XsEe4g4 for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 08:19:10 -0800 (PST)
Received: from authsmtp00.uchicago.edu (authsmtp00.uchicago.edu [128.135.12.120]) by ietfa.amsl.com (Postfix) with ESMTP id 272C21A0458 for <hybi@ietf.org>; Wed, 12 Feb 2014 08:19:10 -0800 (PST)
Received: from [10.0.1.89] (c-68-51-76-178.hsd1.il.comcast.net [68.51.76.178]) (Authenticated sender: zaphoyd) by authsmtp00.uchicago.edu (Postfix) with ESMTP id 8F1754980A2; Wed, 12 Feb 2014 10:19:08 -0600 (CST)
Content-Type: text/plain; charset=windows-1252
Mime-Version: 1.0 (Mac OS X Mail 7.0 \(1822\))
From: Peter Thorson <webmaster@zaphoyd.com>
In-Reply-To: <CACuKZqF+XZs0y04Svxb=rKhFEydwxp=8pwTiaL+3sm6cXF7_5Q@mail.gmail.com>
Date: Wed, 12 Feb 2014 10:19:11 -0600
Content-Transfer-Encoding: quoted-printable
Message-Id: <542F71B7-786C-4965-904E-60EE5369A56A@zaphoyd.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <8984EA8D-C6ED-49BF-92C4-113D69121B30@brandedcode.com> <CAG4zZZAfB_phXSVku+W+VbGRNMJ3+=9YPw8_dD6zSJr5041iEw@mail.gmail.com> <CACuKZqF+XZs0y04Svxb=rKhFEydwxp=8pwTiaL+3sm6cXF7_5Q@mail.gmail.com>
To: Zhong Yu <zhong.j.yu@gmail.com>
X-Mailer: Apple Mail (2.1822)
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 12 Feb 2014 16:19:12 -0000

It doesn=92t. Multiple layers of fixed length framing don=92t =
conceptually affect this sort of memory management. It does make the =
whole operation more complicated though.

On Feb 12, 2014, at 10:09, Zhong Yu <zhong.j.yu@gmail.com> wrote:

> Why does it matter if a frame (of unknown size) is composed of
> fragments (of known sizes), w.r.t. memory management?


From joakim@intalio.com  Wed Feb 12 08:20:02 2014
Return-Path: <joakim@intalio.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 343AF1A040E for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 08:20:02 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.977
X-Spam-Level: 
X-Spam-Status: No, score=-1.977 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_LOW=-0.7] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id MPN-4ftcPoTH for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 08:19:57 -0800 (PST)
Received: from mail-ea0-f182.google.com (mail-ea0-f182.google.com [209.85.215.182]) by ietfa.amsl.com (Postfix) with ESMTP id B0F661A0465 for <hybi@ietf.org>; Wed, 12 Feb 2014 08:19:56 -0800 (PST)
Received: by mail-ea0-f182.google.com with SMTP id r15so4547516ead.27 for <hybi@ietf.org>; Wed, 12 Feb 2014 08:19:55 -0800 (PST)
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=znXBQF/eY2n9tfI9oxlql7b5i2aN4uLEkKF5bg3KssA=; b=AH6bbC8gSAS0IUzU0R6Xm2ygocYsBDi7fg9ARVkEUtWsffv+uzZF6uV5PfNmaf3zly bq98W9fV7nWfgybBlA6Angr8xC1OLkstuhXpdHn2B5TR6YsABHFdz5ocug6+7YZZfkHx 7fonH1Vs1rkNP1TkdVEITQytmqUzfUCL44nZ8QYG0euQisakiVV49GILMEQSPCB3p7gd EUsTZ7YM7ky58Y4ctH4SybXVD5W17ul9tp8S9zbg3+xlJvtyE78cftoIXdptFJaTROOd RPhVWDNVfODRgL+olXXlZU1i/BRZzKvKinOoiqOmHpbHCUTBLD86MxZGFw+g44L1h1cU 6Kgg==
X-Gm-Message-State: ALoCoQmCVjqyr3eomaFCfgSXfNCMQvGuzMDACO0U8TeUkvYY+0/G/w45bXu4LxcYeCCdheg3iu0V
MIME-Version: 1.0
X-Received: by 10.14.119.197 with SMTP id n45mr2011528eeh.93.1392221995413; Wed, 12 Feb 2014 08:19:55 -0800 (PST)
Received: by 10.14.152.202 with HTTP; Wed, 12 Feb 2014 08:19:55 -0800 (PST)
In-Reply-To: <CAG4zZZCr4aTfVpw2coX2g0qw++4kdgNCFVze6tHKZ+fJNqb0aQ@mail.gmail.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <CACuKZqEcA1Pv8RpWfmThMjTzi2BbVMMKXqujs6BxVfxRPZJ9NQ@mail.gmail.com> <CAG4zZZCr4aTfVpw2coX2g0qw++4kdgNCFVze6tHKZ+fJNqb0aQ@mail.gmail.com>
Date: Wed, 12 Feb 2014 09:19:55 -0700
Message-ID: <CAG4zZZA4WR50Ea2e36XEgTgpJURwiMFn+DXeYuxHFfZatYzDOQ@mail.gmail.com>
From: Joakim Erdfelt <joakim@intalio.com>
To: Zhong Yu <zhong.j.yu@gmail.com>
Content-Type: multipart/alternative; boundary=001a1132f7f87c49be04f237f136
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 12 Feb 2014 16:20:02 -0000

--001a1132f7f87c49be04f237f136
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

Eliminating the size from the frame also eliminates the ability to
fast-fail on excessive messages.

Scenario (with current RFC6455 / WS+HTTP/1.1):
  Sending side starts to write out a single message of 100MB.
  Receiving side, using a whole message delivery based API is configured
for max messages of 20MB.
  Receiving side sees the 100MB frame size and automatically closes the
connection with 1009 and doesn't bother reading past the current network
buffer.

Scenario (without frame sizes):
  Sending side starts to write out a single message of 100MB.
  Receiving side, using a whole message delivery based API is configured
for max messages of 20MB.
  Receiving side reads till the 20MB mark, then fails the connection with
1009.  Wasting time, memory, and resources for no good reason.

I really don't see how a discussion of a protocol without sizes is even
going on.
I mean, even the IP layer and TCP layer have sizes for their frames.



--
Joakim Erdfelt <joakim@intalio.com>
webtide.com <http://www.webtide.com/> - intalio.com/jetty
Expert advice, services and support from from the Jetty & CometD experts
eclipse.org/jetty - cometd.org


On Wed, Feb 12, 2014 at 9:13 AM, Joakim Erdfelt <joakim@intalio.com> wrote:

> Scenario:
>   Sending side uses a message stream based API to write out 1 websocket
> message.
>   The sending side's API is configured/implemented to buffer only 16k at =
a
> time, writing out a frame of 16k for each buffer filled.
>   The sending side writes out a 600k message.
>   The receiving side is using a different API, one that is based on
> complete messages.
>   The receiving side is configured to not accept messages over 100k.
>   The receiving side's API/implementation tosses a close code 1009 to the
> sending side, per API.
>   The receiving side's endpoint receives an error and close notification
> indicating a protocol closure due to excessive message size from sending
> side.
>
>
>
> --
> Joakim Erdfelt <joakim@intalio.com>
> webtide.com <http://www.webtide.com/> - intalio.com/jetty
> Expert advice, services and support from from the Jetty & CometD experts
> eclipse.org/jetty - cometd.org
>
>
> On Wed, Feb 12, 2014 at 9:00 AM, Zhong Yu <zhong.j.yu@gmail.com> wrote:
>
>> Why not send the message in one frame? The frame length is then the
>> message length.
>>
>> One argument for fragmentation is to insert control frames while
>> transmitting a long message. However that reasoning is not very
>> convincing:
>>
>> PONG - as long as we are writing data, any data, the peer should see
>> that the connection is alive. It's unreasonable for the peer to think
>> that our end is unresponsive while it is receiving data from us.
>>
>> PING - not necessary while we are writing data. If peer is
>> unresponsive, we'll learn that from TCP layer.
>>
>> CLOSE - closing the connection in the middle of a message is abnormal
>> anyway, it's probably not important to do the close handshake.
>>
>> Zhong Yu
>>
>>
>> On Fri, Feb 7, 2014 at 1:30 AM, Takeshi Yoshino <tyoshino@google.com>
>> wrote:
>> > I remember that in early days of the standardization, somebody was
>> proposing
>> > (*) that we should have a message length field in the first fragment t=
o
>> > allow for allocating buffer for full message at the beginning.
>> >
>> > On Thu, Jan 16, 2014 at 3:27 PM, "Martin J. D=FCrst" <
>> duerst@it.aoyama.ac.jp>
>> > wrote:
>> >>
>> >> On 2014/01/16 15:05, Yutaka Hirano wrote:
>> >>>
>> >>> Hi Joakim,
>> >>>
>> >>> Thanks for the comment.
>> >>>
>> >>> Knowing the size of a frame at the start of the frame, traditional
>> >>>>
>> >>>> websocket behavior, is desired, as it allows us to fast fail for th=
e
>> >>>> 1009
>> >>>> (message too big) use case.
>> >>>
>> >>>
>> >>> Do you know how effective it is?
>> >>>
>> >>> Having no prior knowledge of frame size is undesirable to us.
>> >>>
>> >>> On the other hand, the frame sender have more freedom because they
>> don't
>> >>> have to know the size of the frame in front.
>> >>
>> >>
>> >> My understanding of why messages are split into frames was that this
>> would
>> >> allow to start a message without knowing its length. But if you don't
>> know
>> >> the size of a frame before sending it, then the right solution is to
>> make it
>> >> smaller. Frames are just arbitrary pieces of a message.
>> >
>> >
>> > We should be clear if we want to make it able to
>> > (a) know the length of WebSocket frame on receiving the header of each
>> > fragment
>> > (b) know the length of WebSocket message on receiving the first fragme=
nt
>> >
>> > As the proposal (*) has been rejected and not in the RFC6455 header,
>> (b) is
>> > not true even in RFC6455. What Joakim said should be to keep (a) which
>> is
>> > true now in RFC6455.
>> >
>> > But if a message is fragmented into smaller frames, we cannot reject
>> (and
>> > send 1009) the message on receiving the first frame. "Whether we can
>> > fast-reject a big message on receiving some non-final frame" depends o=
n
>> how
>> > the peer fragments a message.
>> >
>> > In some plans discussed in the main thread, it's planned to drop frame
>> > length info for efficiency. But something else such as the length fiel=
d
>> of
>> > HTTP/2.0 DATA frame tells how much data the fragment (when multiple
>> DATAs
>> > convey a WebSocket frame, each DATA would be called fragments of a
>> frame)
>> > contains. So, we won't be able to know the length of fragments
>> > (original/encapsulated WebSocket frames) on receiving each of them, bu=
t
>> > still be able to know the length of sub-fragments (e.g. HTTP/2.0 data
>> > frames).
>> >
>> > I think this change is acceptable and shouldn't be considered as
>> breaking of
>> > RFC6455 semantics. Even in the mux extension I was editing
>> > (http://tools.ietf.org/html/draft-ietf-hybi-websocket-multiplexing-11)=
,
>> I
>> > made the same change. The length of the encapsulated frame is unknown
>> until
>> > receiving all the fragments of the encapsulating message. No one was
>> > objecting to that when we were discussing the spec, IIRC.
>> >
>> >
>> > _______________________________________________
>> > hybi mailing list
>> > hybi@ietf.org
>> > https://www.ietf.org/mailman/listinfo/hybi
>> >
>>
>> _______________________________________________
>> hybi mailing list
>> hybi@ietf.org
>> https://www.ietf.org/mailman/listinfo/hybi
>>
>
>

--001a1132f7f87c49be04f237f136
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div>Eliminating the size from the frame also eliminates t=
he ability to fast-fail on excessive messages.</div><div><br></div><div>Sce=
nario (with current RFC6455 / WS+HTTP/1.1):</div><div>=A0 Sending side star=
ts to write out a single message of 100MB.</div>
<div>=A0 Receiving side, using a whole message delivery based API is config=
ured for max messages of 20MB.</div><div>=A0 Receiving side sees the 100MB =
frame size and automatically closes the connection with 1009 and doesn&#39;=
t bother reading past the current network buffer.</div>
<div><br></div><div>Scenario (without frame sizes):</div><div>=A0 Sending s=
ide starts to write out a single message of 100MB.</div><div><div>=A0 Recei=
ving side, using a whole message delivery based API is configured for max m=
essages of 20MB.</div>
<div>=A0 Receiving side reads till the 20MB mark, then fails the connection=
 with 1009. =A0Wasting time, memory, and resources for no good reason.</div=
></div><div><br></div>I really don&#39;t see how a discussion of a protocol=
 without sizes is even going on.<div>
I mean, even the IP layer and TCP layer have sizes for their frames.</div><=
div><br></div><div><br></div></div><div class=3D"gmail_extra"><br clear=3D"=
all"><div><div dir=3D"ltr"><div>--</div><div>Joakim Erdfelt &lt;<a href=3D"=
mailto:joakim@intalio.com" target=3D"_blank">joakim@intalio.com</a>&gt;</di=
v>
<div><a href=3D"http://www.webtide.com/" target=3D"_blank">webtide.com</a>=
=A0- <a href=3D"http://intalio.com/jetty" target=3D"_blank">intalio.com/jet=
ty</a></div><div><span>Expert advice, services and support from=A0</span><s=
pan>from the Jetty &amp; CometD experts</span></div>
<div><a href=3D"http://eclipse.org/jetty/" target=3D"_blank">eclipse.org/je=
tty</a>=A0-=A0<a href=3D"http://cometd.org/" target=3D"_blank">cometd.org</=
a></div></div></div>
<br><br><div class=3D"gmail_quote">On Wed, Feb 12, 2014 at 9:13 AM, Joakim =
Erdfelt <span dir=3D"ltr">&lt;<a href=3D"mailto:joakim@intalio.com" target=
=3D"_blank">joakim@intalio.com</a>&gt;</span> wrote:<br><blockquote class=
=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padd=
ing-left:1ex">
<div dir=3D"ltr"><div>Scenario:</div><div>=A0 Sending side uses a message s=
tream based API to write out 1 websocket message.</div><div>=A0 The sending=
 side&#39;s API is configured/implemented to buffer only 16k at a time, wri=
ting out a frame of 16k for each buffer filled.</div>

<div>=A0 The sending side writes out a 600k message.</div><div>=A0 The rece=
iving side is using a different API, one that is based on complete messages=
.</div><div>=A0 The receiving side is configured to not accept messages ove=
r 100k.</div>

<div>=A0 The receiving side&#39;s API/implementation tosses a close code 10=
09 to the sending side, per API.</div><div>=A0 The receiving side&#39;s end=
point receives an error and close notification indicating a protocol closur=
e due to excessive message size from sending side.</div>

<div><br></div><div><br></div></div><div class=3D"gmail_extra"><div class=
=3D""><br clear=3D"all"><div><div dir=3D"ltr"><div>--</div><div>Joakim Erdf=
elt &lt;<a href=3D"mailto:joakim@intalio.com" target=3D"_blank">joakim@inta=
lio.com</a>&gt;</div>
<div>
<a href=3D"http://www.webtide.com/" target=3D"_blank">webtide.com</a>=A0- <=
a href=3D"http://intalio.com/jetty" target=3D"_blank">intalio.com/jetty</a>=
</div><div><span>Expert advice, services and support from=A0</span><span>fr=
om the Jetty &amp; CometD experts</span></div>

<div><a href=3D"http://eclipse.org/jetty/" target=3D"_blank">eclipse.org/je=
tty</a>=A0-=A0<a href=3D"http://cometd.org/" target=3D"_blank">cometd.org</=
a></div></div></div>
<br><br></div><div><div class=3D"h5"><div class=3D"gmail_quote">On Wed, Feb=
 12, 2014 at 9:00 AM, Zhong Yu <span dir=3D"ltr">&lt;<a href=3D"mailto:zhon=
g.j.yu@gmail.com" target=3D"_blank">zhong.j.yu@gmail.com</a>&gt;</span> wro=
te:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
Why not send the message in one frame? The frame length is then the<br>
message length.<br>
<br>
One argument for fragmentation is to insert control frames while<br>
transmitting a long message. However that reasoning is not very<br>
convincing:<br>
<br>
PONG - as long as we are writing data, any data, the peer should see<br>
that the connection is alive. It&#39;s unreasonable for the peer to think<b=
r>
that our end is unresponsive while it is receiving data from us.<br>
<br>
PING - not necessary while we are writing data. If peer is<br>
unresponsive, we&#39;ll learn that from TCP layer.<br>
<br>
CLOSE - closing the connection in the middle of a message is abnormal<br>
anyway, it&#39;s probably not important to do the close handshake.<br>
<span><font color=3D"#888888"><br>
Zhong Yu<br>
</font></span><div><div><br>
<br>
On Fri, Feb 7, 2014 at 1:30 AM, Takeshi Yoshino &lt;<a href=3D"mailto:tyosh=
ino@google.com" target=3D"_blank">tyoshino@google.com</a>&gt; wrote:<br>
&gt; I remember that in early days of the standardization, somebody was pro=
posing<br>
&gt; (*) that we should have a message length field in the first fragment t=
o<br>
&gt; allow for allocating buffer for full message at the beginning.<br>
&gt;<br>
&gt; On Thu, Jan 16, 2014 at 3:27 PM, &quot;Martin J. D=FCrst&quot; &lt;<a =
href=3D"mailto:duerst@it.aoyama.ac.jp" target=3D"_blank">duerst@it.aoyama.a=
c.jp</a>&gt;<br>
&gt; wrote:<br>
&gt;&gt;<br>
&gt;&gt; On 2014/01/16 15:05, Yutaka Hirano wrote:<br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt; Hi Joakim,<br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt; Thanks for the comment.<br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt; Knowing the size of a frame at the start of the frame, traditi=
onal<br>
&gt;&gt;&gt;&gt;<br>
&gt;&gt;&gt;&gt; websocket behavior, is desired, as it allows us to fast fa=
il for the<br>
&gt;&gt;&gt;&gt; 1009<br>
&gt;&gt;&gt;&gt; (message too big) use case.<br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt; Do you know how effective it is?<br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt; Having no prior knowledge of frame size is undesirable to us.<=
br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt; On the other hand, the frame sender have more freedom because =
they don&#39;t<br>
&gt;&gt;&gt; have to know the size of the frame in front.<br>
&gt;&gt;<br>
&gt;&gt;<br>
&gt;&gt; My understanding of why messages are split into frames was that th=
is would<br>
&gt;&gt; allow to start a message without knowing its length. But if you do=
n&#39;t know<br>
&gt;&gt; the size of a frame before sending it, then the right solution is =
to make it<br>
&gt;&gt; smaller. Frames are just arbitrary pieces of a message.<br>
&gt;<br>
&gt;<br>
&gt; We should be clear if we want to make it able to<br>
&gt; (a) know the length of WebSocket frame on receiving the header of each=
<br>
&gt; fragment<br>
&gt; (b) know the length of WebSocket message on receiving the first fragme=
nt<br>
&gt;<br>
&gt; As the proposal (*) has been rejected and not in the RFC6455 header, (=
b) is<br>
&gt; not true even in RFC6455. What Joakim said should be to keep (a) which=
 is<br>
&gt; true now in RFC6455.<br>
&gt;<br>
&gt; But if a message is fragmented into smaller frames, we cannot reject (=
and<br>
&gt; send 1009) the message on receiving the first frame. &quot;Whether we =
can<br>
&gt; fast-reject a big message on receiving some non-final frame&quot; depe=
nds on how<br>
&gt; the peer fragments a message.<br>
&gt;<br>
&gt; In some plans discussed in the main thread, it&#39;s planned to drop f=
rame<br>
&gt; length info for efficiency. But something else such as the length fiel=
d of<br>
&gt; HTTP/2.0 DATA frame tells how much data the fragment (when multiple DA=
TAs<br>
&gt; convey a WebSocket frame, each DATA would be called fragments of a fra=
me)<br>
&gt; contains. So, we won&#39;t be able to know the length of fragments<br>
&gt; (original/encapsulated WebSocket frames) on receiving each of them, bu=
t<br>
&gt; still be able to know the length of sub-fragments (e.g. HTTP/2.0 data<=
br>
&gt; frames).<br>
&gt;<br>
&gt; I think this change is acceptable and shouldn&#39;t be considered as b=
reaking of<br>
&gt; RFC6455 semantics. Even in the mux extension I was editing<br>
&gt; (<a href=3D"http://tools.ietf.org/html/draft-ietf-hybi-websocket-multi=
plexing-11" target=3D"_blank">http://tools.ietf.org/html/draft-ietf-hybi-we=
bsocket-multiplexing-11</a>), I<br>
&gt; made the same change. The length of the encapsulated frame is unknown =
until<br>
&gt; receiving all the fragments of the encapsulating message. No one was<b=
r>
&gt; objecting to that when we were discussing the spec, IIRC.<br>
&gt;<br>
&gt;<br>
</div></div><div><div>&gt; _______________________________________________<=
br>
&gt; hybi mailing list<br>
&gt; <a href=3D"mailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><b=
r>
&gt; <a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blan=
k">https://www.ietf.org/mailman/listinfo/hybi</a><br>
&gt;<br>
<br>
_______________________________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
</div></div></blockquote></div><br></div></div></div>
</blockquote></div><br></div>

--001a1132f7f87c49be04f237f136--


From yhirano@google.com  Wed Feb 12 08:24:59 2014
Return-Path: <yhirano@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id D45231A0325 for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 08:24:59 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.926
X-Spam-Level: 
X-Spam-Status: No, score=-1.926 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id bSLXdQLDAo7k for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 08:24:56 -0800 (PST)
Received: from mail-yh0-x22e.google.com (mail-yh0-x22e.google.com [IPv6:2607:f8b0:4002:c01::22e]) by ietfa.amsl.com (Postfix) with ESMTP id 374A01A0310 for <hybi@ietf.org>; Wed, 12 Feb 2014 08:24:56 -0800 (PST)
Received: by mail-yh0-f46.google.com with SMTP id v1so8700423yhn.19 for <hybi@ietf.org>; Wed, 12 Feb 2014 08:24:55 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=Yg6fIbUB24bG+7926cerjR1ZSp7INnI/3dVK1CdyR+A=; b=fq5gaht7D5lGk6KCmeDkI0gmCH+H+VlZkm7vsRne9jgC78W4lnzi+L/qrAEpNR+otj APsFRs/bqHZCeV4zH6WD0Y2MKah9pqr2adfQlT6HqQB69c2Eew7M/AtCuU8VRyJ+rPxH s0njFVcq9fJKuiWyNzULST+/z91Aq2lZEsrWOwQXc2rzzzCNAIzASbwVzba28vqOga08 RVCO0vSQMA+JqDFOVFjJ6PyzVTV7EI7wEu/HiOQ+4qRI+8BmYEbkj3Be2wqb0DksWqtI rj1TMngOA1Pk99SBC0/aqf/X54XbuDI3bS1QMD6JvLyNypPuyRuxJO141sI9iSBHAAQy Ec5Q==
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=chromium.org; s=google; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=Yg6fIbUB24bG+7926cerjR1ZSp7INnI/3dVK1CdyR+A=; b=Z5NZmX7MvDp26RKmL2dRy6JJLUdBRr6/YNlIZqBVpbzMmG3ajuUQjDNeLk14fs9lqL 1ShWre8Ip7gP2sgXCQ4RfHQZo3pQXu+RZzT7n2fpjle/Hnr2CxVkzumcuCQBT3rY5vzb zwetgQhZAF9jDnigvqt8X70AzT2zVHcVey1vw=
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:sender:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=Yg6fIbUB24bG+7926cerjR1ZSp7INnI/3dVK1CdyR+A=; b=IfTfSah9HBgTgE+/3MuQy6yVH6oc4Hh4cTILqReTAicG5si3oW1NcibKzgPjuU5nXc rOK+JsMfuDnOk1UonyqTGrUWLX/MJcjid1Pv2VtfZMIxucEi93/3X9QcuTDLLlD1RARU ty58tSp0mLongOxCyVFngSOUDRVdK4tg33+kfmQtF5bXFyKSb1ece7eMkr3bxWmS61s7 HuUi8KWlvlqH92vOO13VvhHkB1I7Q7uYIPf6X0LxC1QEmfyFeSn3whaWwDLuTQb0NVXf C/IjLZ2a95iu3tj3fw1SgcO+SHR/lYzLS39MpPwUx00fVcbFDJ4zPZBCIzf7UGFj5FVD T7lQ==
X-Gm-Message-State: ALoCoQk+sUX3KTQF619yzUgfQ/OJDKLg6d0vlj736WUIze0WzaJ1TPgPaS32i+8JwyxcScKnIOTOUCO+klcx4B+SAPX1l9+XRnV6XhQo4DM35XpruAdJx/n0jlh6VplClnzm4TEKv6lqIb+esERP9urkeNoYJpv/d0UiCKTWmSz4a0b/gJWGeGcBF2NSTijrVaNebGWsTR2A
MIME-Version: 1.0
X-Received: by 10.236.131.19 with SMTP id l19mr41122369yhi.0.1392222295155; Wed, 12 Feb 2014 08:24:55 -0800 (PST)
Sender: yhirano@google.com
Received: by 10.170.151.137 with HTTP; Wed, 12 Feb 2014 08:24:55 -0800 (PST)
In-Reply-To: <9766038F-79AF-4CF2-940E-A917D9E7D78F@ericsson.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403DF@EXVMBX020-12.exch020.serverdata.net> <CABihn6H5EtDwhXMYhh-BMb5m6yxL9nYv6=Ud4fN5DeQkjDGyPQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E1@EXVMBX020-12.exch020.serverdata.net> <CABihn6EaE+69q3AJqxEj6vGztBHu4FdOcLCvZ_9Jg-e2cFLwEg@mail.gmail.com> <CAGzyod7ymnpOTiiH7xN3W-aBxD2NNPc=S_Oawdihfhuu84unLw@mail.gmail.com> <634914A010D0B943A035D226786325D4446BED9CDE@EXVMBX020-12.exch020.serverdata.net> <CABihn6EGA3znv_WsJ1gELXyNXgY2fy8weBJ62yD+M2iP7xEnCA@mail.gmail.com> <634914A010D0B943A035D226786325D4446BF99976@EXVMBX020-12.exch020.serverdata.net> <CABihn6GFnGTB9sOZ1=2UnOv-pHPRnRY96DnOb_Q24bNUaxgR=w@mail.gmail.com> <9230B7A2-56AD-4807-A53A-928253364B52@zaphoyd.com> <634914A010D0B943A035D226786325D4446BF99A9B@EXVMBX020-12.exch020.serverdata.net> <F088AE82-E1AF-489B-B220-6E911078C1F3@zaphoyd.com> <6652CC7E-7E92-40EF-BD50-2DDD39E15533@zaphoyd.com> <CABihn6GSrU8oJa4nbMJiDDo2Z+-c4fT69roVFxcHYM4XYFToRQ@mail.gmail.com> <F1F115D4-29EB-457B-B02C-71E46D3E1C81@zaphoyd.com> <CABihn6GXxUwHwLd+926VPEkZxHRZUUi-3dw0PoNaO6aK8X4sEQ@mail.gmail.com> <CABihn6E5h6Xpu2kvZdAAjfp6nKhAqhEwQE2binyai=VC6HkWsw@mail.gmail.com> <FB223CA6-7402-4DF6-905A-1FE2DCBD5BC2@ericsson.com> <CAP+FsNf0R+G6-Wq=Cxg6Ti26tG1MTi1GgZ8X9BHRAvvPbD8JUg@mail.gmail.com> <CABihn6FRSJA85vCHvJUSRtQrvas6TOvFnHWy9JWArp3T6UKVdA@mail.gmail.com> <CAP+FsNcx2LC08n6jKvyiwJQwGc8R7-SELDbZkQrgPAF3BPbo1Q@mail.gmail.com> <CABihn6HVMtHDQwGpSX-ERYSP3Ub1fcuG0gOOCdcndX7hLZZKZQ@mail.gmail.com> <FEBA1A01-4E2C-459F-BB1D-135AFEAD2FAA@ericsson.com> <CABihn6HORhGKppEZvCGHijeR64hS7TQdeG8yPfDQjJyU2JeJtQ@mail.gmail.com> <9766038F-79AF-4CF2-940E-A917D9E7D78F@ericsson.com>
Date: Thu, 13 Feb 2014 01:24:55 +0900
X-Google-Sender-Auth: YA81XjQbkwGURqmXeVBfVUJD6S4
Message-ID: <CABihn6GU8fiWeG+Zy_-i0kBp-YXjLF=N3hPzAU8bo8JAi1+69w@mail.gmail.com>
From: Yutaka Hirano <yhirano@chromium.org>
To: Peter Thorson <webmaster@zaphoyd.com>, Roberto Peon <fenix@google.com>
Content-Type: multipart/alternative; boundary=20cf3010e96d5ac79e04f23803bc
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Discontinuation of mux standardizaton in favor of WS/HTTP/2.0
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 12 Feb 2014 16:25:00 -0000

--20cf3010e96d5ac79e04f23803bc
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

Thank you very much, all!

Roberto gave me comments. Let me share his reply.

> I'm uninterested in requiring all 6455 semantics be replicated. I think o=
f
> frame-level extensions as harmful to the ecosystem. I think we messed up
> there.

I could perhaps be convinced that it it acceptable to do as a compromise,
> but it seems like a kitchen-sink feature that should die.
> Message-level extensions seem reasonable to me.


To Roberto:
I'm not an advocate of frame-level extensions, too.
I once thought a WebSocket connection could be a stream of messages if we
gave up such features, but it was wrong because of the control message
injection, as Takeshi says.
And if we give up RFC6455, we should have two specifications for each
extension, even if it is a message level extension.
Now I think (modified) Plan C is a good plan for you.
It maps WebSocket frame to HTTP/2.0 segment naturally and there is no need
to implement an RFC6455 frame parser, though it isn't widely supported in
this thread yet.

On Fri, Feb 7, 2014 at 4:05 PM, Yutaka Hirano <yhirano@chromium.org> wrote:

> Peter, Roberto, Tobias,
> I think we should shrink the battle line.
> I propose to choose if we preserve the RFC6455 semantics or not (for
> framing).
>
> *Preserve the RFC6455 semantics*
> We won't change the RFC6455 semantics, which means we will discuss how to
> encapsulate WebSocket frame(s) into HTTP/2.0 frame(s).
> All RFC6455 extensions will be supported.
> To preserve the RFC6455 semantics, I modify the proposed plans.
>
> Plan A (WS-TUNNEL with MSG_DONE):
>  - MSG_DONE is used as a FLUSH signal
>  - multiple WebSocket message can be batched in a single HTTP/2.0 frame
> I believe that this preserves the RFC6455 semantics.
> I support this plan.
>
> Plan C (HEADERS + DATAs for a WebSocket frame):
>  - All headers including FIN and length are included in HEADERS.
>  - MSG_DONE is mapped to the end of a **WebSocket frame**
>   - This means that a WebSocket connection is a stream-of-WS_frames, whic=
h
> is completely correct.
> Hence HEADERS + DATAs represent a complete WebSocket frame.
> I support this plan.
> I believe Roberto supports this plan.
>
> Plan D (WS-TUNNEL only): as is
> Tobias and Peter support this plan.
>
> Choosing "Preserve the RFC6455 semantics", means we choose framing method
> from these three plans.
> I know that all of us hate some of the plans, but I think this option is
> better, because
>  - All of us like some of the plans.
>  - We can simplify the discussion.
>  - if "Not preserve the RFC6455 semantics" seems appealing now, at least
> partially that is because it is not concrete.
>
> As the only supporter of Plan A, I promise that I will retract the plan i=
f
> a critical problem is found on it (it is not compatible with RFC6455,
> MSG_DONE can't be used for the flushing purpose, FLUSH can't be introduce=
d
> and so on).
>
> *Not preserve the RFC6455 semantics*
> Choosing this option means we abandon Plan A, C and D.
>
> *Questions*
> As you notice, there are two questions from me.
> 1: Are you OK with deciding NOW if we should preserve the RFC6455
> semantics or not?
> 2: (If the answer for 1 is yes) Should we preserve the RFC6455 semantics?
>
> My answer is, yes & yes.
>
>
>
> On Fri, Feb 7, 2014 at 8:18 AM, Peter Thorson <webmaster@zaphoyd.com>wrot=
e:
>
>>
>> On Feb 6, 2014, at 13:23 , Roberto Peon <fenix@google.com> wrote:
>>
>> > Unfortunately, one can't relax that requirement and still have the
>> backwards-compatible API that exists today: while HTTP/2 can deliver on
>> this, HTTP/1.X cannot.
>> > That one need not change their content to deploy HTTP/2 (etc.) is a
>> major driver of deployment, so it isn't something that could be lightly
>> changed.
>> > IMHO, we'd have to invent a new API (which, honestly I'd be all for,
>> but that is a completely different set of work).
>>
>> This statement gets at the heart of the present discussion.
>>
>> I see the "plan D" option treating WebSocket as HTTP/1.1 is treated by
>> the rest of HTTP/2.0, an existing protocol to be muxed mostly "as is" fo=
r
>> the benefit of backwards compatibility with existing systems. I see a
>> Native Messaging option as the "new API=E2=80=9D that would allow a clea=
ner
>> interface for pure HTTP/2.0 applications going forward.
>>
>> For me, the reality is that any piece of HTTP/2.0 software that I write
>> or deploy in the foreseeable future is going to demux the message stream=
,
>> convert to RFC6455, and forward to existing systems. This is both due to
>> backwards compatibility as well as the nature of a large chunk of my use=
r
>> base (particularly the embedded non-browser clients) having no use for a=
ny
>> of the features that HTTP/2.0 adds over RFC6455 and every reason to want=
 to
>> avoid additional overhead.
>>
>> Plan D provides a straightforward path for this sort of operation and
>> simplifies things considerably for a good chunk of the infrastructure-ty=
pe
>> intermediaries without imposing any additional limitations on anyone (ot=
her
>> than perhaps additional library/protocol dependencies for the more
>> ambitious intermediaries).
>>
>> That said, even if I plan to proxy HTTP/2.0 messages right back into
>> RFC6455. I also see the future benefit to having a cleaner messaging opt=
ion
>> that is aware of HTTP/2.0's new features. My first choice is for this
>> option as it seems more sane long term.. as long as the new protocol is
>> something worth breaking backwards compatibility for. The Plan D case is
>> strong as well and I would not oppose that if the consensus says that
>> backwards compatibility is more important.
>
>
>

--20cf3010e96d5ac79e04f23803bc
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div>Thank you very much, all!</div><div><br></div><div>Ro=
berto gave me comments. Let me share his reply.<br></div><blockquote class=
=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;bo=
rder-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">

<span style=3D"font-family:arial,sans-serif;font-size:13px">I&#39;m uninter=
ested in requiring all 6455 semantics be replicated. I think of frame-level=
 extensions as harmful to the ecosystem. I think we messed up there.</span>=
=C2=A0</blockquote>

<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><span style=3D"font-family:arial,sans-serif;font-size:13px=
"></span>I could perhaps be convinced that it it acceptable to do as a comp=
romise, but it seems like a kitchen-sink feature that should die.<br>

Message-level extensions seem reasonable to me.</blockquote><div><br></div>=
<div>To Roberto:</div><div>I&#39;m not an advocate of frame-level extension=
s, too.</div><div>I once thought a WebSocket connection could be a stream o=
f messages if we gave up such features, but it was wrong because of the con=
trol message injection, as Takeshi says.<br>
</div><div>And if we give up RFC6455, we should have two specifications for=
 each extension, even if it is a message level extension.</div><div>Now I t=
hink (modified) Plan C is a good plan for you.</div><div>It maps WebSocket =
frame to HTTP/2.0 segment naturally and there is no need to implement an RF=
C6455 frame parser, though it isn&#39;t widely supported in this thread yet=
.</div>
<div class=3D"gmail_extra"><br><div class=3D"gmail_quote">On Fri, Feb 7, 20=
14 at 4:05 PM, Yutaka Hirano <span dir=3D"ltr">&lt;<a href=3D"mailto:yhiran=
o@chromium.org" target=3D"_blank">yhirano@chromium.org</a>&gt;</span> wrote=
:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div dir=3D"ltr"><div>Peter, Roberto, Tobias=
,<br></div><div>I think we should shrink the battle line.</div><div>I propo=
se to choose if we preserve the RFC6455 semantics or not (for framing).</di=
v>

<div><br></div><div><b>Preserve the RFC6455 semantics</b></div>

<div>We won&#39;t change the RFC6455 semantics, which means we will discuss=
 how to encapsulate WebSocket frame(s) into HTTP/2.0 frame(s).</div><div>Al=
l RFC6455 extensions will be supported.</div><div>To preserve the RFC6455 s=
emantics, I modify the proposed plans.</div>



<div><br></div><div>Plan A (WS-TUNNEL with MSG_DONE):</div><div>=C2=A0- MSG=
_DONE is used as a FLUSH signal</div><div>=C2=A0- multiple WebSocket messag=
e can be batched in a single HTTP/2.0 frame</div><div>I believe that this p=
reserves the RFC6455 semantics.</div>



<div><div><div>I support this plan.</div></div></div><div><br></div><div>Pl=
an C (HEADERS + DATAs for a WebSocket frame):</div><div>=C2=A0- All headers=
 including FIN and length are included in HEADERS.</div><div>=C2=A0- MSG_DO=
NE is mapped to the end of a **WebSocket frame**</div>



<div>=C2=A0 - This means that a WebSocket connection is a stream-of-WS_fram=
es, which is completely correct.</div><div>Hence HEADERS + DATAs represent =
a complete WebSocket frame.</div><div><div>I support this plan.</div><div>I=
 believe Roberto supports this plan.</div>



<div><br></div></div><div>Plan D (WS-TUNNEL only): as is</div><div>Tobias a=
nd Peter support this plan.</div><div><br></div><div>Choosing &quot;Preserv=
e the RFC6455 semantics&quot;, means we choose framing method from these th=
ree plans.</div>



<div>I know that all of us hate some of the plans, but I think this option =
is better, because</div><div>=C2=A0- All of us like some of the plans.</div=
><div>=C2=A0- We can simplify the discussion.</div><div>=C2=A0- if &quot;No=
t preserve the RFC6455 semantics&quot; seems appealing now, at least partia=
lly that is because it is not concrete.</div>



<div><br></div><div>As the only supporter of Plan A, I promise that I will =
retract the plan if a critical problem is found on it (it is not compatible=
 with RFC6455, MSG_DONE can&#39;t be used for the flushing purpose, FLUSH c=
an&#39;t be introduced and so on).</div>



<div><br></div><div><b>Not preserve the RFC6455 semantics</b><br></div><div=
>Choosing this option means we abandon Plan A, C and D.</div><div><br></div=
><div><b>Questions</b></div><div>As you notice, there are two questions fro=
m me.</div>


<div>1: Are you OK with deciding NOW if we should preserve the RFC6455 sema=
ntics or not?</div>
<div>2: (If the answer for 1 is yes) Should we preserve the RFC6455 semanti=
cs?</div><div><br></div><div>My answer is, yes &amp; yes.</div><div><br></d=
iv></div><div><div><div class=3D"gmail_extra">
<br><br><div class=3D"gmail_quote">On Fri, Feb 7, 2014 at 8:18 AM, Peter Th=
orson <span dir=3D"ltr">&lt;<a href=3D"mailto:webmaster@zaphoyd.com" target=
=3D"_blank">webmaster@zaphoyd.com</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div><br>
On Feb 6, 2014, at 13:23 , Roberto Peon &lt;<a href=3D"mailto:fenix@google.=
com" target=3D"_blank">fenix@google.com</a>&gt; wrote:<br>
<br>
&gt; Unfortunately, one can&#39;t relax that requirement and still have the=
 backwards-compatible API that exists today: while HTTP/2 can deliver on th=
is, HTTP/1.X cannot.<br>
&gt; That one need not change their content to deploy HTTP/2 (etc.) is a ma=
jor driver of deployment, so it isn&#39;t something that could be lightly c=
hanged.<br>
&gt; IMHO, we&#39;d have to invent a new API (which, honestly I&#39;d be al=
l for, but that is a completely different set of work).<br>
<br>
</div>This statement gets at the heart of the present discussion.<br>
<br>
I see the &quot;plan D&quot; option treating WebSocket as HTTP/1.1 is treat=
ed by the rest of HTTP/2.0, an existing protocol to be muxed mostly &quot;a=
s is&quot; for the benefit of backwards compatibility with existing systems=
. I see a Native Messaging option as the &quot;new API=E2=80=9D that would =
allow a cleaner interface for pure HTTP/2.0 applications going forward.<br>



<br>
For me, the reality is that any piece of HTTP/2.0 software that I write or =
deploy in the foreseeable future is going to demux the message stream, conv=
ert to RFC6455, and forward to existing systems. This is both due to backwa=
rds compatibility as well as the nature of a large chunk of my user base (p=
articularly the embedded non-browser clients) having no use for any of the =
features that HTTP/2.0 adds over RFC6455 and every reason to want to avoid =
additional overhead.<br>



<br>
Plan D provides a straightforward path for this sort of operation and simpl=
ifies things considerably for a good chunk of the infrastructure-type inter=
mediaries without imposing any additional limitations on anyone (other than=
 perhaps additional library/protocol dependencies for the more ambitious in=
termediaries).<br>



<br>
That said, even if I plan to proxy HTTP/2.0 messages right back into RFC645=
5. I also see the future benefit to having a cleaner messaging option that =
is aware of HTTP/2.0&#39;s new features. My first choice is for this option=
 as it seems more sane long term.. as long as the new protocol is something=
 worth breaking backwards compatibility for. The Plan D case is strong as w=
ell and I would not oppose that if the consensus says that backwards compat=
ibility is more important.</blockquote>


</div><br></div>
</div></div></blockquote></div><br></div></div>

--20cf3010e96d5ac79e04f23803bc--


From webmaster@zaphoyd.com  Wed Feb 12 08:25:27 2014
Return-Path: <webmaster@zaphoyd.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id AE4281A0447 for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 08:25:27 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.9
X-Spam-Level: 
X-Spam-Status: No, score=-1.9 tagged_above=-999 required=5 tests=[BAYES_00=-1.9] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id ZTLn-8-o3xbR for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 08:25:26 -0800 (PST)
Received: from authsmtp00.uchicago.edu (authsmtp00.uchicago.edu [128.135.12.120]) by ietfa.amsl.com (Postfix) with ESMTP id 2D7331A0310 for <hybi@ietf.org>; Wed, 12 Feb 2014 08:25:26 -0800 (PST)
Received: from [10.0.1.89] (c-68-51-76-178.hsd1.il.comcast.net [68.51.76.178]) (Authenticated sender: zaphoyd) by authsmtp00.uchicago.edu (Postfix) with ESMTP id D0964498088; Wed, 12 Feb 2014 10:25:24 -0600 (CST)
Content-Type: text/plain; charset=windows-1252
Mime-Version: 1.0 (Mac OS X Mail 7.0 \(1822\))
From: Peter Thorson <webmaster@zaphoyd.com>
In-Reply-To: <CACuKZqEcA1Pv8RpWfmThMjTzi2BbVMMKXqujs6BxVfxRPZJ9NQ@mail.gmail.com>
Date: Wed, 12 Feb 2014 10:25:28 -0600
Content-Transfer-Encoding: quoted-printable
Message-Id: <C5F162C5-D2FF-4140-B80B-5BF2DEE80AB1@zaphoyd.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <CACuKZqEcA1Pv8RpWfmThMjTzi2BbVMMKXqujs6BxVfxRPZJ9NQ@mail.gmail.com>
To: Zhong Yu <zhong.j.yu@gmail.com>
X-Mailer: Apple Mail (2.1822)
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 12 Feb 2014 16:25:27 -0000

On Feb 12, 2014, at 10:00, Zhong Yu <zhong.j.yu@gmail.com> wrote:

> Why not send the message in one frame? The frame length is then the
> message length.

The message length might not be known or might not be fully buffer-able =
by the sending endpoint.

> One argument for fragmentation is to insert control frames while
> transmitting a long message. However that reasoning is not very
> convincing:
>=20
> PONG - as long as we are writing data, any data, the peer should see
> that the connection is alive. It's unreasonable for the peer to think
> that our end is unresponsive while it is receiving data from us.
>=20
> PING - not necessary while we are writing data. If peer is
> unresponsive, we'll learn that from TCP layer.

It does provide some separation of concerns and the ability to do things =
like measure RTT during a long message.

> CLOSE - closing the connection in the middle of a message is abnormal
> anyway, it's probably not important to do the close handshake.

It can be used to cleanly close a connection after a message has been =
sent but before it is completed. Say a client initiates a transfer then =
decides to leave halfway through (browser tab closes, etc). The client =
can indicate that it wants to close and that the in flight message =
should be discarded. Otherwise the client would have to either truncate =
the message (which might result in bogus data getting processed by the =
server) or just close the TCP connection (which can leave the connection =
in a bad state on the server and causes the server to have to report an =
unclean close to its application, which isn=92t entirely accurate).=


From zhong.j.yu@gmail.com  Wed Feb 12 08:31:17 2014
Return-Path: <zhong.j.yu@gmail.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 9E6281A03B4 for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 08:31:17 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2
X-Spam-Level: 
X-Spam-Status: No, score=-2 tagged_above=-999 required=5 tests=[BAYES_00=-1.9,  DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FREEMAIL_FROM=0.001, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id jmGQZudMW883 for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 08:31:15 -0800 (PST)
Received: from mail-oa0-x232.google.com (mail-oa0-x232.google.com [IPv6:2607:f8b0:4003:c02::232]) by ietfa.amsl.com (Postfix) with ESMTP id 7B3A91A0325 for <hybi@ietf.org>; Wed, 12 Feb 2014 08:31:15 -0800 (PST)
Received: by mail-oa0-f50.google.com with SMTP id n16so11132949oag.23 for <hybi@ietf.org>; Wed, 12 Feb 2014 08:31:14 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type:content-transfer-encoding; bh=uEZkqqqaOEOg485otGMXqy1rEgUv9Gby4VIq4kLQhs4=; b=ljEJdxYkrHdfCS18wrDTtoEOUVSFCdZXRnd2Dgw+7qBfph4JghG7NLURXsN40SMSaL CDj+nQjPzmJECBXg5zO/phs7G6rhNCXcWlZ6I/ppbv0EpYlgruh0AALRaUOEEqOBWT+J y6adZKzzL2ZZ092Lssz6IQjvMQKvhpwSRwmkC4XFeBwey52946X5pkw48t8gPEQCEL2E ZmUizCgsrc8sxrG/9mq3x84os2jDynk0IWPX3m5D/pPvAuoWZ627iLyG1nTEUO1GuFmp ve7Ux8BF8+/8KCgEcmE/4Q99WKAyDDUw2UnYLNxZ9KURdCWAkyGWW2xY6bRvrgB/QIYB Iitg==
MIME-Version: 1.0
X-Received: by 10.182.142.229 with SMTP id rz5mr37925328obb.12.1392222674457;  Wed, 12 Feb 2014 08:31:14 -0800 (PST)
Received: by 10.76.21.132 with HTTP; Wed, 12 Feb 2014 08:31:14 -0800 (PST)
In-Reply-To: <C5F162C5-D2FF-4140-B80B-5BF2DEE80AB1@zaphoyd.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <CACuKZqEcA1Pv8RpWfmThMjTzi2BbVMMKXqujs6BxVfxRPZJ9NQ@mail.gmail.com> <C5F162C5-D2FF-4140-B80B-5BF2DEE80AB1@zaphoyd.com>
Date: Wed, 12 Feb 2014 10:31:14 -0600
Message-ID: <CACuKZqGMUs5nOEQZvhMrnqTtnPyTTRMG1TvVqQ8O95P_ecnNQw@mail.gmail.com>
From: Zhong Yu <zhong.j.yu@gmail.com>
To: Peter Thorson <webmaster@zaphoyd.com>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 12 Feb 2014 16:31:17 -0000

On Wed, Feb 12, 2014 at 10:25 AM, Peter Thorson <webmaster@zaphoyd.com> wro=
te:
>
> On Feb 12, 2014, at 10:00, Zhong Yu <zhong.j.yu@gmail.com> wrote:
>
>> Why not send the message in one frame? The frame length is then the
>> message length.
>
> The message length might not be known or might not be fully buffer-able b=
y the sending endpoint.
>
>> One argument for fragmentation is to insert control frames while
>> transmitting a long message. However that reasoning is not very
>> convincing:
>>
>> PONG - as long as we are writing data, any data, the peer should see
>> that the connection is alive. It's unreasonable for the peer to think
>> that our end is unresponsive while it is receiving data from us.
>>
>> PING - not necessary while we are writing data. If peer is
>> unresponsive, we'll learn that from TCP layer.
>
> It does provide some separation of concerns and the ability to do things =
like measure RTT during a long message.
>
>> CLOSE - closing the connection in the middle of a message is abnormal
>> anyway, it's probably not important to do the close handshake.
>
> It can be used to cleanly close a connection after a message has been sen=
t but before it is completed. Say a client initiates a transfer then decide=
s to leave halfway through (browser tab closes, etc). The client can indica=
te that it wants to close and that the in flight message should be discarde=
d. Otherwise the client would have to either truncate the message (which mi=
ght result in bogus data getting processed by the server) or just close the=
 TCP connection (which can leave the connection in a bad state on the serve=
r and causes the server to have to report an unclean close to its applicati=
on, which isn't entirely accurate).

I don't quite buy this argument. Browsers initial TCP FIN all the
time, to shutdown persistent HTTP connections for example.


From zhong.j.yu@gmail.com  Wed Feb 12 08:37:10 2014
Return-Path: <zhong.j.yu@gmail.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 7F4D61A02A7 for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 08:37:10 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2
X-Spam-Level: 
X-Spam-Status: No, score=-2 tagged_above=-999 required=5 tests=[BAYES_00=-1.9,  DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FREEMAIL_FROM=0.001, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id t-8j9K8Xz3xO for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 08:37:05 -0800 (PST)
Received: from mail-ob0-x233.google.com (mail-ob0-x233.google.com [IPv6:2607:f8b0:4003:c01::233]) by ietfa.amsl.com (Postfix) with ESMTP id 8AA9C1A0320 for <hybi@ietf.org>; Wed, 12 Feb 2014 08:37:05 -0800 (PST)
Received: by mail-ob0-f179.google.com with SMTP id wo20so10671579obc.24 for <hybi@ietf.org>; Wed, 12 Feb 2014 08:37:04 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type:content-transfer-encoding; bh=7292DR3vwYQPWSOJrx6M5njdEco1ZjDEoVLVeFbj8VM=; b=aso+fbm4V6R0OH0YLtR8u15Y0pYrHqiSQ58Q6+JOPdmdFkV7JAesBj9jNu5qhtCbZh jZkpIDHW76b5akgkDXzypqvJ6twYqP6gDMU/RQUsAgSvwYCtWbM5S+n4k0n7hTetwlJj 54MSquv6aRwXaOpfCSsflo7Qcy9/NLcUmn8werJC8yfxbar89QpcQqcf5G+nNBp65wi5 +Ncnw3qQ694U0OUqo2c9GNESkmRQt93Ttho4RsYWn965R3EXQFJF/o2z02jWtSVej3x5 ZiC45RzxlSynCMsBFosDbiYfITLJ64jcSksqII19YyKB46jpGutInMl62CKivkybBOhB cZXg==
MIME-Version: 1.0
X-Received: by 10.60.229.4 with SMTP id sm4mr38190081oec.9.1392223024637; Wed, 12 Feb 2014 08:37:04 -0800 (PST)
Received: by 10.76.21.132 with HTTP; Wed, 12 Feb 2014 08:37:04 -0800 (PST)
In-Reply-To: <CAG4zZZCr4aTfVpw2coX2g0qw++4kdgNCFVze6tHKZ+fJNqb0aQ@mail.gmail.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <CACuKZqEcA1Pv8RpWfmThMjTzi2BbVMMKXqujs6BxVfxRPZJ9NQ@mail.gmail.com> <CAG4zZZCr4aTfVpw2coX2g0qw++4kdgNCFVze6tHKZ+fJNqb0aQ@mail.gmail.com>
Date: Wed, 12 Feb 2014 10:37:04 -0600
Message-ID: <CACuKZqGNm35jWjND0bPJ6BQK48LBkyj2MR01hX1k1sP+mMyqMQ@mail.gmail.com>
From: Zhong Yu <zhong.j.yu@gmail.com>
To: Joakim Erdfelt <joakim@intalio.com>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 12 Feb 2014 16:37:10 -0000

The sending side does not need to respond with a CLOSE in this
scenario. It is in the middle of sending a huge message in one huge
frame, while it gets a message from the peer saying that the peer is
no longer interested in the message. The sending side can simply drop
the TCP connection here.

On Wed, Feb 12, 2014 at 10:13 AM, Joakim Erdfelt <joakim@intalio.com> wrote=
:
> Scenario:
>   Sending side uses a message stream based API to write out 1 websocket
> message.
>   The sending side's API is configured/implemented to buffer only 16k at =
a
> time, writing out a frame of 16k for each buffer filled.
>   The sending side writes out a 600k message.
>   The receiving side is using a different API, one that is based on compl=
ete
> messages.
>   The receiving side is configured to not accept messages over 100k.
>   The receiving side's API/implementation tosses a close code 1009 to the
> sending side, per API.
>   The receiving side's endpoint receives an error and close notification
> indicating a protocol closure due to excessive message size from sending
> side.
>
>
>
> --
> Joakim Erdfelt <joakim@intalio.com>
> webtide.com - intalio.com/jetty
> Expert advice, services and support from from the Jetty & CometD experts
> eclipse.org/jetty - cometd.org
>
>
> On Wed, Feb 12, 2014 at 9:00 AM, Zhong Yu <zhong.j.yu@gmail.com> wrote:
>>
>> Why not send the message in one frame? The frame length is then the
>> message length.
>>
>> One argument for fragmentation is to insert control frames while
>> transmitting a long message. However that reasoning is not very
>> convincing:
>>
>> PONG - as long as we are writing data, any data, the peer should see
>> that the connection is alive. It's unreasonable for the peer to think
>> that our end is unresponsive while it is receiving data from us.
>>
>> PING - not necessary while we are writing data. If peer is
>> unresponsive, we'll learn that from TCP layer.
>>
>> CLOSE - closing the connection in the middle of a message is abnormal
>> anyway, it's probably not important to do the close handshake.
>>
>> Zhong Yu
>>
>>
>> On Fri, Feb 7, 2014 at 1:30 AM, Takeshi Yoshino <tyoshino@google.com>
>> wrote:
>> > I remember that in early days of the standardization, somebody was
>> > proposing
>> > (*) that we should have a message length field in the first fragment t=
o
>> > allow for allocating buffer for full message at the beginning.
>> >
>> > On Thu, Jan 16, 2014 at 3:27 PM, "Martin J. D=FCrst"
>> > <duerst@it.aoyama.ac.jp>
>> > wrote:
>> >>
>> >> On 2014/01/16 15:05, Yutaka Hirano wrote:
>> >>>
>> >>> Hi Joakim,
>> >>>
>> >>> Thanks for the comment.
>> >>>
>> >>> Knowing the size of a frame at the start of the frame, traditional
>> >>>>
>> >>>> websocket behavior, is desired, as it allows us to fast fail for th=
e
>> >>>> 1009
>> >>>> (message too big) use case.
>> >>>
>> >>>
>> >>> Do you know how effective it is?
>> >>>
>> >>> Having no prior knowledge of frame size is undesirable to us.
>> >>>
>> >>> On the other hand, the frame sender have more freedom because they
>> >>> don't
>> >>> have to know the size of the frame in front.
>> >>
>> >>
>> >> My understanding of why messages are split into frames was that this
>> >> would
>> >> allow to start a message without knowing its length. But if you don't
>> >> know
>> >> the size of a frame before sending it, then the right solution is to
>> >> make it
>> >> smaller. Frames are just arbitrary pieces of a message.
>> >
>> >
>> > We should be clear if we want to make it able to
>> > (a) know the length of WebSocket frame on receiving the header of each
>> > fragment
>> > (b) know the length of WebSocket message on receiving the first fragme=
nt
>> >
>> > As the proposal (*) has been rejected and not in the RFC6455 header, (=
b)
>> > is
>> > not true even in RFC6455. What Joakim said should be to keep (a) which
>> > is
>> > true now in RFC6455.
>> >
>> > But if a message is fragmented into smaller frames, we cannot reject
>> > (and
>> > send 1009) the message on receiving the first frame. "Whether we can
>> > fast-reject a big message on receiving some non-final frame" depends o=
n
>> > how
>> > the peer fragments a message.
>> >
>> > In some plans discussed in the main thread, it's planned to drop frame
>> > length info for efficiency. But something else such as the length fiel=
d
>> > of
>> > HTTP/2.0 DATA frame tells how much data the fragment (when multiple
>> > DATAs
>> > convey a WebSocket frame, each DATA would be called fragments of a
>> > frame)
>> > contains. So, we won't be able to know the length of fragments
>> > (original/encapsulated WebSocket frames) on receiving each of them, bu=
t
>> > still be able to know the length of sub-fragments (e.g. HTTP/2.0 data
>> > frames).
>> >
>> > I think this change is acceptable and shouldn't be considered as
>> > breaking of
>> > RFC6455 semantics. Even in the mux extension I was editing
>> > (http://tools.ietf.org/html/draft-ietf-hybi-websocket-multiplexing-11)=
,
>> > I
>> > made the same change. The length of the encapsulated frame is unknown
>> > until
>> > receiving all the fragments of the encapsulating message. No one was
>> > objecting to that when we were discussing the spec, IIRC.
>> >
>> >
>> > _______________________________________________
>> > hybi mailing list
>> > hybi@ietf.org
>> > https://www.ietf.org/mailman/listinfo/hybi
>> >
>>
>> _______________________________________________
>> hybi mailing list
>> hybi@ietf.org
>> https://www.ietf.org/mailman/listinfo/hybi
>
>


From tobias.oberstein@tavendo.de  Wed Feb 12 08:41:59 2014
Return-Path: <tobias.oberstein@tavendo.de>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 52BBA1A0338 for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 08:41:59 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.9
X-Spam-Level: 
X-Spam-Status: No, score=-1.9 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, RCVD_IN_DNSWL_NONE=-0.0001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id RO3ju_m0ok-C for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 08:41:55 -0800 (PST)
Received: from EXHUB020-1.exch020.serverdata.net (exhub020-1.exch020.serverdata.net [206.225.164.28]) by ietfa.amsl.com (Postfix) with ESMTP id AD3DF1A0310 for <hybi@ietf.org>; Wed, 12 Feb 2014 08:41:55 -0800 (PST)
Received: from EXVMBX020-12.exch020.serverdata.net ([169.254.3.11]) by EXHUB020-1.exch020.serverdata.net ([206.225.164.28]) with mapi; Wed, 12 Feb 2014 08:41:54 -0800
From: Tobias Oberstein <tobias.oberstein@tavendo.de>
To: Zhong Yu <zhong.j.yu@gmail.com>, Joakim Erdfelt <joakim@intalio.com>
Date: Wed, 12 Feb 2014 08:41:46 -0800
Thread-Topic: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
Thread-Index: Ac8oDO7orj6CendaQiavUV2j3bdINQAA9zeg
Message-ID: <634914A010D0B943A035D226786325D4446C44FD11@EXVMBX020-12.exch020.serverdata.net>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <8984EA8D-C6ED-49BF-92C4-113D69121B30@brandedcode.com> <CAG4zZZAfB_phXSVku+W+VbGRNMJ3+=9YPw8_dD6zSJr5041iEw@mail.gmail.com> <CACuKZqF+XZs0y04Svxb=rKhFEydwxp=8pwTiaL+3sm6cXF7_5Q@mail.gmail.com>
In-Reply-To: <CACuKZqF+XZs0y04Svxb=rKhFEydwxp=8pwTiaL+3sm6cXF7_5Q@mail.gmail.com>
Accept-Language: de-DE, en-US
Content-Language: de-DE
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
acceptlanguage: de-DE, en-US
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 12 Feb 2014 16:41:59 -0000

> Why does it matter if a frame (of unknown size) is composed of fragments
> (of known sizes), w.r.t. memory management?

I agree with Joakim that frame size length can be important for memory mana=
gement.

It is also important to be able to fragment a message, for reasons of

- control frame injection (ordered close)
- application data independent keep alive (ping/pong)
- message interleaving

/Tobias

>=20
> On Wed, Feb 12, 2014 at 7:49 AM, Joakim Erdfelt <joakim@intalio.com>
> wrote:
> > For various application level APIs that depend on delivery of complete
> > messages, the need to know the frame level size at the start of the
> > frame becomes important for memory management.  Eclipse Jetty uses
> > this information to implement Java's javax.websocket API for limiting
> > memory use, along with lower level (WS+HTTP/1.1) protocol validation
> (close code 1009).
> >
> > Since this group has a history (now) of making protocol decisions
> > based on existing public APIs (re Javascript WebSocket), I would
> > entreat the members here to think about the other public, industry
> > standard, non-javascript, WebSocket APIs already widely deployed.
> >
> >
> > --
> > Joakim Erdfelt <joakim@intalio.com>
> > webtide.com - intalio.com/jetty
> > Expert advice, services and support from from the Jetty & CometD
> > experts eclipse.org/jetty - cometd.org
> >
> >
> > On Wed, Feb 12, 2014 at 6:34 AM, Micheil Smith
> > <micheil@brandedcode.com>
> > wrote:
> >>
> >> As far as I recall, this was also to enable more efficient reading
> >> and error checking.
> >> Prefixing with the frame size means you don't need to scan every byte
> >> being received in order to find an end of frame delimiter.
> >>
> >> This is a particularly useful thing when doing something like
> >> upstreaming on a websocket server, where all you care about is the
> >> very start of the message, and the length, such that you can then
> >> just read() N bytes which is the size of your frame and just write()
> >> directly.
> >>
> >> In other words, I agree on your reply Takeshi.
> >>
> >> - Micheil
> >>
> >> On 7 Feb 2014, at 7:30 am, Takeshi Yoshino <tyoshino@google.com>
> wrote:
> >>
> >> > I remember that in early days of the standardization, somebody was
> >> > proposing (*) that we should have a message length field in the
> >> > first fragment to allow for allocating buffer for full message at th=
e
> beginning.
> >> >
> >> > On Thu, Jan 16, 2014 at 3:27 PM, "Martin J. D=FCrst"
> >> > <duerst@it.aoyama.ac.jp> wrote:
> >> > On 2014/01/16 15:05, Yutaka Hirano wrote:
> >> > Hi Joakim,
> >> >
> >> > Thanks for the comment.
> >> >
> >> > Knowing the size of a frame at the start of the frame, traditional
> >> > websocket behavior, is desired, as it allows us to fast fail for
> >> > the
> >> > 1009
> >> > (message too big) use case.
> >> >
> >> > Do you know how effective it is?
> >> >
> >> > Having no prior knowledge of frame size is undesirable to us.
> >> >
> >> > On the other hand, the frame sender have more freedom because they
> >> > don't have to know the size of the frame in front.
> >> >
> >> > My understanding of why messages are split into frames was that
> >> > this would allow to start a message without knowing its length. But
> >> > if you don't know the size of a frame before sending it, then the
> >> > right solution is to make it smaller. Frames are just arbitrary piec=
es of a
> message.
> >> >
> >> > We should be clear if we want to make it able to
> >> > (a) know the length of WebSocket frame on receiving the header of
> >> > each fragment
> >> > (b) know the length of WebSocket message on receiving the first
> >> > fragment
> >> >
> >> > As the proposal (*) has been rejected and not in the RFC6455
> >> > header, (b) is not true even in RFC6455. What Joakim said should be
> >> > to keep (a) which is true now in RFC6455.
> >> >
> >> > But if a message is fragmented into smaller frames, we cannot
> >> > reject (and send 1009) the message on receiving the first frame.
> >> > "Whether we can fast-reject a big message on receiving some
> >> > non-final frame" depends on how the peer fragments a message.
> >> >
> >> > In some plans discussed in the main thread, it's planned to drop
> >> > frame length info for efficiency. But something else such as the
> >> > length field of
> >> > HTTP/2.0 DATA frame tells how much data the fragment (when multiple
> >> > DATAs convey a WebSocket frame, each DATA would be called
> fragments
> >> > of a frame) contains. So, we won't be able to know the length of
> >> > fragments (original/encapsulated WebSocket frames) on receiving
> >> > each of them, but still be able to know the length of sub-fragments
> >> > (e.g. HTTP/2.0 data frames).
> >> >
> >> > I think this change is acceptable and shouldn't be considered as
> >> > breaking of RFC6455 semantics. Even in the mux extension I was
> >> > editing
> >> > (http://tools.ietf.org/html/draft-ietf-hybi-websocket-multiplexing-
> >> > 11), I made the same change. The length of the encapsulated frame
> >> > is unknown until receiving all the fragments of the encapsulating
> message. No one was objecting to that when we were discussing the spec,
> IIRC.
> >> >
> >> > _______________________________________________
> >> > hybi mailing list
> >> > hybi@ietf.org
> >> > https://www.ietf.org/mailman/listinfo/hybi
> >>
> >> _______________________________________________
> >> hybi mailing list
> >> hybi@ietf.org
> >> https://www.ietf.org/mailman/listinfo/hybi
> >
> >
> >
> > _______________________________________________
> > hybi mailing list
> > hybi@ietf.org
> > https://www.ietf.org/mailman/listinfo/hybi
> >
>=20
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi


From zhong.j.yu@gmail.com  Wed Feb 12 08:52:34 2014
Return-Path: <zhong.j.yu@gmail.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 307431A05D2 for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 08:52:34 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2
X-Spam-Level: 
X-Spam-Status: No, score=-2 tagged_above=-999 required=5 tests=[BAYES_00=-1.9,  DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FREEMAIL_FROM=0.001, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id hUUpQz34ueTu for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 08:52:31 -0800 (PST)
Received: from mail-ob0-x22d.google.com (mail-ob0-x22d.google.com [IPv6:2607:f8b0:4003:c01::22d]) by ietfa.amsl.com (Postfix) with ESMTP id 275141A09B0 for <hybi@ietf.org>; Wed, 12 Feb 2014 08:52:31 -0800 (PST)
Received: by mail-ob0-f173.google.com with SMTP id vb8so10726969obc.4 for <hybi@ietf.org>; Wed, 12 Feb 2014 08:52:30 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type:content-transfer-encoding; bh=BTMYEKbc5HouLyAVbN6lMe2smh4SCCS4GUjXN+MkJUo=; b=oWwiPnqJNvLb/lvDml+8KaPIFWca8HxQuWnLNrlOSB7n0IgRYrUC0Z5X3Kila1ShKv K9Bk8rBeE9Om8r0lcAy59Edy1UyHaGGG92q/X4fsO9YUmZYdyIFqQyJr5JAxxz4UZDvY 1luBzHn2Q2xh/7KImh9WzJG9Bq0yR0DuFwHapL89QxVOSVhle1WzTDKpcK3TieISCUPz ZJ9AkgN06suoNRZSI6YibhCgl2VhZzIc5u6bSdzanWUjT3+Ch1xAu/1lJ8lPBKSt7fq8 va9/S2jJbcv/y44e7MPFl/5wSYo4dQoJ2CqHnSX5Svtk821gq+98QUJUPxXJugOdWmfe TTrA==
MIME-Version: 1.0
X-Received: by 10.60.228.135 with SMTP id si7mr37455615oec.4.1392223950118; Wed, 12 Feb 2014 08:52:30 -0800 (PST)
Received: by 10.76.21.132 with HTTP; Wed, 12 Feb 2014 08:52:30 -0800 (PST)
In-Reply-To: <634914A010D0B943A035D226786325D4446C44FD11@EXVMBX020-12.exch020.serverdata.net>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <8984EA8D-C6ED-49BF-92C4-113D69121B30@brandedcode.com> <CAG4zZZAfB_phXSVku+W+VbGRNMJ3+=9YPw8_dD6zSJr5041iEw@mail.gmail.com> <CACuKZqF+XZs0y04Svxb=rKhFEydwxp=8pwTiaL+3sm6cXF7_5Q@mail.gmail.com> <634914A010D0B943A035D226786325D4446C44FD11@EXVMBX020-12.exch020.serverdata.net>
Date: Wed, 12 Feb 2014 10:52:30 -0600
Message-ID: <CACuKZqFbL5vv3Kn-m=8NO8dtGvTS7uteGTDy_h9ReUFszR4Nhg@mail.gmail.com>
From: Zhong Yu <zhong.j.yu@gmail.com>
To: Tobias Oberstein <tobias.oberstein@tavendo.de>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 12 Feb 2014 16:52:34 -0000

On Wed, Feb 12, 2014 at 10:41 AM, Tobias Oberstein
<tobias.oberstein@tavendo.de> wrote:
>> Why does it matter if a frame (of unknown size) is composed of fragments
>> (of known sizes), w.r.t. memory management?
>
> I agree with Joakim that frame size length can be important for memory ma=
nagement.
>
> It is also important to be able to fragment a message, for reasons of
>
> - control frame injection (ordered close)
> - application data independent keep alive (ping/pong)
> - message interleaving

Fragmentation is certainly necessary for mux. But if we are talking
about a simple RFC6455 connection, it seems to me that it's ok to send
a message (of known size) in one frame, regardless of its size.

>
> /Tobias
>
>>
>> On Wed, Feb 12, 2014 at 7:49 AM, Joakim Erdfelt <joakim@intalio.com>
>> wrote:
>> > For various application level APIs that depend on delivery of complete
>> > messages, the need to know the frame level size at the start of the
>> > frame becomes important for memory management.  Eclipse Jetty uses
>> > this information to implement Java's javax.websocket API for limiting
>> > memory use, along with lower level (WS+HTTP/1.1) protocol validation
>> (close code 1009).
>> >
>> > Since this group has a history (now) of making protocol decisions
>> > based on existing public APIs (re Javascript WebSocket), I would
>> > entreat the members here to think about the other public, industry
>> > standard, non-javascript, WebSocket APIs already widely deployed.
>> >
>> >
>> > --
>> > Joakim Erdfelt <joakim@intalio.com>
>> > webtide.com - intalio.com/jetty
>> > Expert advice, services and support from from the Jetty & CometD
>> > experts eclipse.org/jetty - cometd.org
>> >
>> >
>> > On Wed, Feb 12, 2014 at 6:34 AM, Micheil Smith
>> > <micheil@brandedcode.com>
>> > wrote:
>> >>
>> >> As far as I recall, this was also to enable more efficient reading
>> >> and error checking.
>> >> Prefixing with the frame size means you don't need to scan every byte
>> >> being received in order to find an end of frame delimiter.
>> >>
>> >> This is a particularly useful thing when doing something like
>> >> upstreaming on a websocket server, where all you care about is the
>> >> very start of the message, and the length, such that you can then
>> >> just read() N bytes which is the size of your frame and just write()
>> >> directly.
>> >>
>> >> In other words, I agree on your reply Takeshi.
>> >>
>> >> - Micheil
>> >>
>> >> On 7 Feb 2014, at 7:30 am, Takeshi Yoshino <tyoshino@google.com>
>> wrote:
>> >>
>> >> > I remember that in early days of the standardization, somebody was
>> >> > proposing (*) that we should have a message length field in the
>> >> > first fragment to allow for allocating buffer for full message at t=
he
>> beginning.
>> >> >
>> >> > On Thu, Jan 16, 2014 at 3:27 PM, "Martin J. D=FCrst"
>> >> > <duerst@it.aoyama.ac.jp> wrote:
>> >> > On 2014/01/16 15:05, Yutaka Hirano wrote:
>> >> > Hi Joakim,
>> >> >
>> >> > Thanks for the comment.
>> >> >
>> >> > Knowing the size of a frame at the start of the frame, traditional
>> >> > websocket behavior, is desired, as it allows us to fast fail for
>> >> > the
>> >> > 1009
>> >> > (message too big) use case.
>> >> >
>> >> > Do you know how effective it is?
>> >> >
>> >> > Having no prior knowledge of frame size is undesirable to us.
>> >> >
>> >> > On the other hand, the frame sender have more freedom because they
>> >> > don't have to know the size of the frame in front.
>> >> >
>> >> > My understanding of why messages are split into frames was that
>> >> > this would allow to start a message without knowing its length. But
>> >> > if you don't know the size of a frame before sending it, then the
>> >> > right solution is to make it smaller. Frames are just arbitrary pie=
ces of a
>> message.
>> >> >
>> >> > We should be clear if we want to make it able to
>> >> > (a) know the length of WebSocket frame on receiving the header of
>> >> > each fragment
>> >> > (b) know the length of WebSocket message on receiving the first
>> >> > fragment
>> >> >
>> >> > As the proposal (*) has been rejected and not in the RFC6455
>> >> > header, (b) is not true even in RFC6455. What Joakim said should be
>> >> > to keep (a) which is true now in RFC6455.
>> >> >
>> >> > But if a message is fragmented into smaller frames, we cannot
>> >> > reject (and send 1009) the message on receiving the first frame.
>> >> > "Whether we can fast-reject a big message on receiving some
>> >> > non-final frame" depends on how the peer fragments a message.
>> >> >
>> >> > In some plans discussed in the main thread, it's planned to drop
>> >> > frame length info for efficiency. But something else such as the
>> >> > length field of
>> >> > HTTP/2.0 DATA frame tells how much data the fragment (when multiple
>> >> > DATAs convey a WebSocket frame, each DATA would be called
>> fragments
>> >> > of a frame) contains. So, we won't be able to know the length of
>> >> > fragments (original/encapsulated WebSocket frames) on receiving
>> >> > each of them, but still be able to know the length of sub-fragments
>> >> > (e.g. HTTP/2.0 data frames).
>> >> >
>> >> > I think this change is acceptable and shouldn't be considered as
>> >> > breaking of RFC6455 semantics. Even in the mux extension I was
>> >> > editing
>> >> > (http://tools.ietf.org/html/draft-ietf-hybi-websocket-multiplexing-
>> >> > 11), I made the same change. The length of the encapsulated frame
>> >> > is unknown until receiving all the fragments of the encapsulating
>> message. No one was objecting to that when we were discussing the spec,
>> IIRC.
>> >> >
>> >> > _______________________________________________
>> >> > hybi mailing list
>> >> > hybi@ietf.org
>> >> > https://www.ietf.org/mailman/listinfo/hybi
>> >>
>> >> _______________________________________________
>> >> hybi mailing list
>> >> hybi@ietf.org
>> >> https://www.ietf.org/mailman/listinfo/hybi
>> >
>> >
>> >
>> > _______________________________________________
>> > hybi mailing list
>> > hybi@ietf.org
>> > https://www.ietf.org/mailman/listinfo/hybi
>> >
>>
>> _______________________________________________
>> hybi mailing list
>> hybi@ietf.org
>> https://www.ietf.org/mailman/listinfo/hybi


From fenix@google.com  Wed Feb 12 09:08:55 2014
Return-Path: <fenix@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 1A0341A09D4 for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 09:08:55 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.326
X-Spam-Level: 
X-Spam-Status: No, score=-1.326 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, J_CHICKENPOX_27=0.6, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id dtZYYRkxaJAo for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 09:08:52 -0800 (PST)
Received: from mail-qc0-x22b.google.com (mail-qc0-x22b.google.com [IPv6:2607:f8b0:400d:c01::22b]) by ietfa.amsl.com (Postfix) with ESMTP id 0111C1A0974 for <hybi@ietf.org>; Wed, 12 Feb 2014 09:08:51 -0800 (PST)
Received: by mail-qc0-f171.google.com with SMTP id n7so15778659qcx.2 for <hybi@ietf.org>; Wed, 12 Feb 2014 09:08:50 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=CubRpcEIT/Zr1e5Y7hsy4pPAW2q9i2zdTuC/V3h7fM0=; b=mhsneC57kF+qhGc5y1Qf9l3hZULQN2Jb7Zpz6PP1eezAfjO2M+DTgQSBm/KfHImrV7 n4XcyNdfgH1aWu/JuARIq4sESo3iBYI6eSzKFC8Zk5F1Lsgt/v6jczWzrqM0TOrMqlFD JTJU4Aq2I4HRFtH/lT2GwZgrCdJ0cBtoBTGK37yTclIHB9e9atz9Tz8lfhuz/9AbQjIM Ch7MMoeQsPMCewtgVXmxp801EWtYsWKHh6Q+CJ6h9YawTpHVv5UkOrjbt1f/sRB7uZTL TyXMF4pOSAx0DlqFRH5u+zYEWQ+aESZEH4xHbz2LfH/h9a2f38s381k97DK9d2JWgiBv pVzw==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=CubRpcEIT/Zr1e5Y7hsy4pPAW2q9i2zdTuC/V3h7fM0=; b=aeA4iuRl5qiNEqGAoOdS+Zysq6muvElPHZTJI7ypytTyUON/4YYHsjvbUzSTi3O0ri XeDRAJEL4s1aoCimIaurA6EI2pTmAecW9iX+gZTzaNwN172MjNrn+CKW5SamJRAXJsBO /PP1ENV3qSWJ16rh7RhXyp128DvgdpGnm+FOAAcnedyDj6KgSbU9HIhlQ1qos9eIdhE8 MkQcpG/btYGCH+mj8rVUCArGoK1KaDYPPRBzjRx3E/N0/WUrVKW+a6iRG6Jmn75Pv1So LCxqqEa61YZGv4xn3Yeb+7k6WjvM6wQ7fKoEnyWyJGcGguSUMwsUmAOzZCYEAlOynTDC gWQA==
X-Gm-Message-State: ALoCoQmgIeLFeLCMrBtPQgwkv6WDLBDNwDHa0mGrZ45sH70ltUh5twrFVMA3sFVG+kojf89b3N5dUQT6VggvWs/KlV9vnregEr7gv+5ynaN5ETxKYaw2W3dc3qbYb8+65xLOLUzD5fDfrn+hvWpTHTAgHLo07qeqVNnBSdKtFZOD1KwsF9I1Odd7zU9abgc0edCzEEJztkq/
MIME-Version: 1.0
X-Received: by 10.224.127.129 with SMTP id g1mr34582620qas.22.1392224930735; Wed, 12 Feb 2014 09:08:50 -0800 (PST)
Received: by 10.229.13.7 with HTTP; Wed, 12 Feb 2014 09:08:50 -0800 (PST)
In-Reply-To: <CAH9hSJZfEoYv9uxH31R-s1zbXy7PqQW7MsA475VJJdFrjDhj_w@mail.gmail.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E0@EXVMBX020-12.exch020.serverdata.net> <52CE8D59.3070802@it.aoyama.ac.jp> <634914A010D0B943A035D226786325D4446BE403EB@EXVMBX020-12.exch020.serverdata.net> <CAH9hSJZfEoYv9uxH31R-s1zbXy7PqQW7MsA475VJJdFrjDhj_w@mail.gmail.com>
Date: Wed, 12 Feb 2014 09:08:50 -0800
Message-ID: <CAGzyod6i6HnOETSm1j_nhbQDp-129g06ZhR1Ey=KdGzzCQ+hJw@mail.gmail.com>
From: Roberto Peon <fenix@google.com>
To: Takeshi Yoshino <tyoshino@google.com>
Content-Type: multipart/alternative; boundary=001a11c1e05471df3e04f238a053
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Detection of broken WS connections
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 12 Feb 2014 17:08:55 -0000

--001a11c1e05471df3e04f238a053
Content-Type: text/plain; charset=UTF-8

On should steer away from keep-alive (which is anathema on mobile links),
and use liveness detection when one needs to solve these problems.
(I believe that was what you essentially proposed, Martin)

A PING (application-layer equivalent if necessary) should be automatically
generated when the client has been idle for some time and then has any
message to send. If the PONG is not received within some reasonable timeout
(preferably ~2-3RT or so, or within some other human-scale constant), the
connection should be assumed dead, and another should be created.

imho, this should be done for you by the browser (not that it necessarily
*is*, but it *should*)

Keep-alive is fairly evil at the infrastructure layer, and incents
browsers, servers and carriers to do things to the tear down connection
quickly/prematurely.

Liveness-detection doesn't solve the server notification problem (which
requires at least some level of high-level polling, though hopefully on a
longer timescale), but keep in mind that in seeking perfection, one is
likely to cause a significant less than perfect response from the system
and the players within it.

(It is pretty annoying that we've not developed good APIs or behaviors for
doing decent things on mobile devices...)

-=R


On Tue, Feb 11, 2014 at 7:50 PM, Takeshi Yoshino <tyoshino@google.com>wrote:

> On Thu, Jan 9, 2014 at 9:13 PM, Tobias Oberstein <
> tobias.oberstein@tavendo.de> wrote:
>
>> > The WS API is explicit about the fact that there is no way from
>> JavaScript to
>> > generate PINGs or check for PONGs. So do I just create my own,
>> application
>> > level equivalent, and send a ping-like message every so often and assume
>>
>> You can have a WS server that does automatic ping-pong for connection
>> keep-alive/loss-detection.
>>
>>
> This works for server. For client, app level work is needed.
>
>
>> Or you can do it on app level from JS.
>>
>> > the connection is dead when I don't get a reply "soonish"? Are there
>> libraries
>> > that support such a mechanism? Or is every browser supported to
>> regularly
>> > PINGPONG with the server, and call the onclose function when there's no
>>
>> As far as I know, there is no JS API where you could request the browser
>> to do automatic ping-pong upon opening a WS connection.
>>
>> > answer?
>> >
>>
>
> Right. We've been saying that we should start discussing such a change on
> the API, but haven't.
>
> // Send a ping (w/o body) every 5 sec
> ws.pingInterval = 5;
> // _Close the WebSocket connection_ when it passed 10 sec without
> receiving any byte (not limited to PONG)
> ws.timeout = 10;
>
>
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi
>
>

--001a11c1e05471df3e04f238a053
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">On should steer away from keep-alive (which is anathema on=
 mobile links), and use liveness detection when one needs to solve these pr=
oblems.<div>(I believe that was what you essentially proposed, Martin)<br>
<div><br><div>A PING (application-layer equivalent if necessary) should be =
automatically generated when the client has been idle for some time and the=
n has any message to send. If the PONG is not received within some reasonab=
le timeout (preferably ~2-3RT or so, or within some other human-scale const=
ant), the connection should be assumed dead, and another should be created.=
</div>

<div><br></div><div>imho, this should be done for you by the browser (not t=
hat it necessarily *is*, but it *should*)</div><div><br></div><div>Keep-ali=
ve is fairly evil at the infrastructure layer, and incents browsers, server=
s and carriers to do things to the tear down connection quickly/prematurely=
.</div>
<div><br></div><div>Liveness-detection doesn&#39;t solve the server notific=
ation problem (which requires at least some level of high-level polling, th=
ough hopefully on a longer timescale), but keep in mind that in seeking per=
fection, one is likely to cause a significant less than perfect response fr=
om the system and the players within it.=C2=A0</div>
<div><br></div><div>(It is pretty annoying that we&#39;ve not developed goo=
d APIs or behaviors for doing decent things on mobile devices...)</div><div=
><br></div><div>-=3DR</div></div></div></div><div class=3D"gmail_extra"><br=
>
<br><div class=3D"gmail_quote">On Tue, Feb 11, 2014 at 7:50 PM, Takeshi Yos=
hino <span dir=3D"ltr">&lt;<a href=3D"mailto:tyoshino@google.com" target=3D=
"_blank">tyoshino@google.com</a>&gt;</span> wrote:<br><blockquote class=3D"=
gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-=
left:1ex">
<div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote"><div=
 class=3D"">On Thu, Jan 9, 2014 at 9:13 PM, Tobias Oberstein <span dir=3D"l=
tr">&lt;<a href=3D"mailto:tobias.oberstein@tavendo.de" target=3D"_blank">to=
bias.oberstein@tavendo.de</a>&gt;</span> wrote:<br>


<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div>&gt; The WS API is explicit about the f=
act that there is no way from JavaScript to<br>
&gt; generate PINGs or check for PONGs. So do I just create my own, applica=
tion<br>
&gt; level equivalent, and send a ping-like message every so often and assu=
me<br>
<br>
</div>You can have a WS server that does automatic ping-pong for connection=
 keep-alive/loss-detection.<br>
<br></blockquote><div><br></div></div><div>This works for server. For clien=
t, app level work is needed.</div><div class=3D""><div>=C2=A0</div><blockqu=
ote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc s=
olid;padding-left:1ex">

Or you can do it on app level from JS.<br>
<div><br>
&gt; the connection is dead when I don&#39;t get a reply &quot;soonish&quot=
;? Are there libraries<br>
&gt; that support such a mechanism? Or is every browser supported to regula=
rly<br>
&gt; PINGPONG with the server, and call the onclose function when there&#39=
;s no<br>
<br>
</div>As far as I know, there is no JS API where you could request the brow=
ser to do automatic ping-pong upon opening a WS connection.<br>
<div><div><br>
&gt; answer?<br>
&gt;<br></div></div></blockquote><div><br></div></div><div>Right. We&#39;ve=
 been saying that we should start discussing such a change on the API, but =
haven&#39;t.</div><div><br></div><div>// Send a ping (w/o body) every 5 sec=
</div>


<div>ws.pingInterval =3D 5;</div><div>// _Close the WebSocket connection_ w=
hen it passed 10 sec without receiving any byte (not limited to PONG)</div>=
<div>ws.timeout =3D 10;</div><div>=C2=A0</div></div></div></div>
<br>_______________________________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
<br></blockquote></div><br></div>

--001a11c1e05471df3e04f238a053--


From tyoshino@google.com  Wed Feb 12 20:14:41 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id D78971A00DC for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 20:14:41 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.926
X-Spam-Level: 
X-Spam-Status: No, score=-1.926 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id YbxEj0qTtcGf for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 20:14:40 -0800 (PST)
Received: from mail-we0-x22c.google.com (mail-we0-x22c.google.com [IPv6:2a00:1450:400c:c03::22c]) by ietfa.amsl.com (Postfix) with ESMTP id B4D771A00D5 for <hybi@ietf.org>; Wed, 12 Feb 2014 20:14:39 -0800 (PST)
Received: by mail-we0-f172.google.com with SMTP id p61so6977664wes.3 for <hybi@ietf.org>; Wed, 12 Feb 2014 20:14:38 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc:content-type; bh=R5j51nv15t/obb4rBwDhxcUjNt5b5LClvK2n4r+7EQI=; b=acfOKt+utuaSf1HdEIlAGa6rk1MWdhLVX0AgQIsh45iGFM52aIMbR5jrJA87iFyBa8 W15AdbitjeE+gF7ed/Bprvd77kAI++p8q3hAz//jBl9nmuPqqi96PZ9SQJwV1DoPkNlZ HbbdDYQQl2I0FElba58PJUHKR4bwGzkI8doBKTiw5Hz2yECMOSqS76NJvKnLQo6ze44H kwsA0NzQCiQp9W8iBjKFyX6bE29TfBHXHuZfdcv2fPKzeUQ8K1lVdTIHdB1/Hzbbtrgg zP3FMOlX8NNufIVi3fn0OAxx7BEXr6DyH1MCG+86AgW5OasBfccDe5syadl6DtVXxrYG W1yg==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc:content-type; bh=R5j51nv15t/obb4rBwDhxcUjNt5b5LClvK2n4r+7EQI=; b=SeGbZISAQ22yjZc+Z1txJjRBFmZ14+H5RZA4jG3W2wdTqG46DawrumgGen5wL/pbLF iQvbtMz21wyuBHeWrivm2JXmH4tp0EnuAHtHWxx3WbxT+hC8Gvh2qTphM6rug5bx53Ky hjY1sQa7FJmRfAPDWp9mNwJj1eeJPRQGp745uFEVSAVXLIRvVm0RyrAxvkIggOQyDEBV 8UN6b3v+Yfm1yVCcQm+lk//tinvDoc+Qo/SJ6gJa3DRa4VlUscdQEvaxRH3WyCWEDyoI Jqtmw1btzmY7FtDeB4mHv+N5pFVma8ntp0zNycwQwmdkhOIXasX0sEkfQjHOgrO91EKk mp6A==
X-Gm-Message-State: ALoCoQln4KadcBhANC/EXV03DeevPfBzy0gbDGJehw5D4U23p1HAmgKm6XoLu9Vr+MVQGDDpl1XiqxDaIPmyE0j9VISh3lzJyE3y5KItxudbJK4xHwJ9uQl51UR0UF9tIyfnJQSXn6hOdKRzpdMIWAPj9kTwkBWMQwbmdKe9TGW6j+29Tzq1UPUFN+UDCuvkDzZ08qtgS3e6
X-Received: by 10.180.73.173 with SMTP id m13mr4711543wiv.52.1392264878157; Wed, 12 Feb 2014 20:14:38 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Wed, 12 Feb 2014 20:14:18 -0800 (PST)
In-Reply-To: <C5F162C5-D2FF-4140-B80B-5BF2DEE80AB1@zaphoyd.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <CACuKZqEcA1Pv8RpWfmThMjTzi2BbVMMKXqujs6BxVfxRPZJ9NQ@mail.gmail.com> <C5F162C5-D2FF-4140-B80B-5BF2DEE80AB1@zaphoyd.com>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Thu, 13 Feb 2014 13:14:18 +0900
Message-ID: <CAH9hSJZ3KKViOTiwxjKjTe=bkoPtFC8cbkT=-RtJ5D9Bi9C03A@mail.gmail.com>
To: Peter Thorson <webmaster@zaphoyd.com>
Content-Type: multipart/alternative; boundary=f46d043c07e07f411b04f241edcb
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 13 Feb 2014 04:14:42 -0000

--f46d043c07e07f411b04f241edcb
Content-Type: text/plain; charset=UTF-8

On Thu, Feb 13, 2014 at 1:25 AM, Peter Thorson <webmaster@zaphoyd.com>wrote:

> > PING - not necessary while we are writing data. If peer is
>  > unresponsive, we'll learn that from TCP layer.
>
> It does provide some separation of concerns and the ability to do things
> like measure RTT during a long message.


There can be an intermediary which silently terminates TCP and relay
WebSocket traffic. In such case, app level request for pong is useful for
liveness detection. The intermediary still can relay unresponsiveness of
the peer endpoint to the endpoint (by disconnecting the TCP connection) and
maybe it needs to do so due to limited buffer. So, I admit that this
argument is not so strong support for necessity of PING.

--f46d043c07e07f411b04f241edcb
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote">On T=
hu, Feb 13, 2014 at 1:25 AM, Peter Thorson <span dir=3D"ltr">&lt;<a href=3D=
"mailto:webmaster@zaphoyd.com" target=3D"_blank">webmaster@zaphoyd.com</a>&=
gt;</span> wrote:<br>


<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div>&gt; PING - not necessary while we are writing data. =
If peer is<br>


</div><div>
&gt; unresponsive, we&#39;ll learn that from TCP layer.<br>
<br>
</div>It does provide some separation of concerns and the ability to do thi=
ngs like measure RTT during a long message.</blockquote><div><br></div><div=
>There can be an intermediary which silently terminates TCP and relay WebSo=
cket traffic. In such case, app level request for pong is useful for livene=
ss detection. The intermediary still can relay unresponsiveness of the peer=
 endpoint to the endpoint (by disconnecting the TCP connection) and maybe i=
t needs to do so due to limited buffer. So, I admit that this argument is n=
ot so strong support for necessity of PING.</div>


</div></div></div>

--f46d043c07e07f411b04f241edcb--


From tyoshino@google.com  Wed Feb 12 20:38:47 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id EA01C1A00E9 for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 20:38:47 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.926
X-Spam-Level: 
X-Spam-Status: No, score=-1.926 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id yZx_UBbeIigG for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 20:38:44 -0800 (PST)
Received: from mail-wi0-x22b.google.com (mail-wi0-x22b.google.com [IPv6:2a00:1450:400c:c05::22b]) by ietfa.amsl.com (Postfix) with ESMTP id 6D0361A00EB for <hybi@ietf.org>; Wed, 12 Feb 2014 20:38:43 -0800 (PST)
Received: by mail-wi0-f171.google.com with SMTP id cc10so7983902wib.16 for <hybi@ietf.org>; Wed, 12 Feb 2014 20:38:41 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc:content-type; bh=RYIB61UZ0UCld+V/iPX/WNH/dv91umbIVE70pldQgy8=; b=fvMW7sZpcjT+TxExIMhiM5a0hmdHPIrQ3zsM+ljHmsiYhMBCWRPZPHx/tGOQaVvYuz S/3sRBEVfCg/KjGkOogTdxi3ZgDIkkXNHZhDT5MvgOTxCSIgoZ33N89ZI7V1PTclfPz9 M/Iw4x8o7feN9DsmLjf5pXBJnm6m8Hj+8CfTVx667/94wR1IplhN4XGOgwvk5YeNzUrB ttHRj69aArfxRO/rPkWCqcsWFLU2qLp93IQvTA4DleOaZCRoH2Z2Lj0hcbENI6xo651S 2+4NNlYmT62ZsGQZFg/cjKxMg6OHjBBs5w+cgScDd7JoZ7cG4wP25sNX/xFOrkmoZCWN CqLw==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc:content-type; bh=RYIB61UZ0UCld+V/iPX/WNH/dv91umbIVE70pldQgy8=; b=a9lcgmBUk5IO2iejfDh0dGdhjVe5E5gX+/ev9CnqFI4FhX6siDUzEiuV9NjeWNqYxP bKBaAcvHLDsfkTXC0OS5s/2G9krtRum/cUbj1WKUakuHm7xBOcnL0TkSnMLnXBj0V5+n VKJ9rdnzqoDesResW3msElNoLQyhR2/73hlyCel9ntpSK/xvCFUXCsTb9W20yLAktjD4 OYuepBVOrj9qo7TbmD2ED8/5l37ph09xXbYsn9VhgaKO2JTELQqSJMYkVkC6Kgi4Z4Wj xT9ivx8h1ZTaTSblmQVuRN0Kilu70BwpsynaJ3e+O/ptnxw9UhOHlrwpiDnULYKy41W8 +Auw==
X-Gm-Message-State: ALoCoQmJkpKRRJ8hvw7w09EQltcIabaIpUzwHQalPE8doWOcARiulISe4EJlHVWjsckdYxcLk8nz8nKJ48LMPTPupnk3i3ZOorjpls9sEcOpxtoQfZ2EsyCRIT/B6V/1KTwAVcPpAD+TI7Bape+IKDD2jOntb8zKyjM5Lf9XjvGsVIfyJ2daXM6EAvq3OZTvVdZfxNuAsBVj
X-Received: by 10.194.110.41 with SMTP id hx9mr4205658wjb.28.1392266321516; Wed, 12 Feb 2014 20:38:41 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Wed, 12 Feb 2014 20:38:21 -0800 (PST)
In-Reply-To: <CAG4zZZA4WR50Ea2e36XEgTgpJURwiMFn+DXeYuxHFfZatYzDOQ@mail.gmail.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <CACuKZqEcA1Pv8RpWfmThMjTzi2BbVMMKXqujs6BxVfxRPZJ9NQ@mail.gmail.com> <CAG4zZZCr4aTfVpw2coX2g0qw++4kdgNCFVze6tHKZ+fJNqb0aQ@mail.gmail.com> <CAG4zZZA4WR50Ea2e36XEgTgpJURwiMFn+DXeYuxHFfZatYzDOQ@mail.gmail.com>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Thu, 13 Feb 2014 13:38:21 +0900
Message-ID: <CAH9hSJYv5VXGjS7AfG12-ArEvE6Uj_eE6pPxAiQcchcgV8vHcg@mail.gmail.com>
To: Joakim Erdfelt <joakim@intalio.com>
Content-Type: multipart/alternative; boundary=089e010d89e0871e2904f24243e9
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 13 Feb 2014 04:38:48 -0000

--089e010d89e0871e2904f24243e9
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

Joakim,

In my post, I was doubting necessity of frame length header IN THE CONTEXT
OF WebSocket over HTTP/2.0 discussion as the HTTP/2.0 DATA frame also has a
frame length that is immediately decodable and you can choose to send a big
one DATA frame to send a big message to tell the peer that it's a big frame
at the start of arrival of the frame. Not doubting necessity of frame
length in general.

However, I noticed that I didn't consider that the size of the frame length
field of the current HTTP/2.0 spec is very small. It's 14 bit. Old SPDYs
used 24 bits (and the latest SPDY/4 uses 16 bits).

So, without help by WebSocket level length field, even by using an
un-fragmented WebSocket frame (at WebSocket semantics level), we cannot
tell the peer that "we're sending a message which is N (N > 16k) bytes
long" on the first DATA frame. Sorry that I didn't notice this point.


On Thu, Feb 13, 2014 at 1:19 AM, Joakim Erdfelt <joakim@intalio.com> wrote:

> Eliminating the size from the frame also eliminates the ability to
> fast-fail on excessive messages.
>
> Scenario (with current RFC6455 / WS+HTTP/1.1):
>   Sending side starts to write out a single message of 100MB.
>   Receiving side, using a whole message delivery based API is configured
> for max messages of 20MB.
>   Receiving side sees the 100MB frame size and automatically closes the
> connection with 1009 and doesn't bother reading past the current network
> buffer.
>
> Scenario (without frame sizes):
>   Sending side starts to write out a single message of 100MB.
>   Receiving side, using a whole message delivery based API is configured
> for max messages of 20MB.
>   Receiving side reads till the 20MB mark, then fails the connection with
> 1009.  Wasting time, memory, and resources for no good reason.
>
> I really don't see how a discussion of a protocol without sizes is even
> going on.
> I mean, even the IP layer and TCP layer have sizes for their frames.
>
>
>
> --
> Joakim Erdfelt <joakim@intalio.com>
> webtide.com <http://www.webtide.com/> - intalio.com/jetty
> Expert advice, services and support from from the Jetty & CometD experts
> eclipse.org/jetty - cometd.org
>
>
> On Wed, Feb 12, 2014 at 9:13 AM, Joakim Erdfelt <joakim@intalio.com>wrote=
:
>
>> Scenario:
>>   Sending side uses a message stream based API to write out 1 websocket
>> message.
>>   The sending side's API is configured/implemented to buffer only 16k at
>> a time, writing out a frame of 16k for each buffer filled.
>>   The sending side writes out a 600k message.
>>   The receiving side is using a different API, one that is based on
>> complete messages.
>>   The receiving side is configured to not accept messages over 100k.
>>   The receiving side's API/implementation tosses a close code 1009 to th=
e
>> sending side, per API.
>>   The receiving side's endpoint receives an error and close notification
>> indicating a protocol closure due to excessive message size from sending
>> side.
>>
>>
>>
>> --
>> Joakim Erdfelt <joakim@intalio.com>
>>  webtide.com <http://www.webtide.com/> - intalio.com/jetty
>> Expert advice, services and support from from the Jetty & CometD experts
>> eclipse.org/jetty - cometd.org
>>
>>
>> On Wed, Feb 12, 2014 at 9:00 AM, Zhong Yu <zhong.j.yu@gmail.com> wrote:
>>
>>> Why not send the message in one frame? The frame length is then the
>>> message length.
>>>
>>> One argument for fragmentation is to insert control frames while
>>> transmitting a long message. However that reasoning is not very
>>> convincing:
>>>
>>> PONG - as long as we are writing data, any data, the peer should see
>>> that the connection is alive. It's unreasonable for the peer to think
>>> that our end is unresponsive while it is receiving data from us.
>>>
>>> PING - not necessary while we are writing data. If peer is
>>> unresponsive, we'll learn that from TCP layer.
>>>
>>> CLOSE - closing the connection in the middle of a message is abnormal
>>> anyway, it's probably not important to do the close handshake.
>>>
>>> Zhong Yu
>>>
>>>
>>> On Fri, Feb 7, 2014 at 1:30 AM, Takeshi Yoshino <tyoshino@google.com>
>>> wrote:
>>> > I remember that in early days of the standardization, somebody was
>>> proposing
>>> > (*) that we should have a message length field in the first fragment =
to
>>> > allow for allocating buffer for full message at the beginning.
>>> >
>>> > On Thu, Jan 16, 2014 at 3:27 PM, "Martin J. D=C3=BCrst" <
>>> duerst@it.aoyama.ac.jp>
>>> > wrote:
>>> >>
>>> >> On 2014/01/16 15:05, Yutaka Hirano wrote:
>>> >>>
>>> >>> Hi Joakim,
>>> >>>
>>> >>> Thanks for the comment.
>>> >>>
>>> >>> Knowing the size of a frame at the start of the frame, traditional
>>> >>>>
>>> >>>> websocket behavior, is desired, as it allows us to fast fail for t=
he
>>> >>>> 1009
>>> >>>> (message too big) use case.
>>> >>>
>>> >>>
>>> >>> Do you know how effective it is?
>>> >>>
>>> >>> Having no prior knowledge of frame size is undesirable to us.
>>> >>>
>>> >>> On the other hand, the frame sender have more freedom because they
>>> don't
>>> >>> have to know the size of the frame in front.
>>> >>
>>> >>
>>> >> My understanding of why messages are split into frames was that this
>>> would
>>> >> allow to start a message without knowing its length. But if you don'=
t
>>> know
>>> >> the size of a frame before sending it, then the right solution is to
>>> make it
>>> >> smaller. Frames are just arbitrary pieces of a message.
>>> >
>>> >
>>> > We should be clear if we want to make it able to
>>> > (a) know the length of WebSocket frame on receiving the header of eac=
h
>>> > fragment
>>> > (b) know the length of WebSocket message on receiving the first
>>> fragment
>>> >
>>> > As the proposal (*) has been rejected and not in the RFC6455 header,
>>> (b) is
>>> > not true even in RFC6455. What Joakim said should be to keep (a) whic=
h
>>> is
>>> > true now in RFC6455.
>>> >
>>> > But if a message is fragmented into smaller frames, we cannot reject
>>> (and
>>> > send 1009) the message on receiving the first frame. "Whether we can
>>> > fast-reject a big message on receiving some non-final frame" depends
>>> on how
>>> > the peer fragments a message.
>>> >
>>> > In some plans discussed in the main thread, it's planned to drop fram=
e
>>> > length info for efficiency. But something else such as the length
>>> field of
>>> > HTTP/2.0 DATA frame tells how much data the fragment (when multiple
>>> DATAs
>>> > convey a WebSocket frame, each DATA would be called fragments of a
>>> frame)
>>> > contains. So, we won't be able to know the length of fragments
>>> > (original/encapsulated WebSocket frames) on receiving each of them, b=
ut
>>> > still be able to know the length of sub-fragments (e.g. HTTP/2.0 data
>>> > frames).
>>> >
>>> > I think this change is acceptable and shouldn't be considered as
>>> breaking of
>>> > RFC6455 semantics. Even in the mux extension I was editing
>>> > (http://tools.ietf.org/html/draft-ietf-hybi-websocket-multiplexing-11=
),
>>> I
>>> > made the same change. The length of the encapsulated frame is unknown
>>> until
>>> > receiving all the fragments of the encapsulating message. No one was
>>> > objecting to that when we were discussing the spec, IIRC.
>>> >
>>> >
>>> > _______________________________________________
>>> > hybi mailing list
>>> > hybi@ietf.org
>>> > https://www.ietf.org/mailman/listinfo/hybi
>>> >
>>>
>>> _______________________________________________
>>> hybi mailing list
>>> hybi@ietf.org
>>> https://www.ietf.org/mailman/listinfo/hybi
>>>
>>
>>
>

--089e010d89e0871e2904f24243e9
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Joakim,<div><br></div><div>In my post, I was doubting nece=
ssity of frame length header IN THE CONTEXT OF WebSocket over HTTP/2.0 disc=
ussion as the HTTP/2.0 DATA frame also has a frame length that is immediate=
ly decodable and you can choose to send a big one DATA frame to send a big =
message to tell the peer that it&#39;s a big frame at the start of arrival =
of the frame. Not doubting necessity of frame length in general.</div>

<div><br></div><div>However, I noticed that I didn&#39;t consider that the =
size of the frame length field of the current HTTP/2.0 spec is very small. =
It&#39;s 14 bit. Old SPDYs used 24 bits (and the latest SPDY/4 uses 16 bits=
).</div>

<div><br></div><div>So, without help by WebSocket level length field, even =
by using an un-fragmented WebSocket frame (at WebSocket semantics level), w=
e cannot tell the peer that &quot;we&#39;re sending a message which is N (N=
 &gt; 16k) bytes long&quot; on the first DATA frame. Sorry that I didn&#39;=
t notice this point.</div>

<div><br></div><div><br></div><div class=3D"gmail_extra"><div class=3D"gmai=
l_quote">On Thu, Feb 13, 2014 at 1:19 AM, Joakim Erdfelt <span dir=3D"ltr">=
&lt;<a href=3D"mailto:joakim@intalio.com" target=3D"_blank">joakim@intalio.=
com</a>&gt;</span> wrote:<br>

<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div dir=3D"ltr"><div>Eliminating the size f=
rom the frame also eliminates the ability to fast-fail on excessive message=
s.</div>

<div><br></div><div>Scenario (with current RFC6455 / WS+HTTP/1.1):</div><di=
v>=C2=A0 Sending side starts to write out a single message of 100MB.</div>
<div>=C2=A0 Receiving side, using a whole message delivery based API is con=
figured for max messages of 20MB.</div><div>=C2=A0 Receiving side sees the =
100MB frame size and automatically closes the connection with 1009 and does=
n&#39;t bother reading past the current network buffer.</div>


<div><br></div><div>Scenario (without frame sizes):</div><div>=C2=A0 Sendin=
g side starts to write out a single message of 100MB.</div><div><div>=C2=A0=
 Receiving side, using a whole message delivery based API is configured for=
 max messages of 20MB.</div>


<div>=C2=A0 Receiving side reads till the 20MB mark, then fails the connect=
ion with 1009. =C2=A0Wasting time, memory, and resources for no good reason=
.</div></div><div><br></div>I really don&#39;t see how a discussion of a pr=
otocol without sizes is even going on.<div>


I mean, even the IP layer and TCP layer have sizes for their frames.</div><=
div><br></div><div><br></div></div><div class=3D"gmail_extra"><div class=3D=
"im"><br clear=3D"all"><div><div dir=3D"ltr"><div>--</div><div>Joakim Erdfe=
lt &lt;<a href=3D"mailto:joakim@intalio.com" target=3D"_blank">joakim@intal=
io.com</a>&gt;</div>


<div><a href=3D"http://www.webtide.com/" target=3D"_blank">webtide.com</a>=
=C2=A0- <a href=3D"http://intalio.com/jetty" target=3D"_blank">intalio.com/=
jetty</a></div><div><span>Expert advice, services and support from=C2=A0</s=
pan><span>from the Jetty &amp; CometD experts</span></div>


<div><a href=3D"http://eclipse.org/jetty/" target=3D"_blank">eclipse.org/je=
tty</a>=C2=A0-=C2=A0<a href=3D"http://cometd.org/" target=3D"_blank">cometd=
.org</a></div></div></div>
<br><br></div><div><div class=3D"h5"><div class=3D"gmail_quote">On Wed, Feb=
 12, 2014 at 9:13 AM, Joakim Erdfelt <span dir=3D"ltr">&lt;<a href=3D"mailt=
o:joakim@intalio.com" target=3D"_blank">joakim@intalio.com</a>&gt;</span> w=
rote:<br>

<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
<div dir=3D"ltr"><div>Scenario:</div><div>=C2=A0 Sending side uses a messag=
e stream based API to write out 1 websocket message.</div><div>=C2=A0 The s=
ending side&#39;s API is configured/implemented to buffer only 16k at a tim=
e, writing out a frame of 16k for each buffer filled.</div>



<div>=C2=A0 The sending side writes out a 600k message.</div><div>=C2=A0 Th=
e receiving side is using a different API, one that is based on complete me=
ssages.</div><div>=C2=A0 The receiving side is configured to not accept mes=
sages over 100k.</div>



<div>=C2=A0 The receiving side&#39;s API/implementation tosses a close code=
 1009 to the sending side, per API.</div><div>=C2=A0 The receiving side&#39=
;s endpoint receives an error and close notification indicating a protocol =
closure due to excessive message size from sending side.</div>



<div><br></div><div><br></div></div><div class=3D"gmail_extra"><div><br cle=
ar=3D"all"><div><div dir=3D"ltr"><div>--</div><div>Joakim Erdfelt &lt;<a hr=
ef=3D"mailto:joakim@intalio.com" target=3D"_blank">joakim@intalio.com</a>&g=
t;</div>


<div>
<a href=3D"http://www.webtide.com/" target=3D"_blank">webtide.com</a>=C2=A0=
- <a href=3D"http://intalio.com/jetty" target=3D"_blank">intalio.com/jetty<=
/a></div><div><span>Expert advice, services and support from=C2=A0</span><s=
pan>from the Jetty &amp; CometD experts</span></div>



<div><a href=3D"http://eclipse.org/jetty/" target=3D"_blank">eclipse.org/je=
tty</a>=C2=A0-=C2=A0<a href=3D"http://cometd.org/" target=3D"_blank">cometd=
.org</a></div></div></div>
<br><br></div><div><div><div class=3D"gmail_quote">On Wed, Feb 12, 2014 at =
9:00 AM, Zhong Yu <span dir=3D"ltr">&lt;<a href=3D"mailto:zhong.j.yu@gmail.=
com" target=3D"_blank">zhong.j.yu@gmail.com</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
Why not send the message in one frame? The frame length is then the<br>
message length.<br>
<br>
One argument for fragmentation is to insert control frames while<br>
transmitting a long message. However that reasoning is not very<br>
convincing:<br>
<br>
PONG - as long as we are writing data, any data, the peer should see<br>
that the connection is alive. It&#39;s unreasonable for the peer to think<b=
r>
that our end is unresponsive while it is receiving data from us.<br>
<br>
PING - not necessary while we are writing data. If peer is<br>
unresponsive, we&#39;ll learn that from TCP layer.<br>
<br>
CLOSE - closing the connection in the middle of a message is abnormal<br>
anyway, it&#39;s probably not important to do the close handshake.<br>
<span><font color=3D"#888888"><br>
Zhong Yu<br>
</font></span><div><div><br>
<br>
On Fri, Feb 7, 2014 at 1:30 AM, Takeshi Yoshino &lt;<a href=3D"mailto:tyosh=
ino@google.com" target=3D"_blank">tyoshino@google.com</a>&gt; wrote:<br>
&gt; I remember that in early days of the standardization, somebody was pro=
posing<br>
&gt; (*) that we should have a message length field in the first fragment t=
o<br>
&gt; allow for allocating buffer for full message at the beginning.<br>
&gt;<br>
&gt; On Thu, Jan 16, 2014 at 3:27 PM, &quot;Martin J. D=C3=BCrst&quot; &lt;=
<a href=3D"mailto:duerst@it.aoyama.ac.jp" target=3D"_blank">duerst@it.aoyam=
a.ac.jp</a>&gt;<br>
&gt; wrote:<br>
&gt;&gt;<br>
&gt;&gt; On 2014/01/16 15:05, Yutaka Hirano wrote:<br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt; Hi Joakim,<br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt; Thanks for the comment.<br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt; Knowing the size of a frame at the start of the frame, traditi=
onal<br>
&gt;&gt;&gt;&gt;<br>
&gt;&gt;&gt;&gt; websocket behavior, is desired, as it allows us to fast fa=
il for the<br>
&gt;&gt;&gt;&gt; 1009<br>
&gt;&gt;&gt;&gt; (message too big) use case.<br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt; Do you know how effective it is?<br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt; Having no prior knowledge of frame size is undesirable to us.<=
br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt; On the other hand, the frame sender have more freedom because =
they don&#39;t<br>
&gt;&gt;&gt; have to know the size of the frame in front.<br>
&gt;&gt;<br>
&gt;&gt;<br>
&gt;&gt; My understanding of why messages are split into frames was that th=
is would<br>
&gt;&gt; allow to start a message without knowing its length. But if you do=
n&#39;t know<br>
&gt;&gt; the size of a frame before sending it, then the right solution is =
to make it<br>
&gt;&gt; smaller. Frames are just arbitrary pieces of a message.<br>
&gt;<br>
&gt;<br>
&gt; We should be clear if we want to make it able to<br>
&gt; (a) know the length of WebSocket frame on receiving the header of each=
<br>
&gt; fragment<br>
&gt; (b) know the length of WebSocket message on receiving the first fragme=
nt<br>
&gt;<br>
&gt; As the proposal (*) has been rejected and not in the RFC6455 header, (=
b) is<br>
&gt; not true even in RFC6455. What Joakim said should be to keep (a) which=
 is<br>
&gt; true now in RFC6455.<br>
&gt;<br>
&gt; But if a message is fragmented into smaller frames, we cannot reject (=
and<br>
&gt; send 1009) the message on receiving the first frame. &quot;Whether we =
can<br>
&gt; fast-reject a big message on receiving some non-final frame&quot; depe=
nds on how<br>
&gt; the peer fragments a message.<br>
&gt;<br>
&gt; In some plans discussed in the main thread, it&#39;s planned to drop f=
rame<br>
&gt; length info for efficiency. But something else such as the length fiel=
d of<br>
&gt; HTTP/2.0 DATA frame tells how much data the fragment (when multiple DA=
TAs<br>
&gt; convey a WebSocket frame, each DATA would be called fragments of a fra=
me)<br>
&gt; contains. So, we won&#39;t be able to know the length of fragments<br>
&gt; (original/encapsulated WebSocket frames) on receiving each of them, bu=
t<br>
&gt; still be able to know the length of sub-fragments (e.g. HTTP/2.0 data<=
br>
&gt; frames).<br>
&gt;<br>
&gt; I think this change is acceptable and shouldn&#39;t be considered as b=
reaking of<br>
&gt; RFC6455 semantics. Even in the mux extension I was editing<br>
&gt; (<a href=3D"http://tools.ietf.org/html/draft-ietf-hybi-websocket-multi=
plexing-11" target=3D"_blank">http://tools.ietf.org/html/draft-ietf-hybi-we=
bsocket-multiplexing-11</a>), I<br>
&gt; made the same change. The length of the encapsulated frame is unknown =
until<br>
&gt; receiving all the fragments of the encapsulating message. No one was<b=
r>
&gt; objecting to that when we were discussing the spec, IIRC.<br>
&gt;<br>
&gt;<br>
</div></div><div><div>&gt; _______________________________________________<=
br>
&gt; hybi mailing list<br>
&gt; <a href=3D"mailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><b=
r>
&gt; <a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blan=
k">https://www.ietf.org/mailman/listinfo/hybi</a><br>
&gt;<br>
<br>
_______________________________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
</div></div></blockquote></div><br></div></div></div>
</blockquote></div><br></div></div></div>
</blockquote></div><br></div></div>

--089e010d89e0871e2904f24243e9--


From tyoshino@google.com  Wed Feb 12 21:14:29 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 014081A00F8 for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 21:14:29 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.326
X-Spam-Level: 
X-Spam-Status: No, score=-1.326 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, J_CHICKENPOX_27=0.6, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 0R-sgqQjWo-h for <hybi@ietfa.amsl.com>; Wed, 12 Feb 2014 21:14:26 -0800 (PST)
Received: from mail-we0-x234.google.com (mail-we0-x234.google.com [IPv6:2a00:1450:400c:c03::234]) by ietfa.amsl.com (Postfix) with ESMTP id 4F8601A00F5 for <hybi@ietf.org>; Wed, 12 Feb 2014 21:14:26 -0800 (PST)
Received: by mail-we0-f180.google.com with SMTP id u57so6880441wes.39 for <hybi@ietf.org>; Wed, 12 Feb 2014 21:14:24 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc:content-type; bh=Fa1co7qioZie0YiP47vWCzxoYB8zjTyiGY3fgQsSOjE=; b=UfvbxmVKItp5e36+tVs/b347+GGG7Shyx6Ajb/j894HimHVJFdBMrEcuwoMDISBjfq LuRQPoIIiJ1OhOYXQM/24dEi3es6ztwaW2fy3/G6kTd1nqGtqREYbpL1H1OYze/nkhtl hTXqFkxNqseO0+IT9vUDNR1n7rV+bVcE5HABbEmdD75UR6vCaOpzJLJMwq8HsHvNb9v2 iAx7n/0lG4qB44fzc/8bhA2sYOZ2s69ANgx2GjtJ0ppKFMTLHQq1qfFwghCwlOfX02/5 IYbsNXRLxA55oF54K7JaQkwP6aqPPKkQevB9enC5gYkCY5iXIuCFxEeeIZLrXLJpibor ejXg==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc:content-type; bh=Fa1co7qioZie0YiP47vWCzxoYB8zjTyiGY3fgQsSOjE=; b=YV9KZs4vfk705pHT2DAo/P77CUEho7N6cwdWUbyO0tTd4/sk1HsfWRisJCzDrbcO4d 7bIRGi7JQJMH6V9n9Yn/eme7ZKvGFREA27V2Y2mecFD0GWDsHWjixfBhKAjTZ400vgCN ZlS4SEq3l7B11rqnNIxSE+VQUn37okDRrRtwZ3gVFr7vZUG0T1l9LByGvn5ldDvuM6et 3omeQn4tLwKW3KTn5/SmB1EUjWhREbx6c6Co4rcdg06bjgs937IWQ6o5L8hGj9Zx5B6B tIEacnriaMhqQ1KATmF5WRFLqdI4qDMmE6HaUc0SfdzAaAC7rDubG1KZ/JCaquwN52A8 2YTQ==
X-Gm-Message-State: ALoCoQkzAmxzs05haTQRxJJoEkOazVkYGJjtEz5+xU4nfT1ZRFDBI/GcXFI3FcnVVSza1Eim/yFOkmkRmtSZQhY57+kUq1BOtjWfXXuuILNvZ7JzXlLz4HgP4CsfghYlk8SNVQnSnYcgtVG4KxU8fw4scfezXkWK638jv6lktosJ27A6DAcxSPbNpNmws6A52jdI/FrKa2tb
X-Received: by 10.194.175.202 with SMTP id cc10mr28573wjc.48.1392268464717; Wed, 12 Feb 2014 21:14:24 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Wed, 12 Feb 2014 21:14:04 -0800 (PST)
In-Reply-To: <CAGzyod6i6HnOETSm1j_nhbQDp-129g06ZhR1Ey=KdGzzCQ+hJw@mail.gmail.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E0@EXVMBX020-12.exch020.serverdata.net> <52CE8D59.3070802@it.aoyama.ac.jp> <634914A010D0B943A035D226786325D4446BE403EB@EXVMBX020-12.exch020.serverdata.net> <CAH9hSJZfEoYv9uxH31R-s1zbXy7PqQW7MsA475VJJdFrjDhj_w@mail.gmail.com> <CAGzyod6i6HnOETSm1j_nhbQDp-129g06ZhR1Ey=KdGzzCQ+hJw@mail.gmail.com>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Thu, 13 Feb 2014 14:14:04 +0900
Message-ID: <CAH9hSJYaRLbXsXXx=dcTyDqC+GJ8vj=E=oDg6vSQ+549DniV3Q@mail.gmail.com>
To: Roberto Peon <fenix@google.com>
Content-Type: multipart/alternative; boundary=089e013d0f4845f56d04f242c323
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Detection of broken WS connections
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 13 Feb 2014 05:14:29 -0000

--089e013d0f4845f56d04f242c323
Content-Type: text/plain; charset=UTF-8

On Thu, Feb 13, 2014 at 2:08 AM, Roberto Peon <fenix@google.com> wrote:

> On should steer away from keep-alive (which is anathema on mobile links),
> and use liveness detection when one needs to solve these problems.
> (I believe that was what you essentially proposed, Martin)
>
>
I want to confirm my understanding on your terms. Do you mean that
- it's kinda illusion that we are able to keep an end-to-end connectivity
alive
- but should just have a way to detect the peer is dead when wanted


> A PING (application-layer equivalent if necessary) should be automatically
> generated when the client has been idle for some time and then has any
> message to send. If the PONG is not received within some reasonable timeout
> (preferably ~2-3RT or so, or within some other human-scale constant), the
> connection should be assumed dead, and another should be created.
>
> imho, this should be done for you by the browser (not that it necessarily
> *is*, but it *should*)
>

Thinking of what browser vendors can do at this point:
a) have built-in default-on ping-generate/timeout-close (try to do
something intelligent like TCP's RTO calculation?) feature in browser. one
concern is that people may have skipped implementation of ping/pong
b) same as (a) but default-off and can be turned on by calling some method
on WebSocket object
c) have one-shot ping sending method on WebSocket object with timeout arg
d) have one-shot ping sending method and pong received event. Sounds
overkilling to me
e) periodical version of (c) or (d)


>
> Keep-alive is fairly evil at the infrastructure layer, and incents
> browsers, servers and carriers to do things to the tear down connection
> quickly/prematurely.
>
> Liveness-detection doesn't solve the server notification problem (which
> requires at least some level of high-level polling, though hopefully on a
> longer timescale), but keep in mind that in seeking perfection, one is
> likely to cause a significant less than perfect response from the system
> and the players within it.
>
> (It is pretty annoying that we've not developed good APIs or behaviors for
> doing decent things on mobile devices...)
>
> -=R
>
>
> On Tue, Feb 11, 2014 at 7:50 PM, Takeshi Yoshino <tyoshino@google.com>wrote:
>
>> On Thu, Jan 9, 2014 at 9:13 PM, Tobias Oberstein <
>> tobias.oberstein@tavendo.de> wrote:
>>
>>> > The WS API is explicit about the fact that there is no way from
>>> JavaScript to
>>> > generate PINGs or check for PONGs. So do I just create my own,
>>> application
>>> > level equivalent, and send a ping-like message every so often and
>>> assume
>>>
>>> You can have a WS server that does automatic ping-pong for connection
>>> keep-alive/loss-detection.
>>>
>>>
>> This works for server. For client, app level work is needed.
>>
>>
>>> Or you can do it on app level from JS.
>>>
>>> > the connection is dead when I don't get a reply "soonish"? Are there
>>> libraries
>>> > that support such a mechanism? Or is every browser supported to
>>> regularly
>>> > PINGPONG with the server, and call the onclose function when there's no
>>>
>>> As far as I know, there is no JS API where you could request the browser
>>> to do automatic ping-pong upon opening a WS connection.
>>>
>>> > answer?
>>> >
>>>
>>
>> Right. We've been saying that we should start discussing such a change on
>> the API, but haven't.
>>
>> // Send a ping (w/o body) every 5 sec
>> ws.pingInterval = 5;
>> // _Close the WebSocket connection_ when it passed 10 sec without
>> receiving any byte (not limited to PONG)
>> ws.timeout = 10;
>>
>>
>> _______________________________________________
>> hybi mailing list
>> hybi@ietf.org
>> https://www.ietf.org/mailman/listinfo/hybi
>>
>>
>

--089e013d0f4845f56d04f242c323
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote">On T=
hu, Feb 13, 2014 at 2:08 AM, Roberto Peon <span dir=3D"ltr">&lt;<a href=3D"=
mailto:fenix@google.com" target=3D"_blank">fenix@google.com</a>&gt;</span> =
wrote:<br>

<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div dir=3D"ltr">On should steer away from keep-alive (whi=
ch is anathema on mobile links), and use liveness detection when one needs =
to solve these problems.<div>

(I believe that was what you essentially proposed, Martin)<br>
<div><br></div></div></div></blockquote><div><br></div><div>I want to confi=
rm my understanding on your terms. Do you mean that</div><div>- it&#39;s ki=
nda illusion that we are able to keep an end-to-end connectivity alive</div=
>

<div>- but should just have a way to detect the peer is dead when wanted</d=
iv><div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0=
px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);borde=
r-left-style:solid;padding-left:1ex">

<div dir=3D"ltr"><div><div><div>A PING (application-layer equivalent if nec=
essary) should be automatically generated when the client has been idle for=
 some time and then has any message to send. If the PONG is not received wi=
thin some reasonable timeout (preferably ~2-3RT or so, or within some other=
 human-scale constant), the connection should be assumed dead, and another =
should be created.</div>



<div><br></div><div>imho, this should be done for you by the browser (not t=
hat it necessarily *is*, but it *should*)</div></div></div></div></blockquo=
te><div><br></div><div>Thinking of what browser vendors can do at this poin=
t:</div>

<div>a) have built-in default-on ping-generate/timeout-close (try to do som=
ething intelligent like TCP&#39;s RTO calculation?) feature in browser. one=
 concern is that people may have skipped implementation of ping/pong</div>

<div>b) same as (a) but default-off and can be turned on by calling some me=
thod on WebSocket object</div><div>c) have one-shot ping sending method on =
WebSocket object with timeout arg</div><div>d) have one-shot ping sending m=
ethod and pong received event. Sounds overkilling to me</div>

<div>e) periodical version of (c) or (d)</div><div>=C2=A0</div><blockquote =
class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1=
px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left:=
1ex">

<div dir=3D"ltr"><div><div><div><br></div><div>Keep-alive is fairly evil at=
 the infrastructure layer, and incents browsers, servers and carriers to do=
 things to the tear down connection quickly/prematurely.</div>
<div><br></div><div>Liveness-detection doesn&#39;t solve the server notific=
ation problem (which requires at least some level of high-level polling, th=
ough hopefully on a longer timescale), but keep in mind that in seeking per=
fection, one is likely to cause a significant less than perfect response fr=
om the system and the players within it.=C2=A0</div>


<div><br></div><div>(It is pretty annoying that we&#39;ve not developed goo=
d APIs or behaviors for doing decent things on mobile devices...)</div><spa=
n class=3D""><font color=3D"#888888"><div><br></div><div>-=3DR</div></font>=
</span></div>

</div></div><div class=3D"gmail_extra"><br>
<br><div class=3D"gmail_quote"><div><div class=3D"h5">On Tue, Feb 11, 2014 =
at 7:50 PM, Takeshi Yoshino <span dir=3D"ltr">&lt;<a href=3D"mailto:tyoshin=
o@google.com" target=3D"_blank">tyoshino@google.com</a>&gt;</span> wrote:<b=
r></div>

</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;b=
order-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:s=
olid;padding-left:1ex"><div><div class=3D"h5">
<div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote"><div=
>On Thu, Jan 9, 2014 at 9:13 PM, Tobias Oberstein <span dir=3D"ltr">&lt;<a =
href=3D"mailto:tobias.oberstein@tavendo.de" target=3D"_blank">tobias.oberst=
ein@tavendo.de</a>&gt;</span> wrote:<br>




<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div>&gt; The WS API is explicit about the fact that there=
 is no way from JavaScript to<br>


&gt; generate PINGs or check for PONGs. So do I just create my own, applica=
tion<br>
&gt; level equivalent, and send a ping-like message every so often and assu=
me<br>
<br>
</div>You can have a WS server that does automatic ping-pong for connection=
 keep-alive/loss-detection.<br>
<br></blockquote><div><br></div></div><div>This works for server. For clien=
t, app level work is needed.</div><div><div>=C2=A0</div><blockquote class=
=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;bo=
rder-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">



Or you can do it on app level from JS.<br>
<div><br>
&gt; the connection is dead when I don&#39;t get a reply &quot;soonish&quot=
;? Are there libraries<br>
&gt; that support such a mechanism? Or is every browser supported to regula=
rly<br>
&gt; PINGPONG with the server, and call the onclose function when there&#39=
;s no<br>
<br>
</div>As far as I know, there is no JS API where you could request the brow=
ser to do automatic ping-pong upon opening a WS connection.<br>
<div><div><br>
&gt; answer?<br>
&gt;<br></div></div></blockquote><div><br></div></div><div>Right. We&#39;ve=
 been saying that we should start discussing such a change on the API, but =
haven&#39;t.</div><div><br></div><div>// Send a ping (w/o body) every 5 sec=
</div>




<div>ws.pingInterval =3D 5;</div><div>// _Close the WebSocket connection_ w=
hen it passed 10 sec without receiving any byte (not limited to PONG)</div>=
<div>ws.timeout =3D 10;</div><div>=C2=A0</div></div></div></div>
<br></div></div><div class=3D"im">_________________________________________=
______<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
<br></div></blockquote></div><br></div>
</blockquote></div><br></div></div>

--089e013d0f4845f56d04f242c323--


From nobody Thu Feb 13 11:37:04 2014
Return-Path: <fenix@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 6ABDE1A0405 for <hybi@ietfa.amsl.com>; Thu, 13 Feb 2014 11:37:02 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.326
X-Spam-Level: 
X-Spam-Status: No, score=-1.326 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, J_CHICKENPOX_27=0.6, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id I8OTbtablF2T for <hybi@ietfa.amsl.com>; Thu, 13 Feb 2014 11:36:59 -0800 (PST)
Received: from mail-qc0-x234.google.com (mail-qc0-x234.google.com [IPv6:2607:f8b0:400d:c01::234]) by ietfa.amsl.com (Postfix) with ESMTP id 236651A03FA for <hybi@ietf.org>; Thu, 13 Feb 2014 11:36:59 -0800 (PST)
Received: by mail-qc0-f180.google.com with SMTP id i17so18571255qcy.39 for <hybi@ietf.org>; Thu, 13 Feb 2014 11:36:57 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=UvntJeX9lLWK0G7nIHFV4AT4ZUfN9FcyOzdH/VARwhc=; b=lsInk4s/Hh8B135DOUtDS6pN/ECufkLwtkblYaBh8eTRavni3EgEdB9+ymKVfyEIyA FPNw3kNAI8iR0fgKQum/5d8mjL3vugJfitR3Hhmvi24X0YQAYwJIS/dSqIH8ri8YrTnJ GRoC6/uguuNhb8CMz/x5XtSwUJCoGqDoa9Ku7CcaLDk5ONT6UvhgdXTFqLlpFx2xepVc VljdnsmIl1l/dmp3RvPfMhl2KzFw+ZJauQnHs7k6XwbwzKwZ0wGOGl0fVgbNNda66CxE +JOGwyj+O6jtL1gCuQJKAaDve02/vuBYfOcCLDbQNvU8Oeq5bPGA+9CqfqEHxzlvSasJ zlAQ==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=UvntJeX9lLWK0G7nIHFV4AT4ZUfN9FcyOzdH/VARwhc=; b=KjRYNwjgNe4WxrhkWphD2kxCA8c0cfAMJW04CddWIa+98GfhhdDeUcAceRU/q/pgS8 8LY3SPb6A1ydTSruFH/lj69MMhqB3xA2Q7OzMUmEbmiu6Gcoc6ofkIZNo3iEPgRdy0u9 QmBd6Yd3T/gZyAkijJNMrWdDxqb8XtWGlmEOkZzmsUL1VwvX9/dg+4ZzbxF3bexTAN15 OJEM5PatixNRAuaVF2T1dao/1+rx2deSzCmZ663o6n0ih6QAJHTHVbLlEXDGfqKFOfV3 mTNlw7/uqNDUK5Ch70EJaj8K0he1w8bu68gt1qqvGZNWcdgK0fcx50Uv75BPILM6WbD2 u9ag==
X-Gm-Message-State: ALoCoQlIikqv2rNPRJi3BWBIJ4PTgqKVexvmsTGjsHd+QQosmOGIDBddj4LHjmUKOuCB/1hlEiwPjdBrAeUCVGXKIKzEFSUb8kd5WVo+LA/xCXWdFkUx2UYKsBeseDjrHwcpGt89NjFUlQug4go7HIMsuQDkZGYKiOUingwkccti1zzgf5OZXBUk7mAu5lSteR68H+OT2IAt
MIME-Version: 1.0
X-Received: by 10.140.96.180 with SMTP id k49mr5414280qge.4.1392320217720; Thu, 13 Feb 2014 11:36:57 -0800 (PST)
Received: by 10.229.13.7 with HTTP; Thu, 13 Feb 2014 11:36:57 -0800 (PST)
In-Reply-To: <CAH9hSJYaRLbXsXXx=dcTyDqC+GJ8vj=E=oDg6vSQ+549DniV3Q@mail.gmail.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E0@EXVMBX020-12.exch020.serverdata.net> <52CE8D59.3070802@it.aoyama.ac.jp> <634914A010D0B943A035D226786325D4446BE403EB@EXVMBX020-12.exch020.serverdata.net> <CAH9hSJZfEoYv9uxH31R-s1zbXy7PqQW7MsA475VJJdFrjDhj_w@mail.gmail.com> <CAGzyod6i6HnOETSm1j_nhbQDp-129g06ZhR1Ey=KdGzzCQ+hJw@mail.gmail.com> <CAH9hSJYaRLbXsXXx=dcTyDqC+GJ8vj=E=oDg6vSQ+549DniV3Q@mail.gmail.com>
Date: Thu, 13 Feb 2014 11:36:57 -0800
Message-ID: <CAGzyod4Ga0Gy9f9f+chufqhWa4XswmOdAtNU7rkXY6ZzayTa-g@mail.gmail.com>
From: Roberto Peon <fenix@google.com>
To: Takeshi Yoshino <tyoshino@google.com>
Content-Type: multipart/alternative; boundary=001a113ac3d0fe15e304f24ecf5a
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/9LNMsc5h8NDiZlcA0_cZBC6xZHI
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Detection of broken WS connections
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 13 Feb 2014 19:37:02 -0000

--001a113ac3d0fe15e304f24ecf5a
Content-Type: text/plain; charset=UTF-8

On Wed, Feb 12, 2014 at 9:14 PM, Takeshi Yoshino <tyoshino@google.com>wrote:

> On Thu, Feb 13, 2014 at 2:08 AM, Roberto Peon <fenix@google.com> wrote:
>
>> On should steer away from keep-alive (which is anathema on mobile links),
>> and use liveness detection when one needs to solve these problems.
>> (I believe that was what you essentially proposed, Martin)
>>
>>
> I want to confirm my understanding on your terms. Do you mean that
> - it's kinda illusion that we are able to keep an end-to-end connectivity
> alive
> - but should just have a way to detect the peer is dead when wanted
>

yup.
Essentially, we don't want traffic black-holed (because the connection is
dead but we haven't detected it), and we don't want to incent "the network"
to kill connections unnecessarily (this is particularly an issue on shared
bandwidth mediums like mobile networks).


>
>
>> A PING (application-layer equivalent if necessary) should be
>> automatically generated when the client has been idle for some time and
>> then has any message to send. If the PONG is not received within some
>> reasonable timeout (preferably ~2-3RT or so, or within some other
>> human-scale constant), the connection should be assumed dead, and another
>> should be created.
>>
>> imho, this should be done for you by the browser (not that it necessarily
>> *is*, but it *should*)
>>
>
> Thinking of what browser vendors can do at this point:
> a) have built-in default-on ping-generate/timeout-close (try to do
> something intelligent like TCP's RTO calculation?) feature in browser. one
> concern is that people may have skipped implementation of ping/pong
> b) same as (a) but default-off and can be turned on by calling some method
> on WebSocket object
> c) have one-shot ping sending method on WebSocket object with timeout arg
> d) have one-shot ping sending method and pong received event. Sounds
> overkilling to me
> e) periodical version of (c) or (d)
>

We have implemenation experience for (a) for HTTP/2/SPDY, and it seems to
work pretty well.
HTTP2/SPDY has the same problem: if the connection is dead, it is a poor
user experience to attempt to keep sending all traffic down it!

The design rationale behind the choice of (a) was:
it transparently fixed the biggest problem case without requiring any user
involvement or knowledge.
It was important to use ping, because ping/pong requires no interaction
with the application-layer logic, and thus is most likely to be consistent
and have timing similar to TCP's PING.

-=R


>
>
>>
>> Keep-alive is fairly evil at the infrastructure layer, and incents
>> browsers, servers and carriers to do things to the tear down connection
>> quickly/prematurely.
>>
>> Liveness-detection doesn't solve the server notification problem (which
>> requires at least some level of high-level polling, though hopefully on a
>> longer timescale), but keep in mind that in seeking perfection, one is
>> likely to cause a significant less than perfect response from the system
>> and the players within it.
>>
>> (It is pretty annoying that we've not developed good APIs or behaviors
>> for doing decent things on mobile devices...)
>>
>> -=R
>>
>>
>> On Tue, Feb 11, 2014 at 7:50 PM, Takeshi Yoshino <tyoshino@google.com>wrote:
>>
>>> On Thu, Jan 9, 2014 at 9:13 PM, Tobias Oberstein <
>>> tobias.oberstein@tavendo.de> wrote:
>>>
>>>> > The WS API is explicit about the fact that there is no way from
>>>> JavaScript to
>>>> > generate PINGs or check for PONGs. So do I just create my own,
>>>> application
>>>> > level equivalent, and send a ping-like message every so often and
>>>> assume
>>>>
>>>> You can have a WS server that does automatic ping-pong for connection
>>>> keep-alive/loss-detection.
>>>>
>>>>
>>> This works for server. For client, app level work is needed.
>>>
>>>
>>>> Or you can do it on app level from JS.
>>>>
>>>> > the connection is dead when I don't get a reply "soonish"? Are there
>>>> libraries
>>>> > that support such a mechanism? Or is every browser supported to
>>>> regularly
>>>> > PINGPONG with the server, and call the onclose function when there's
>>>> no
>>>>
>>>> As far as I know, there is no JS API where you could request the
>>>> browser to do automatic ping-pong upon opening a WS connection.
>>>>
>>>> > answer?
>>>> >
>>>>
>>>
>>> Right. We've been saying that we should start discussing such a change
>>> on the API, but haven't.
>>>
>>> // Send a ping (w/o body) every 5 sec
>>> ws.pingInterval = 5;
>>> // _Close the WebSocket connection_ when it passed 10 sec without
>>> receiving any byte (not limited to PONG)
>>> ws.timeout = 10;
>>>
>>>
>>> _______________________________________________
>>> hybi mailing list
>>> hybi@ietf.org
>>> https://www.ietf.org/mailman/listinfo/hybi
>>>
>>>
>>
>

--001a113ac3d0fe15e304f24ecf5a
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><br><div class=3D"gmail_extra"><br><br><div class=3D"gmail=
_quote">On Wed, Feb 12, 2014 at 9:14 PM, Takeshi Yoshino <span dir=3D"ltr">=
&lt;<a href=3D"mailto:tyoshino@google.com" target=3D"_blank">tyoshino@googl=
e.com</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div dir=3D"ltr"><div class=3D"gmail_extra">=
<div class=3D"gmail_quote"><div class=3D"">On Thu, Feb 13, 2014 at 2:08 AM,=
 Roberto Peon <span dir=3D"ltr">&lt;<a href=3D"mailto:fenix@google.com" tar=
get=3D"_blank">fenix@google.com</a>&gt;</span> wrote:<br>


<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div dir=3D"ltr">On should steer away from keep-alive (whi=
ch is anathema on mobile links), and use liveness detection when one needs =
to solve these problems.<div>


(I believe that was what you essentially proposed, Martin)<br>
<div><br></div></div></div></blockquote><div><br></div></div><div>I want to=
 confirm my understanding on your terms. Do you mean that</div><div>- it&#3=
9;s kinda illusion that we are able to keep an end-to-end connectivity aliv=
e</div>


<div>- but should just have a way to detect the peer is dead when wanted</d=
iv></div></div></div></blockquote><div><br></div><div>yup.</div><div>Essent=
ially, we don&#39;t want traffic black-holed (because the connection is dea=
d but we haven&#39;t detected it), and we don&#39;t want to incent &quot;th=
e network&quot; to kill connections unnecessarily (this is particularly an =
issue on shared bandwidth mediums like mobile networks).</div>
<div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8=
ex;border-left:1px #ccc solid;padding-left:1ex"><div dir=3D"ltr"><div class=
=3D"gmail_extra"><div class=3D"gmail_quote"><div class=3D""><div>=C2=A0</di=
v><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;borde=
r-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid=
;padding-left:1ex">


<div dir=3D"ltr"><div><div><div>A PING (application-layer equivalent if nec=
essary) should be automatically generated when the client has been idle for=
 some time and then has any message to send. If the PONG is not received wi=
thin some reasonable timeout (preferably ~2-3RT or so, or within some other=
 human-scale constant), the connection should be assumed dead, and another =
should be created.</div>




<div><br></div><div>imho, this should be done for you by the browser (not t=
hat it necessarily *is*, but it *should*)</div></div></div></div></blockquo=
te><div><br></div></div><div>Thinking of what browser vendors can do at thi=
s point:</div>


<div>a) have built-in default-on ping-generate/timeout-close (try to do som=
ething intelligent like TCP&#39;s RTO calculation?) feature in browser. one=
 concern is that people may have skipped implementation of ping/pong</div>


<div>b) same as (a) but default-off and can be turned on by calling some me=
thod on WebSocket object</div><div>c) have one-shot ping sending method on =
WebSocket object with timeout arg</div><div>d) have one-shot ping sending m=
ethod and pong received event. Sounds overkilling to me</div>


<div>e) periodical version of (c) or (d)</div></div></div></div></blockquot=
e><div><br>We have implemenation experience for (a) for HTTP/2/SPDY, and it=
 seems to work pretty well.</div><div>HTTP2/SPDY has the same problem: if t=
he connection is dead, it is a poor user experience to attempt to keep send=
ing all traffic down it!</div>
<div><br></div><div>The design rationale behind the choice of (a) was:</div=
><div>it transparently fixed the biggest problem case without requiring any=
 user involvement or knowledge.</div><div>It was important to use ping, bec=
ause ping/pong requires no interaction with the application-layer logic, an=
d thus is most likely to be consistent and have timing similar to TCP&#39;s=
 PING.</div>
<div><br></div><div>-=3DR</div><div>=C2=A0</div><blockquote class=3D"gmail_=
quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1=
ex"><div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote">=
<div><div class=3D"h5">
<div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px =
0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-l=
eft-style:solid;padding-left:1ex">

<div dir=3D"ltr"><div><div><div><br></div><div>Keep-alive is fairly evil at=
 the infrastructure layer, and incents browsers, servers and carriers to do=
 things to the tear down connection quickly/prematurely.</div>
<div><br></div><div>Liveness-detection doesn&#39;t solve the server notific=
ation problem (which requires at least some level of high-level polling, th=
ough hopefully on a longer timescale), but keep in mind that in seeking per=
fection, one is likely to cause a significant less than perfect response fr=
om the system and the players within it.=C2=A0</div>



<div><br></div><div>(It is pretty annoying that we&#39;ve not developed goo=
d APIs or behaviors for doing decent things on mobile devices...)</div><spa=
n><font color=3D"#888888"><div><br></div><div>-=3DR</div></font></span></di=
v>


</div></div><div class=3D"gmail_extra"><br>
<br><div class=3D"gmail_quote"><div><div>On Tue, Feb 11, 2014 at 7:50 PM, T=
akeshi Yoshino <span dir=3D"ltr">&lt;<a href=3D"mailto:tyoshino@google.com"=
 target=3D"_blank">tyoshino@google.com</a>&gt;</span> wrote:<br></div>

</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;b=
order-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:s=
olid;padding-left:1ex"><div><div>
<div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote"><div=
>On Thu, Jan 9, 2014 at 9:13 PM, Tobias Oberstein <span dir=3D"ltr">&lt;<a =
href=3D"mailto:tobias.oberstein@tavendo.de" target=3D"_blank">tobias.oberst=
ein@tavendo.de</a>&gt;</span> wrote:<br>





<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div>&gt; The WS API is explicit about the fact that there=
 is no way from JavaScript to<br>



&gt; generate PINGs or check for PONGs. So do I just create my own, applica=
tion<br>
&gt; level equivalent, and send a ping-like message every so often and assu=
me<br>
<br>
</div>You can have a WS server that does automatic ping-pong for connection=
 keep-alive/loss-detection.<br>
<br></blockquote><div><br></div></div><div>This works for server. For clien=
t, app level work is needed.</div><div><div>=C2=A0</div><blockquote class=
=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;bo=
rder-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">




Or you can do it on app level from JS.<br>
<div><br>
&gt; the connection is dead when I don&#39;t get a reply &quot;soonish&quot=
;? Are there libraries<br>
&gt; that support such a mechanism? Or is every browser supported to regula=
rly<br>
&gt; PINGPONG with the server, and call the onclose function when there&#39=
;s no<br>
<br>
</div>As far as I know, there is no JS API where you could request the brow=
ser to do automatic ping-pong upon opening a WS connection.<br>
<div><div><br>
&gt; answer?<br>
&gt;<br></div></div></blockquote><div><br></div></div><div>Right. We&#39;ve=
 been saying that we should start discussing such a change on the API, but =
haven&#39;t.</div><div><br></div><div>// Send a ping (w/o body) every 5 sec=
</div>





<div>ws.pingInterval =3D 5;</div><div>// _Close the WebSocket connection_ w=
hen it passed 10 sec without receiving any byte (not limited to PONG)</div>=
<div>ws.timeout =3D 10;</div><div>=C2=A0</div></div></div></div>
<br></div></div><div>_______________________________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
<br></div></blockquote></div><br></div>
</blockquote></div></div></div><br></div></div>
</blockquote></div><br></div></div>

--001a113ac3d0fe15e304f24ecf5a--


From nobody Fri Feb 14 02:07:07 2014
Return-Path: <ricea@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id A67541A01EC for <hybi@ietfa.amsl.com>; Fri, 14 Feb 2014 02:07:05 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.326
X-Spam-Level: 
X-Spam-Status: No, score=-1.326 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, J_CHICKENPOX_27=0.6, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id BI8UavttP-S2 for <hybi@ietfa.amsl.com>; Fri, 14 Feb 2014 02:07:03 -0800 (PST)
Received: from mail-oa0-x229.google.com (mail-oa0-x229.google.com [IPv6:2607:f8b0:4003:c02::229]) by ietfa.amsl.com (Postfix) with ESMTP id 2FD2B1A00F9 for <hybi@ietf.org>; Fri, 14 Feb 2014 02:07:03 -0800 (PST)
Received: by mail-oa0-f41.google.com with SMTP id j17so14420514oag.0 for <hybi@ietf.org>; Fri, 14 Feb 2014 02:07:01 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=2cY++ZBYUojGLa55e0Vwn6oM/3d/5zDUjtOdvRek4nI=; b=Yi1wsgQqp1GY7I9qnI0RwIgFlsM1noE8a6BK6g+axu9D/hFv+wL6VZmi669n/q2Jgp vhZCmHkAZBw0/67ypiyqKIS3X3rU1/IIB8rRDgkFaP/7ARXmgfrUX9Wo25Z0Gy32XVOQ 6dpkcyelnwnadi5MyMXieksEct+bmypHcqPSy2wEdwzxkKRfin1b8BeOTcym4S77aNtx vQvjgF08frRiYe9XktkY6FegFYkRL2ckwF/761RDDpTPRCzUUlaf+3aHy1ynEe0dzFk0 3kQAagzQs+/MNBVownhWV40REIM6px4OvJ1qn5bWUP4zKGo8sgemLqB9i8KETkvERgEx g7TQ==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=2cY++ZBYUojGLa55e0Vwn6oM/3d/5zDUjtOdvRek4nI=; b=GM0MHPpYyvFNDOaxpNfgPanu0qB7Qrq2C/ow0aVG6o8N/kxK8e5XTBbfcNXwidncut lcPQ3Tk9L1ksHgjAl5TPX5+hRbC5arr+51mIOe7/JAl/M7JNKEK4lEyqIIsHdqVqEGWu QWi//1Vg0RYSg2fBUO2UVyE01tjv3j/h0v+BpZ/MHH2+dIeFxdwUfdAJpjTXisSEccql BSQb8GTvYhUBNf1t8kTe1J8tkUw7U9jHqlOEiFlgn2sJac9D/5MrZzgGx98n/Zp2Z9+B 1ScPqKNPLKzUyPKODL+2hVnHmCsFy/na7HXy2tydMgpVSSg+vMlwYyWvi8Vp9s0HahlK Iitg==
X-Gm-Message-State: ALoCoQkYniX8+aQtHYW8x4dnuvalAy3tbN/XzPLLC7W1xQfS+2D+bzrJMVjRCtnoPsnxkWmm+jmKkrCXyMn1G7LEdkZXUGYlvLi5Y5bWCrDdQLNTo8f8OSa/mMeuQjqRv7llR/Ta1HksOiXBYfU9dp+3LgbkKdHHsXKl6Nm130EuGjBHEqkAjQ7Owa11L5wTCXF2QPUQHbsc
MIME-Version: 1.0
X-Received: by 10.60.55.227 with SMTP id v3mr219487oep.67.1392372421582; Fri, 14 Feb 2014 02:07:01 -0800 (PST)
Received: by 10.182.118.41 with HTTP; Fri, 14 Feb 2014 02:07:01 -0800 (PST)
In-Reply-To: <CAGzyod4Ga0Gy9f9f+chufqhWa4XswmOdAtNU7rkXY6ZzayTa-g@mail.gmail.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E0@EXVMBX020-12.exch020.serverdata.net> <52CE8D59.3070802@it.aoyama.ac.jp> <634914A010D0B943A035D226786325D4446BE403EB@EXVMBX020-12.exch020.serverdata.net> <CAH9hSJZfEoYv9uxH31R-s1zbXy7PqQW7MsA475VJJdFrjDhj_w@mail.gmail.com> <CAGzyod6i6HnOETSm1j_nhbQDp-129g06ZhR1Ey=KdGzzCQ+hJw@mail.gmail.com> <CAH9hSJYaRLbXsXXx=dcTyDqC+GJ8vj=E=oDg6vSQ+549DniV3Q@mail.gmail.com> <CAGzyod4Ga0Gy9f9f+chufqhWa4XswmOdAtNU7rkXY6ZzayTa-g@mail.gmail.com>
Date: Fri, 14 Feb 2014 19:07:01 +0900
Message-ID: <CAHixhFpzoToMF1L+bchnLfpWd2u2mB9-1Z41vPMs=aO_13W2Hg@mail.gmail.com>
From: Adam Rice <ricea@google.com>
To: Roberto Peon <fenix@google.com>
Content-Type: multipart/alternative; boundary=089e0111bbd695781404f25af77e
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/zPTxSDOASVvb9MQlAo9rzkwPhIs
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Detection of broken WS connections
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 14 Feb 2014 10:07:05 -0000

--089e0111bbd695781404f25af77e
Content-Type: text/plain; charset=UTF-8

I don't think anyone has yet made the point that WebSocket connections
cannot be assumed to be stateless, so browsers cannot implement transparent
reconnection without an extension with some server-side state.

I have not so far seen a compelling argument that reconnection logic is
better implemented as a protocol extension rather than at the application
level. Hypothetically mobile networks could provide some kind of WebSocket
proxy with suspend/resume capabilities, but so far no such thing exists.

Currently WebSocket applications can control keepalive behaviour by sending
(or not) pings from the server.

I see two potential areas for improvement:
1) Client side detection of loss of connection
2) More efficient liveness detection

1) could be resolved with something as simple as an "expect_ping_when_idle"
extension which told the client about whatever keepalive policy the server
was attempting to enforce.

Roberto seems to be implying that 2) may in fact be counter-productive,
although I'm not sure I fully understand the argument.

Adam Rice


On 14 February 2014 04:36, Roberto Peon <fenix@google.com> wrote:

>
>
>
> On Wed, Feb 12, 2014 at 9:14 PM, Takeshi Yoshino <tyoshino@google.com>wrote:
>
>> On Thu, Feb 13, 2014 at 2:08 AM, Roberto Peon <fenix@google.com> wrote:
>>
>>> On should steer away from keep-alive (which is anathema on mobile
>>> links), and use liveness detection when one needs to solve these problems.
>>> (I believe that was what you essentially proposed, Martin)
>>>
>>>
>> I want to confirm my understanding on your terms. Do you mean that
>> - it's kinda illusion that we are able to keep an end-to-end connectivity
>> alive
>> - but should just have a way to detect the peer is dead when wanted
>>
>
> yup.
> Essentially, we don't want traffic black-holed (because the connection is
> dead but we haven't detected it), and we don't want to incent "the network"
> to kill connections unnecessarily (this is particularly an issue on shared
> bandwidth mediums like mobile networks).
>
>
>>
>>
>>> A PING (application-layer equivalent if necessary) should be
>>> automatically generated when the client has been idle for some time and
>>> then has any message to send. If the PONG is not received within some
>>> reasonable timeout (preferably ~2-3RT or so, or within some other
>>> human-scale constant), the connection should be assumed dead, and another
>>> should be created.
>>>
>>> imho, this should be done for you by the browser (not that it
>>> necessarily *is*, but it *should*)
>>>
>>
>> Thinking of what browser vendors can do at this point:
>> a) have built-in default-on ping-generate/timeout-close (try to do
>> something intelligent like TCP's RTO calculation?) feature in browser. one
>> concern is that people may have skipped implementation of ping/pong
>> b) same as (a) but default-off and can be turned on by calling some
>> method on WebSocket object
>> c) have one-shot ping sending method on WebSocket object with timeout arg
>> d) have one-shot ping sending method and pong received event. Sounds
>> overkilling to me
>> e) periodical version of (c) or (d)
>>
>
> We have implemenation experience for (a) for HTTP/2/SPDY, and it seems to
> work pretty well.
> HTTP2/SPDY has the same problem: if the connection is dead, it is a poor
> user experience to attempt to keep sending all traffic down it!
>
> The design rationale behind the choice of (a) was:
> it transparently fixed the biggest problem case without requiring any user
> involvement or knowledge.
> It was important to use ping, because ping/pong requires no interaction
> with the application-layer logic, and thus is most likely to be consistent
> and have timing similar to TCP's PING.
>
> -=R
>
>
>>
>>
>>>
>>> Keep-alive is fairly evil at the infrastructure layer, and incents
>>> browsers, servers and carriers to do things to the tear down connection
>>> quickly/prematurely.
>>>
>>> Liveness-detection doesn't solve the server notification problem (which
>>> requires at least some level of high-level polling, though hopefully on a
>>> longer timescale), but keep in mind that in seeking perfection, one is
>>> likely to cause a significant less than perfect response from the system
>>> and the players within it.
>>>
>>> (It is pretty annoying that we've not developed good APIs or behaviors
>>> for doing decent things on mobile devices...)
>>>
>>> -=R
>>>
>>>
>>> On Tue, Feb 11, 2014 at 7:50 PM, Takeshi Yoshino <tyoshino@google.com>wrote:
>>>
>>>> On Thu, Jan 9, 2014 at 9:13 PM, Tobias Oberstein <
>>>> tobias.oberstein@tavendo.de> wrote:
>>>>
>>>>> > The WS API is explicit about the fact that there is no way from
>>>>> JavaScript to
>>>>> > generate PINGs or check for PONGs. So do I just create my own,
>>>>> application
>>>>> > level equivalent, and send a ping-like message every so often and
>>>>> assume
>>>>>
>>>>> You can have a WS server that does automatic ping-pong for connection
>>>>> keep-alive/loss-detection.
>>>>>
>>>>>
>>>> This works for server. For client, app level work is needed.
>>>>
>>>>
>>>>> Or you can do it on app level from JS.
>>>>>
>>>>> > the connection is dead when I don't get a reply "soonish"? Are there
>>>>> libraries
>>>>> > that support such a mechanism? Or is every browser supported to
>>>>> regularly
>>>>> > PINGPONG with the server, and call the onclose function when there's
>>>>> no
>>>>>
>>>>> As far as I know, there is no JS API where you could request the
>>>>> browser to do automatic ping-pong upon opening a WS connection.
>>>>>
>>>>> > answer?
>>>>> >
>>>>>
>>>>
>>>> Right. We've been saying that we should start discussing such a change
>>>> on the API, but haven't.
>>>>
>>>> // Send a ping (w/o body) every 5 sec
>>>> ws.pingInterval = 5;
>>>> // _Close the WebSocket connection_ when it passed 10 sec without
>>>> receiving any byte (not limited to PONG)
>>>> ws.timeout = 10;
>>>>
>>>>
>>>> _______________________________________________
>>>> hybi mailing list
>>>> hybi@ietf.org
>>>> https://www.ietf.org/mailman/listinfo/hybi
>>>>
>>>>
>>>
>>
>
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi
>
>

--089e0111bbd695781404f25af77e
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">I don&#39;t think anyone has yet made the point that WebSo=
cket connections cannot be assumed to be stateless, so browsers cannot impl=
ement transparent reconnection without an extension with some server-side s=
tate.<div>
<br></div><div>I have not so far seen a compelling argument that reconnecti=
on logic is better implemented as a protocol extension rather than at the a=
pplication level. Hypothetically mobile networks could provide some kind of=
 WebSocket proxy with suspend/resume capabilities, but so far no such thing=
 exists.</div>
<div><br></div><div>Currently WebSocket applications can control keepalive =
behaviour by sending (or not) pings from the server.</div><div><br></div><d=
iv>I see two potential areas for improvement:</div><div>1) Client side dete=
ction of loss of connection</div>
<div>2) More efficient liveness detection</div><div><br></div><div>1) could=
 be resolved with something as simple as an &quot;expect_ping_when_idle&quo=
t; extension which told the client about whatever keepalive policy the serv=
er was attempting to enforce.</div>
<div><br></div><div>Roberto seems to be implying that 2) may in fact be cou=
nter-productive, although I&#39;m not sure I fully understand the argument.=
</div><div><br></div><div>Adam Rice</div></div><div class=3D"gmail_extra">
<br><br><div class=3D"gmail_quote">On 14 February 2014 04:36, Roberto Peon =
<span dir=3D"ltr">&lt;<a href=3D"mailto:fenix@google.com" target=3D"_blank"=
>fenix@google.com</a>&gt;</span> wrote:<br><blockquote class=3D"gmail_quote=
" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<div dir=3D"ltr"><br><div class=3D"gmail_extra"><br><br><div class=3D"gmail=
_quote"><div class=3D"">On Wed, Feb 12, 2014 at 9:14 PM, Takeshi Yoshino <s=
pan dir=3D"ltr">&lt;<a href=3D"mailto:tyoshino@google.com" target=3D"_blank=
">tyoshino@google.com</a>&gt;</span> wrote:<br>

<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div dir=3D"ltr"><div class=3D"gmail_extra">=
<div class=3D"gmail_quote"><div>On Thu, Feb 13, 2014 at 2:08 AM, Roberto Pe=
on <span dir=3D"ltr">&lt;<a href=3D"mailto:fenix@google.com" target=3D"_bla=
nk">fenix@google.com</a>&gt;</span> wrote:<br>



<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div dir=3D"ltr">On should steer away from keep-alive (whi=
ch is anathema on mobile links), and use liveness detection when one needs =
to solve these problems.<div>



(I believe that was what you essentially proposed, Martin)<br>
<div><br></div></div></div></blockquote><div><br></div></div><div>I want to=
 confirm my understanding on your terms. Do you mean that</div><div>- it&#3=
9;s kinda illusion that we are able to keep an end-to-end connectivity aliv=
e</div>



<div>- but should just have a way to detect the peer is dead when wanted</d=
iv></div></div></div></blockquote><div><br></div></div><div>yup.</div><div>=
Essentially, we don&#39;t want traffic black-holed (because the connection =
is dead but we haven&#39;t detected it), and we don&#39;t want to incent &q=
uot;the network&quot; to kill connections unnecessarily (this is particular=
ly an issue on shared bandwidth mediums like mobile networks).</div>
<div class=3D"">
<div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8=
ex;border-left:1px #ccc solid;padding-left:1ex"><div dir=3D"ltr"><div class=
=3D"gmail_extra"><div class=3D"gmail_quote"><div><div>=C2=A0</div><blockquo=
te class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-widt=
h:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-le=
ft:1ex">



<div dir=3D"ltr"><div><div><div>A PING (application-layer equivalent if nec=
essary) should be automatically generated when the client has been idle for=
 some time and then has any message to send. If the PONG is not received wi=
thin some reasonable timeout (preferably ~2-3RT or so, or within some other=
 human-scale constant), the connection should be assumed dead, and another =
should be created.</div>





<div><br></div><div>imho, this should be done for you by the browser (not t=
hat it necessarily *is*, but it *should*)</div></div></div></div></blockquo=
te><div><br></div></div><div>Thinking of what browser vendors can do at thi=
s point:</div>



<div>a) have built-in default-on ping-generate/timeout-close (try to do som=
ething intelligent like TCP&#39;s RTO calculation?) feature in browser. one=
 concern is that people may have skipped implementation of ping/pong</div>



<div>b) same as (a) but default-off and can be turned on by calling some me=
thod on WebSocket object</div><div>c) have one-shot ping sending method on =
WebSocket object with timeout arg</div><div>d) have one-shot ping sending m=
ethod and pong received event. Sounds overkilling to me</div>



<div>e) periodical version of (c) or (d)</div></div></div></div></blockquot=
e></div><div><br>We have implemenation experience for (a) for HTTP/2/SPDY, =
and it seems to work pretty well.</div><div>HTTP2/SPDY has the same problem=
: if the connection is dead, it is a poor user experience to attempt to kee=
p sending all traffic down it!</div>

<div><br></div><div>The design rationale behind the choice of (a) was:</div=
><div>it transparently fixed the biggest problem case without requiring any=
 user involvement or knowledge.</div><div>It was important to use ping, bec=
ause ping/pong requires no interaction with the application-layer logic, an=
d thus is most likely to be consistent and have timing similar to TCP&#39;s=
 PING.</div>
<span class=3D"HOEnZb"><font color=3D"#888888">
<div><br></div><div>-=3DR</div></font></span><div><div class=3D"h5"><div>=
=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;bo=
rder-left:1px #ccc solid;padding-left:1ex"><div dir=3D"ltr"><div class=3D"g=
mail_extra"><div class=3D"gmail_quote">
<div><div>
<div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px =
0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-l=
eft-style:solid;padding-left:1ex">

<div dir=3D"ltr"><div><div><div><br></div><div>Keep-alive is fairly evil at=
 the infrastructure layer, and incents browsers, servers and carriers to do=
 things to the tear down connection quickly/prematurely.</div>
<div><br></div><div>Liveness-detection doesn&#39;t solve the server notific=
ation problem (which requires at least some level of high-level polling, th=
ough hopefully on a longer timescale), but keep in mind that in seeking per=
fection, one is likely to cause a significant less than perfect response fr=
om the system and the players within it.=C2=A0</div>




<div><br></div><div>(It is pretty annoying that we&#39;ve not developed goo=
d APIs or behaviors for doing decent things on mobile devices...)</div><spa=
n><font color=3D"#888888"><div><br></div><div>-=3DR</div></font></span></di=
v>



</div></div><div class=3D"gmail_extra"><br>
<br><div class=3D"gmail_quote"><div><div>On Tue, Feb 11, 2014 at 7:50 PM, T=
akeshi Yoshino <span dir=3D"ltr">&lt;<a href=3D"mailto:tyoshino@google.com"=
 target=3D"_blank">tyoshino@google.com</a>&gt;</span> wrote:<br></div>

</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;b=
order-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:s=
olid;padding-left:1ex"><div><div>
<div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote"><div=
>On Thu, Jan 9, 2014 at 9:13 PM, Tobias Oberstein <span dir=3D"ltr">&lt;<a =
href=3D"mailto:tobias.oberstein@tavendo.de" target=3D"_blank">tobias.oberst=
ein@tavendo.de</a>&gt;</span> wrote:<br>






<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div>&gt; The WS API is explicit about the fact that there=
 is no way from JavaScript to<br>




&gt; generate PINGs or check for PONGs. So do I just create my own, applica=
tion<br>
&gt; level equivalent, and send a ping-like message every so often and assu=
me<br>
<br>
</div>You can have a WS server that does automatic ping-pong for connection=
 keep-alive/loss-detection.<br>
<br></blockquote><div><br></div></div><div>This works for server. For clien=
t, app level work is needed.</div><div><div>=C2=A0</div><blockquote class=
=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;bo=
rder-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">





Or you can do it on app level from JS.<br>
<div><br>
&gt; the connection is dead when I don&#39;t get a reply &quot;soonish&quot=
;? Are there libraries<br>
&gt; that support such a mechanism? Or is every browser supported to regula=
rly<br>
&gt; PINGPONG with the server, and call the onclose function when there&#39=
;s no<br>
<br>
</div>As far as I know, there is no JS API where you could request the brow=
ser to do automatic ping-pong upon opening a WS connection.<br>
<div><div><br>
&gt; answer?<br>
&gt;<br></div></div></blockquote><div><br></div></div><div>Right. We&#39;ve=
 been saying that we should start discussing such a change on the API, but =
haven&#39;t.</div><div><br></div><div>// Send a ping (w/o body) every 5 sec=
</div>






<div>ws.pingInterval =3D 5;</div><div>// _Close the WebSocket connection_ w=
hen it passed 10 sec without receiving any byte (not limited to PONG)</div>=
<div>ws.timeout =3D 10;</div><div>=C2=A0</div></div></div></div>
<br></div></div><div>_______________________________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
<br></div></blockquote></div><br></div>
</blockquote></div></div></div><br></div></div>
</blockquote></div></div></div><br></div></div>
<br>_______________________________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
<br></blockquote></div><br></div>

--089e0111bbd695781404f25af77e--


From nobody Fri Feb 14 03:52:51 2014
Return-Path: <salvatore.loreto@ericsson.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 1BF351A0203 for <hybi@ietfa.amsl.com>; Fri, 14 Feb 2014 03:52:47 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -3.551
X-Spam-Level: 
X-Spam-Status: No, score=-3.551 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HELO_EQ_SE=0.35, MIME_8BIT_HEADER=0.3, RCVD_IN_DNSWL_MED=-2.3, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id jXTB_4VoX68h for <hybi@ietfa.amsl.com>; Fri, 14 Feb 2014 03:52:44 -0800 (PST)
Received: from mailgw1.ericsson.se (mailgw1.ericsson.se [193.180.251.45]) by ietfa.amsl.com (Postfix) with ESMTP id C3BE51A0204 for <hybi@ietf.org>; Fri, 14 Feb 2014 03:52:43 -0800 (PST)
X-AuditID: c1b4fb2d-b7f5d8e000002a7b-36-52fe0389b0d3
Received: from ESESSHC019.ericsson.se (Unknown_Domain [153.88.253.124]) by mailgw1.ericsson.se (Symantec Mail Security) with SMTP id AE.7F.10875.9830EF25; Fri, 14 Feb 2014 12:52:41 +0100 (CET)
Received: from ESESSMB109.ericsson.se ([169.254.9.236]) by ESESSHC019.ericsson.se ([153.88.183.75]) with mapi id 14.02.0387.000; Fri, 14 Feb 2014 12:52:41 +0100
From: Salvatore Loreto <salvatore.loreto@ericsson.com>
To: =?iso-8859-1?Q?Martin_J=2E_D=FCrst?= <duerst@it.aoyama.ac.jp>
Thread-Topic: [hybi] Detection of broken WS connections
Thread-Index: AQHPDTE+HEAGpMqP8EKcj7MJA6woXJq0y2GA
Date: Fri, 14 Feb 2014 11:52:40 +0000
Message-ID: <2A1EB399-A5CF-481C-81E8-74E70C7FE1F9@ericsson.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E0@EXVMBX020-12.exch020.serverdata.net> <52CE8D59.3070802@it.aoyama.ac.jp>
In-Reply-To: <52CE8D59.3070802@it.aoyama.ac.jp>
Accept-Language: en-US
Content-Language: en-US
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
x-originating-ip: [153.88.183.149]
Content-Type: text/plain; charset="iso-8859-1"
Content-ID: <9884CA0E5DDF4048AD39AE300ED8A2A4@ericsson.com>
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0
X-Brightmail-Tracker: H4sIAAAAAAAAA+NgFnrKLMWRmVeSWpSXmKPExsUyM+JvjW4n878gg9YH6hbtd6+wW7x/uY3J gcljyZKfTB5Nl2cwBTBFcdmkpOZklqUW6dslcGVcfK5fcJ6nYuKjaUwNjLO5uhg5OSQETCSW 3nzOCGGLSVy4t56ti5GLQ0jgEKPEus232CGcJYwSL1pfsIBUsQmYSTx/uIUZxBYRcJR4/ucS O4jNLKAscfXYCrAaYaCaxpY+dogac4kZdw6zQthGEn8/dIBtYxFQlZi+ehaYzStgL9F6/g0j xLITjBIHb3SCDeIU0Jc403eICcRmBDrv+6k1TBDLxCVuPZnPBHG2gMSSPeeZIWxRiZeP/7FC 2EoSaw9vZ4Go15O4MXUKG4RtLfHhawNUXFti2cLXzBBHCEqcnPmEZQKj+CwkK2YhaZ+FpH0W kvZZSNoXMLKuYmTPTczMSS833MQIjKqDW37r7mA8dU7kEKM0B4uSOO+Ht85BQgLpiSWp2amp BalF8UWlOanFhxiZODilGhgjudRSpPY5n7x5tObula+Pn72oEHdwVS1fdc6hfm7QtZudLAX3 CyZpsLhYVM+Jf7Wr8fp+w+sLqtW6rzTddZ+Xsi/EVeVR/P68XVbPw3b8n6iVw3vZ4eT+dxtf sFxwe+qVI8+4zbRoXdSZyxsYlYVWifa1HD29X943tl1p5SH7+9r5e7WDBOWVWIozEg21mIuK EwEj0m2WeAIAAA==
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/xKCog5CPagf2xBpAIT2GpZ3Uk_4
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Detection of broken WS connections
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 14 Feb 2014 11:52:47 -0000

at some point a couple of years ago=20
we were trying to work on something at the http level:
http://tools.ietf.org/id/draft-loreto-http-timeout-00.txt

but we didn't have enough energy at time to finalise it

/salvatore

On Jan 9, 2014, at 1:51 PM, Martin J. D=FCrst <duerst@it.aoyama.ac.jp> wrot=
e:

> This is a problem that I have already encountered in three different appl=
ications, and think there must/should be a well-known solution for. It's no=
t strictly standards-related, but I'm sending it to this lists because I th=
ink that's where the experts are.
>=20
> It's about detecting broken connections on the application level. Setting=
 up a WS connection and using it while it's working is very easy. However, =
what is supposed to happen when e.g. the server dies or the cable gets cut =
(directly or figuratively), or the PC is put to sleep for a while, and then=
 wakes up again?
>=20
> The WS API is explicit about the fact that there is no way from JavaScrip=
t to generate PINGs or check for PONGs. So do I just create my own, applica=
tion level equivalent, and send a ping-like message every so often and assu=
me the connection is dead when I don't get a reply "soonish"? Are there lib=
raries that support such a mechanism? Or is every browser supported to regu=
larly PINGPONG with the server, and call the onclose function when there's =
no answer?
>=20
> Any advice or pointers appreciated!
>=20
> Regards,   Martin.
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi


From nobody Fri Feb 14 04:01:49 2014
Return-Path: <salvatore.loreto@ericsson.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 242C31A021A for <hybi@ietfa.amsl.com>; Fri, 14 Feb 2014 04:01:46 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -3.25
X-Spam-Level: 
X-Spam-Status: No, score=-3.25 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HELO_EQ_SE=0.35, HTML_MESSAGE=0.001, J_CHICKENPOX_27=0.6, RCVD_IN_DNSWL_MED=-2.3, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id DtU8K2450Jl2 for <hybi@ietfa.amsl.com>; Fri, 14 Feb 2014 04:01:43 -0800 (PST)
Received: from mailgw1.ericsson.se (mailgw1.ericsson.se [193.180.251.45]) by ietfa.amsl.com (Postfix) with ESMTP id F200E1A0222 for <hybi@ietf.org>; Fri, 14 Feb 2014 04:01:41 -0800 (PST)
X-AuditID: c1b4fb2d-b7f5d8e000002a7b-25-52fe05a350e4
Received: from ESESSHC007.ericsson.se (Unknown_Domain [153.88.253.124]) by mailgw1.ericsson.se (Symantec Mail Security) with SMTP id 3F.E0.10875.3A50EF25; Fri, 14 Feb 2014 13:01:40 +0100 (CET)
Received: from ESESSMB109.ericsson.se ([169.254.9.236]) by ESESSHC007.ericsson.se ([153.88.183.39]) with mapi id 14.02.0387.000; Fri, 14 Feb 2014 13:01:39 +0100
From: Salvatore Loreto <salvatore.loreto@ericsson.com>
To: Adam Rice <ricea@google.com>
Thread-Topic: [hybi] Detection of broken WS connections
Thread-Index: AQHPDTE+HEAGpMqP8EKcj7MJA6woXJp8PVGAgDTil4CAAN8kAIAAyqEAgADxFoCAAPMYgIAAIAeA
Date: Fri, 14 Feb 2014 12:01:38 +0000
Message-ID: <027FE739-50E0-4480-A52F-6661A36D769F@ericsson.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E0@EXVMBX020-12.exch020.serverdata.net> <52CE8D59.3070802@it.aoyama.ac.jp> <634914A010D0B943A035D226786325D4446BE403EB@EXVMBX020-12.exch020.serverdata.net> <CAH9hSJZfEoYv9uxH31R-s1zbXy7PqQW7MsA475VJJdFrjDhj_w@mail.gmail.com> <CAGzyod6i6HnOETSm1j_nhbQDp-129g06ZhR1Ey=KdGzzCQ+hJw@mail.gmail.com> <CAH9hSJYaRLbXsXXx=dcTyDqC+GJ8vj=E=oDg6vSQ+549DniV3Q@mail.gmail.com> <CAGzyod4Ga0Gy9f9f+chufqhWa4XswmOdAtNU7rkXY6ZzayTa-g@mail.gmail.com> <CAHixhFpzoToMF1L+bchnLfpWd2u2mB9-1Z41vPMs=aO_13W2Hg@mail.gmail.com>
In-Reply-To: <CAHixhFpzoToMF1L+bchnLfpWd2u2mB9-1Z41vPMs=aO_13W2Hg@mail.gmail.com>
Accept-Language: en-US
Content-Language: en-US
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
x-originating-ip: [153.88.183.150]
Content-Type: multipart/alternative; boundary="_000_027FE73950E04480A52F6661A36D769Fericssoncom_"
MIME-Version: 1.0
X-Brightmail-Tracker: H4sIAAAAAAAAA+NgFvrMLMWRmVeSWpSXmKPExsUyM+Jvje4S1n9BBk99LP73GFi8f7mNyeLs dw8HZo8Fm0o9liz5yRTAFMVlk5Kak1mWWqRvl8CV8XPdRtaCd/2MFT9m5DQwzqvsYuTkkBAw kXj7YicLhC0mceHeerYuRi4OIYFDjBJzdv1nh3CWMEqcenyaHaSKTcBM4vnDLcwgtoiAgsSh w6eYQGxmAReJA1O/gNnCQDWNLX3sEDXmEjPuHGaFsKMkDu24CdbLIqAqsfL+W7A4r4C9RPOE fUwQy/6xSMx//gRsEKdAoMT+SZ/ABjECnff91BqoZeISt57MZ4I4W0BiyZ7zzBC2qMTLx/9Y IWwliRXbLzFC1CdL/JuzkAVimaDEyZlPWCYwis5CMmoWkrJZSMog4joSC3Z/YoOwtSWWLXzN DGOfOfAYqtdaonPVaXZkNQsYOVYxsucmZuaklxtuYgRG3MEtv3V3MJ46J3KIUZqDRUmc98Nb 5yAhgfTEktTs1NSC1KL4otKc1OJDjEwcnFINjAUCfab9r/bp/J1ezXbzX8D9tfM7r7jeFXFh P1zPtLzAca+9xkvTmTkpwXGOqhwp3cl/nrw1mBhSfPPdRZPTe1relT9r5+B3bLvEss1QwNxk 5e7NNyvudhYwfrl5akvUwYVZl/67e17cdN3zQFCedkPt+2JVp+Pi52OeWh6e2b16Z13AvFen pZRYijMSDbWYi4oTAfvtJ9iGAgAA
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/Yre_NYTnHisqLyE2BHA0-DdxMKM
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Detection of broken WS connections
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 14 Feb 2014 12:01:46 -0000

--_000_027FE73950E04480A52F6661A36D769Fericssoncom_
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable


On Feb 14, 2014, at 12:07 PM, Adam Rice <ricea@google.com<mailto:ricea@goog=
le.com>> wrote:

I don't think anyone has yet made the point that WebSocket connections cann=
ot be assumed to be stateless, so browsers cannot implement transparent rec=
onnection without an extension with some server-side state.

I have not so far seen a compelling argument that reconnection logic is bet=
ter implemented as a protocol extension rather than at the application leve=
l. Hypothetically mobile networks could provide some kind of WebSocket prox=
y with suspend/resume capabilities, but so far no such thing exists.

that is a great business idea, but unfortunately it would only work for Web=
Socket in clear (i.e. on port 80)

/Sal


Currently WebSocket applications can control keepalive behaviour by sending=
 (or not) pings from the server.

I see two potential areas for improvement:
1) Client side detection of loss of connection
2) More efficient liveness detection

1) could be resolved with something as simple as an "expect_ping_when_idle"=
 extension which told the client about whatever keepalive policy the server=
 was attempting to enforce.

Roberto seems to be implying that 2) may in fact be counter-productive, alt=
hough I'm not sure I fully understand the argument.

Adam Rice


On 14 February 2014 04:36, Roberto Peon <fenix@google.com<mailto:fenix@goog=
le.com>> wrote:



On Wed, Feb 12, 2014 at 9:14 PM, Takeshi Yoshino <tyoshino@google.com<mailt=
o:tyoshino@google.com>> wrote:
On Thu, Feb 13, 2014 at 2:08 AM, Roberto Peon <fenix@google.com<mailto:feni=
x@google.com>> wrote:
On should steer away from keep-alive (which is anathema on mobile links), a=
nd use liveness detection when one needs to solve these problems.
(I believe that was what you essentially proposed, Martin)


I want to confirm my understanding on your terms. Do you mean that
- it's kinda illusion that we are able to keep an end-to-end connectivity a=
live
- but should just have a way to detect the peer is dead when wanted

yup.
Essentially, we don't want traffic black-holed (because the connection is d=
ead but we haven't detected it), and we don't want to incent "the network" =
to kill connections unnecessarily (this is particularly an issue on shared =
bandwidth mediums like mobile networks).


A PING (application-layer equivalent if necessary) should be automatically =
generated when the client has been idle for some time and then has any mess=
age to send. If the PONG is not received within some reasonable timeout (pr=
eferably ~2-3RT or so, or within some other human-scale constant), the conn=
ection should be assumed dead, and another should be created.

imho, this should be done for you by the browser (not that it necessarily *=
is*, but it *should*)

Thinking of what browser vendors can do at this point:
a) have built-in default-on ping-generate/timeout-close (try to do somethin=
g intelligent like TCP's RTO calculation?) feature in browser. one concern =
is that people may have skipped implementation of ping/pong
b) same as (a) but default-off and can be turned on by calling some method =
on WebSocket object
c) have one-shot ping sending method on WebSocket object with timeout arg
d) have one-shot ping sending method and pong received event. Sounds overki=
lling to me
e) periodical version of (c) or (d)

We have implemenation experience for (a) for HTTP/2/SPDY, and it seems to w=
ork pretty well.
HTTP2/SPDY has the same problem: if the connection is dead, it is a poor us=
er experience to attempt to keep sending all traffic down it!

The design rationale behind the choice of (a) was:
it transparently fixed the biggest problem case without requiring any user =
involvement or knowledge.
It was important to use ping, because ping/pong requires no interaction wit=
h the application-layer logic, and thus is most likely to be consistent and=
 have timing similar to TCP's PING.

-=3DR



Keep-alive is fairly evil at the infrastructure layer, and incents browsers=
, servers and carriers to do things to the tear down connection quickly/pre=
maturely.

Liveness-detection doesn't solve the server notification problem (which req=
uires at least some level of high-level polling, though hopefully on a long=
er timescale), but keep in mind that in seeking perfection, one is likely t=
o cause a significant less than perfect response from the system and the pl=
ayers within it.

(It is pretty annoying that we've not developed good APIs or behaviors for =
doing decent things on mobile devices...)

-=3DR


On Tue, Feb 11, 2014 at 7:50 PM, Takeshi Yoshino <tyoshino@google.com<mailt=
o:tyoshino@google.com>> wrote:
On Thu, Jan 9, 2014 at 9:13 PM, Tobias Oberstein <tobias.oberstein@tavendo.=
de<mailto:tobias.oberstein@tavendo.de>> wrote:
> The WS API is explicit about the fact that there is no way from JavaScrip=
t to
> generate PINGs or check for PONGs. So do I just create my own, applicatio=
n
> level equivalent, and send a ping-like message every so often and assume

You can have a WS server that does automatic ping-pong for connection keep-=
alive/loss-detection.


This works for server. For client, app level work is needed.

Or you can do it on app level from JS.

> the connection is dead when I don't get a reply "soonish"? Are there libr=
aries
> that support such a mechanism? Or is every browser supported to regularly
> PINGPONG with the server, and call the onclose function when there's no

As far as I know, there is no JS API where you could request the browser to=
 do automatic ping-pong upon opening a WS connection.

> answer?
>

Right. We've been saying that we should start discussing such a change on t=
he API, but haven't.

// Send a ping (w/o body) every 5 sec
ws.pingInterval =3D 5;
// _Close the WebSocket connection_ when it passed 10 sec without receiving=
 any byte (not limited to PONG)
ws.timeout =3D 10;


_______________________________________________
hybi mailing list
hybi@ietf.org<mailto:hybi@ietf.org>
https://www.ietf.org/mailman/listinfo/hybi





_______________________________________________
hybi mailing list
hybi@ietf.org<mailto:hybi@ietf.org>
https://www.ietf.org/mailman/listinfo/hybi


_______________________________________________
hybi mailing list
hybi@ietf.org<mailto:hybi@ietf.org>
https://www.ietf.org/mailman/listinfo/hybi


--_000_027FE73950E04480A52F6661A36D769Fericssoncom_
Content-Type: text/html; charset="us-ascii"
Content-ID: <875DB163EA982145901DC88DB40D1D58@ericsson.com>
Content-Transfer-Encoding: quoted-printable

<html>
<head>
<meta http-equiv=3D"Content-Type" content=3D"text/html; charset=3Dus-ascii"=
>
</head>
<body style=3D"word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-lin=
e-break: after-white-space; ">
<br>
<div>
<div>On Feb 14, 2014, at 12:07 PM, Adam Rice &lt;<a href=3D"mailto:ricea@go=
ogle.com">ricea@google.com</a>&gt; wrote:</div>
<br class=3D"Apple-interchange-newline">
<blockquote type=3D"cite">
<div dir=3D"ltr">I don't think anyone has yet made the point that WebSocket=
 connections cannot be assumed to be stateless, so browsers cannot implemen=
t transparent reconnection without an extension with some server-side state=
.
<div><br>
</div>
<div>I have not so far seen a compelling argument that reconnection logic i=
s better implemented as a protocol extension rather than at the application=
 level. Hypothetically mobile networks could provide some kind of WebSocket=
 proxy with suspend/resume capabilities,
 but so far no such thing exists.</div>
</div>
</blockquote>
<div><br>
</div>
<div>that is a great business idea, but unfortunately it would only work fo=
r WebSocket in clear (i.e. on port 80)</div>
<div><br>
</div>
<div>/Sal</div>
<div><br>
</div>
<blockquote type=3D"cite">
<div dir=3D"ltr">
<div><br>
</div>
<div>Currently WebSocket applications can control keepalive behaviour by se=
nding (or not) pings from the server.</div>
<div><br>
</div>
<div>I see two potential areas for improvement:</div>
<div>1) Client side detection of loss of connection</div>
<div>2) More efficient liveness detection</div>
<div><br>
</div>
<div>1) could be resolved with something as simple as an &quot;expect_ping_=
when_idle&quot; extension which told the client about whatever keepalive po=
licy the server was attempting to enforce.</div>
<div><br>
</div>
<div>Roberto seems to be implying that 2) may in fact be counter-productive=
, although I'm not sure I fully understand the argument.</div>
<div><br>
</div>
<div>Adam Rice</div>
</div>
<div class=3D"gmail_extra"><br>
<br>
<div class=3D"gmail_quote">On 14 February 2014 04:36, Roberto Peon <span di=
r=3D"ltr">
&lt;<a href=3D"mailto:fenix@google.com" target=3D"_blank">fenix@google.com<=
/a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
<div dir=3D"ltr"><br>
<div class=3D"gmail_extra"><br>
<br>
<div class=3D"gmail_quote">
<div class=3D"">On Wed, Feb 12, 2014 at 9:14 PM, Takeshi Yoshino <span dir=
=3D"ltr">&lt;<a href=3D"mailto:tyoshino@google.com" target=3D"_blank">tyosh=
ino@google.com</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
<div dir=3D"ltr">
<div class=3D"gmail_extra">
<div class=3D"gmail_quote">
<div>On Thu, Feb 13, 2014 at 2:08 AM, Roberto Peon <span dir=3D"ltr">&lt;<a=
 href=3D"mailto:fenix@google.com" target=3D"_blank">fenix@google.com</a>&gt=
;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex">
<div dir=3D"ltr">On should steer away from keep-alive (which is anathema on=
 mobile links), and use liveness detection when one needs to solve these pr=
oblems.
<div>(I believe that was what you essentially proposed, Martin)<br>
<div><br>
</div>
</div>
</div>
</blockquote>
<div><br>
</div>
</div>
<div>I want to confirm my understanding on your terms. Do you mean that</di=
v>
<div>- it's kinda illusion that we are able to keep an end-to-end connectiv=
ity alive</div>
<div>- but should just have a way to detect the peer is dead when wanted</d=
iv>
</div>
</div>
</div>
</blockquote>
<div><br>
</div>
</div>
<div>yup.</div>
<div>Essentially, we don't want traffic black-holed (because the connection=
 is dead but we haven't detected it), and we don't want to incent &quot;the=
 network&quot; to kill connections unnecessarily (this is particularly an i=
ssue on shared bandwidth mediums like mobile
 networks).</div>
<div class=3D"">
<div>&nbsp;</div>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
<div dir=3D"ltr">
<div class=3D"gmail_extra">
<div class=3D"gmail_quote">
<div>
<div>&nbsp;</div>
<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex">
<div dir=3D"ltr">
<div>A PING (application-layer equivalent if necessary) should be automatic=
ally generated when the client has been idle for some time and then has any=
 message to send. If the PONG is not received within some reasonable timeou=
t (preferably ~2-3RT or so, or within
 some other human-scale constant), the connection should be assumed dead, a=
nd another should be created.</div>
<div><br>
</div>
<div>imho, this should be done for you by the browser (not that it necessar=
ily *is*, but it *should*)</div>
</div>
</blockquote>
<div><br>
</div>
</div>
<div>Thinking of what browser vendors can do at this point:</div>
<div>a) have built-in default-on ping-generate/timeout-close (try to do som=
ething intelligent like TCP's RTO calculation?) feature in browser. one con=
cern is that people may have skipped implementation of ping/pong</div>
<div>b) same as (a) but default-off and can be turned on by calling some me=
thod on WebSocket object</div>
<div>c) have one-shot ping sending method on WebSocket object with timeout =
arg</div>
<div>d) have one-shot ping sending method and pong received event. Sounds o=
verkilling to me</div>
<div>e) periodical version of (c) or (d)</div>
</div>
</div>
</div>
</blockquote>
</div>
<div><br>
We have implemenation experience for (a) for HTTP/2/SPDY, and it seems to w=
ork pretty well.</div>
<div>HTTP2/SPDY has the same problem: if the connection is dead, it is a po=
or user experience to attempt to keep sending all traffic down it!</div>
<div><br>
</div>
<div>The design rationale behind the choice of (a) was:</div>
<div>it transparently fixed the biggest problem case without requiring any =
user involvement or knowledge.</div>
<div>It was important to use ping, because ping/pong requires no interactio=
n with the application-layer logic, and thus is most likely to be consisten=
t and have timing similar to TCP's PING.</div>
<span class=3D"HOEnZb"><font color=3D"#888888">
<div><br>
</div>
<div>-=3DR</div>
</font></span>
<div>
<div class=3D"h5">
<div>&nbsp;</div>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
<div dir=3D"ltr">
<div class=3D"gmail_extra">
<div class=3D"gmail_quote">
<div>
<div>&nbsp;</div>
<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex">
<div dir=3D"ltr">
<div>
<div><br>
</div>
<div>Keep-alive is fairly evil at the infrastructure layer, and incents bro=
wsers, servers and carriers to do things to the tear down connection quickl=
y/prematurely.</div>
<div><br>
</div>
<div>Liveness-detection doesn't solve the server notification problem (whic=
h requires at least some level of high-level polling, though hopefully on a=
 longer timescale), but keep in mind that in seeking perfection, one is lik=
ely to cause a significant less
 than perfect response from the system and the players within it.&nbsp;</di=
v>
<div><br>
</div>
<div>(It is pretty annoying that we've not developed good APIs or behaviors=
 for doing decent things on mobile devices...)</div>
<span><font color=3D"#888888">
<div><br>
</div>
<div>-=3DR</div>
</font></span></div>
</div>
<div class=3D"gmail_extra"><br>
<br>
<div class=3D"gmail_quote">
<div>
<div>On Tue, Feb 11, 2014 at 7:50 PM, Takeshi Yoshino <span dir=3D"ltr">&lt=
;<a href=3D"mailto:tyoshino@google.com" target=3D"_blank">tyoshino@google.c=
om</a>&gt;</span> wrote:<br>
</div>
</div>
<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex">
<div>
<div dir=3D"ltr">
<div class=3D"gmail_extra">
<div class=3D"gmail_quote">
<div>On Thu, Jan 9, 2014 at 9:13 PM, Tobias Oberstein <span dir=3D"ltr">&lt=
;<a href=3D"mailto:tobias.oberstein@tavendo.de" target=3D"_blank">tobias.ob=
erstein@tavendo.de</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex">
<div>&gt; The WS API is explicit about the fact that there is no way from J=
avaScript to<br>
&gt; generate PINGs or check for PONGs. So do I just create my own, applica=
tion<br>
&gt; level equivalent, and send a ping-like message every so often and assu=
me<br>
<br>
</div>
You can have a WS server that does automatic ping-pong for connection keep-=
alive/loss-detection.<br>
<br>
</blockquote>
<div><br>
</div>
</div>
<div>This works for server. For client, app level work is needed.</div>
<div>
<div>&nbsp;</div>
<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex">
Or you can do it on app level from JS.<br>
<div><br>
&gt; the connection is dead when I don't get a reply &quot;soonish&quot;? A=
re there libraries<br>
&gt; that support such a mechanism? Or is every browser supported to regula=
rly<br>
&gt; PINGPONG with the server, and call the onclose function when there's n=
o<br>
<br>
</div>
As far as I know, there is no JS API where you could request the browser to=
 do automatic ping-pong upon opening a WS connection.<br>
<div><br>
&gt; answer?<br>
&gt;<br>
</div>
</blockquote>
<div><br>
</div>
</div>
<div>Right. We've been saying that we should start discussing such a change=
 on the API, but haven't.</div>
<div><br>
</div>
<div>// Send a ping (w/o body) every 5 sec</div>
<div>ws.pingInterval =3D 5;</div>
<div>// _Close the WebSocket connection_ when it passed 10 sec without rece=
iving any byte (not limited to PONG)</div>
<div>ws.timeout =3D 10;</div>
<div>&nbsp;</div>
</div>
</div>
</div>
<br>
</div>
<div>_______________________________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
<br>
</div>
</blockquote>
</div>
<br>
</div>
</blockquote>
</div>
</div>
<br>
</div>
</div>
</blockquote>
</div>
</div>
</div>
<br>
</div>
</div>
<br>
_______________________________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
<br>
</blockquote>
</div>
<br>
</div>
_______________________________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org">hybi@ietf.org</a><br>
https://www.ietf.org/mailman/listinfo/hybi<br>
</blockquote>
</div>
<br>
</body>
</html>

--_000_027FE73950E04480A52F6661A36D769Fericssoncom_--


From nobody Fri Feb 14 11:21:18 2014
Return-Path: <fenix@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id E51701A0309 for <hybi@ietfa.amsl.com>; Fri, 14 Feb 2014 11:21:13 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.326
X-Spam-Level: 
X-Spam-Status: No, score=-1.326 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, J_CHICKENPOX_27=0.6, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id evseMPlWaCW0 for <hybi@ietfa.amsl.com>; Fri, 14 Feb 2014 11:21:10 -0800 (PST)
Received: from mail-qc0-x235.google.com (mail-qc0-x235.google.com [IPv6:2607:f8b0:400d:c01::235]) by ietfa.amsl.com (Postfix) with ESMTP id 5203E1A0122 for <hybi@ietf.org>; Fri, 14 Feb 2014 11:21:10 -0800 (PST)
Received: by mail-qc0-f181.google.com with SMTP id e9so20547184qcy.40 for <hybi@ietf.org>; Fri, 14 Feb 2014 11:21:08 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=7b9KbzloA4l934vzEdPHKj2zmOUFzZdQJ9fzok5R4cg=; b=TF4Dp6rEhsAjrZm1jUWuJseRJwY74r2OwF8HCDgZ77LnDAQJRzYXmXINyy3ax3NhEr fOaQ4AR+bD1MwF9inLouCuZ59qZSqxscmEzxalmoryYrWvCrIuZ3bCv9zNNklrhQhPUK v+g7HkPxNkQqGOYdG02OkN6U47r8qdMj8aK8b1/VyNVxGmcxsMKSJegS2sZU2M3z9Mdb MITIhZ3XufsdowggSOIinR3qfR0das092HeQZplAtGpEryz+dHUXRpB/9712JtQCHLmq pzSyZyhbW5bEyTstSPU4MH7+9mJcKMa2gvO1qt6V0vjVnyVlGX8iDhJ+BahnGIp+b7vL SdsQ==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=7b9KbzloA4l934vzEdPHKj2zmOUFzZdQJ9fzok5R4cg=; b=LM5OxtPo3xQXWCGEWSrb/BNxwqfxCEUTNdsy6wb2zC5uImtwAYEWL5yMdTgHjx3qR4 BofNMChDawUze+s2xfEiJwdzeKFjyteQzZr7bwSfy7OCt+FMnpP3sa20wtcoU1m7f1kW iD0fn797qlWFFSv3LFdOUsNjIk4bcdgitorM2Yi6FXLglJnueNqzkfpHUVPKPGJWS8qu a4WSFRkUVvGArhWbGSW0RbcOEmu9f2INY8BegX65pSjKct/YK3SK+cCcANpMreoKOomi i3nZLbv4CDxCkidJpSi7HGoLcVThsM84Fz5xhzuOexRGpcJ3Gzvv/m9c76icfMR9nh8a AcVg==
X-Gm-Message-State: ALoCoQmBg5DWnlqwpLDVnNuP28HrLF5WLDIP8hT7fFVTznccc9hydxrgLpwfM0hSNdxJ99Usn2y4RA0q25iAja6OWk4rYyaxXWfNticKSCHsZhmy26YrVsiz2bOGN2Rf0JbA8i8D+t8q15bNlV9TjWOTbr4O7bxaPBo936XpN8+gXAbtmNro6MgovgEgrBckG2B2LUcnJC5G
MIME-Version: 1.0
X-Received: by 10.229.139.199 with SMTP id f7mr16341750qcu.2.1392405668499; Fri, 14 Feb 2014 11:21:08 -0800 (PST)
Received: by 10.229.13.7 with HTTP; Fri, 14 Feb 2014 11:21:08 -0800 (PST)
In-Reply-To: <CAHixhFpzoToMF1L+bchnLfpWd2u2mB9-1Z41vPMs=aO_13W2Hg@mail.gmail.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E0@EXVMBX020-12.exch020.serverdata.net> <52CE8D59.3070802@it.aoyama.ac.jp> <634914A010D0B943A035D226786325D4446BE403EB@EXVMBX020-12.exch020.serverdata.net> <CAH9hSJZfEoYv9uxH31R-s1zbXy7PqQW7MsA475VJJdFrjDhj_w@mail.gmail.com> <CAGzyod6i6HnOETSm1j_nhbQDp-129g06ZhR1Ey=KdGzzCQ+hJw@mail.gmail.com> <CAH9hSJYaRLbXsXXx=dcTyDqC+GJ8vj=E=oDg6vSQ+549DniV3Q@mail.gmail.com> <CAGzyod4Ga0Gy9f9f+chufqhWa4XswmOdAtNU7rkXY6ZzayTa-g@mail.gmail.com> <CAHixhFpzoToMF1L+bchnLfpWd2u2mB9-1Z41vPMs=aO_13W2Hg@mail.gmail.com>
Date: Fri, 14 Feb 2014 11:21:08 -0800
Message-ID: <CAGzyod5OgpYR=pVpa5HVjhj=VJYb0uPUJEkONUrLGB7x2fe8Sw@mail.gmail.com>
From: Roberto Peon <fenix@google.com>
To: Adam Rice <ricea@google.com>
Content-Type: multipart/alternative; boundary=001a11c3e45441340704f262b5ce
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/QxQGG5tVfwjhwnEdAddHICf_4ZA
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Detection of broken WS connections
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 14 Feb 2014 19:21:14 -0000

--001a11c3e45441340704f262b5ce
Content-Type: text/plain; charset=UTF-8

Basically, keep-alives ties up enough network resources and/or device
battery life that OS providers and network providers have strong incentive
to consider a connection idle (and close it/blackhole it) for smaller and
smaller time periods, eventually resulting in an arms race which increases
inefficiencies everywhere.

-=R


On Fri, Feb 14, 2014 at 2:07 AM, Adam Rice <ricea@google.com> wrote:

> I don't think anyone has yet made the point that WebSocket connections
> cannot be assumed to be stateless, so browsers cannot implement transparent
> reconnection without an extension with some server-side state.
>
> I have not so far seen a compelling argument that reconnection logic is
> better implemented as a protocol extension rather than at the application
> level. Hypothetically mobile networks could provide some kind of WebSocket
> proxy with suspend/resume capabilities, but so far no such thing exists.
>
> Currently WebSocket applications can control keepalive behaviour by
> sending (or not) pings from the server.
>
> I see two potential areas for improvement:
> 1) Client side detection of loss of connection
> 2) More efficient liveness detection
>
> 1) could be resolved with something as simple as an
> "expect_ping_when_idle" extension which told the client about whatever
> keepalive policy the server was attempting to enforce.
>
> Roberto seems to be implying that 2) may in fact be counter-productive,
> although I'm not sure I fully understand the argument.
>
> Adam Rice
>
>
> On 14 February 2014 04:36, Roberto Peon <fenix@google.com> wrote:
>
>>
>>
>>
>> On Wed, Feb 12, 2014 at 9:14 PM, Takeshi Yoshino <tyoshino@google.com>wrote:
>>
>>> On Thu, Feb 13, 2014 at 2:08 AM, Roberto Peon <fenix@google.com> wrote:
>>>
>>>> On should steer away from keep-alive (which is anathema on mobile
>>>> links), and use liveness detection when one needs to solve these problems.
>>>> (I believe that was what you essentially proposed, Martin)
>>>>
>>>>
>>> I want to confirm my understanding on your terms. Do you mean that
>>> - it's kinda illusion that we are able to keep an end-to-end
>>> connectivity alive
>>> - but should just have a way to detect the peer is dead when wanted
>>>
>>
>> yup.
>> Essentially, we don't want traffic black-holed (because the connection is
>> dead but we haven't detected it), and we don't want to incent "the network"
>> to kill connections unnecessarily (this is particularly an issue on shared
>> bandwidth mediums like mobile networks).
>>
>>
>>>
>>>
>>>> A PING (application-layer equivalent if necessary) should be
>>>> automatically generated when the client has been idle for some time and
>>>> then has any message to send. If the PONG is not received within some
>>>> reasonable timeout (preferably ~2-3RT or so, or within some other
>>>> human-scale constant), the connection should be assumed dead, and another
>>>> should be created.
>>>>
>>>> imho, this should be done for you by the browser (not that it
>>>> necessarily *is*, but it *should*)
>>>>
>>>
>>> Thinking of what browser vendors can do at this point:
>>> a) have built-in default-on ping-generate/timeout-close (try to do
>>> something intelligent like TCP's RTO calculation?) feature in browser. one
>>> concern is that people may have skipped implementation of ping/pong
>>> b) same as (a) but default-off and can be turned on by calling some
>>> method on WebSocket object
>>> c) have one-shot ping sending method on WebSocket object with timeout arg
>>> d) have one-shot ping sending method and pong received event. Sounds
>>> overkilling to me
>>> e) periodical version of (c) or (d)
>>>
>>
>> We have implemenation experience for (a) for HTTP/2/SPDY, and it seems to
>> work pretty well.
>> HTTP2/SPDY has the same problem: if the connection is dead, it is a poor
>> user experience to attempt to keep sending all traffic down it!
>>
>> The design rationale behind the choice of (a) was:
>> it transparently fixed the biggest problem case without requiring any
>> user involvement or knowledge.
>> It was important to use ping, because ping/pong requires no interaction
>> with the application-layer logic, and thus is most likely to be consistent
>> and have timing similar to TCP's PING.
>>
>> -=R
>>
>>
>>>
>>>
>>>>
>>>> Keep-alive is fairly evil at the infrastructure layer, and incents
>>>> browsers, servers and carriers to do things to the tear down connection
>>>> quickly/prematurely.
>>>>
>>>> Liveness-detection doesn't solve the server notification problem (which
>>>> requires at least some level of high-level polling, though hopefully on a
>>>> longer timescale), but keep in mind that in seeking perfection, one is
>>>> likely to cause a significant less than perfect response from the system
>>>> and the players within it.
>>>>
>>>> (It is pretty annoying that we've not developed good APIs or behaviors
>>>> for doing decent things on mobile devices...)
>>>>
>>>> -=R
>>>>
>>>>
>>>> On Tue, Feb 11, 2014 at 7:50 PM, Takeshi Yoshino <tyoshino@google.com>wrote:
>>>>
>>>>> On Thu, Jan 9, 2014 at 9:13 PM, Tobias Oberstein <
>>>>> tobias.oberstein@tavendo.de> wrote:
>>>>>
>>>>>> > The WS API is explicit about the fact that there is no way from
>>>>>> JavaScript to
>>>>>> > generate PINGs or check for PONGs. So do I just create my own,
>>>>>> application
>>>>>> > level equivalent, and send a ping-like message every so often and
>>>>>> assume
>>>>>>
>>>>>> You can have a WS server that does automatic ping-pong for connection
>>>>>> keep-alive/loss-detection.
>>>>>>
>>>>>>
>>>>> This works for server. For client, app level work is needed.
>>>>>
>>>>>
>>>>>> Or you can do it on app level from JS.
>>>>>>
>>>>>> > the connection is dead when I don't get a reply "soonish"? Are
>>>>>> there libraries
>>>>>> > that support such a mechanism? Or is every browser supported to
>>>>>> regularly
>>>>>> > PINGPONG with the server, and call the onclose function when
>>>>>> there's no
>>>>>>
>>>>>> As far as I know, there is no JS API where you could request the
>>>>>> browser to do automatic ping-pong upon opening a WS connection.
>>>>>>
>>>>>> > answer?
>>>>>> >
>>>>>>
>>>>>
>>>>> Right. We've been saying that we should start discussing such a change
>>>>> on the API, but haven't.
>>>>>
>>>>> // Send a ping (w/o body) every 5 sec
>>>>> ws.pingInterval = 5;
>>>>> // _Close the WebSocket connection_ when it passed 10 sec without
>>>>> receiving any byte (not limited to PONG)
>>>>> ws.timeout = 10;
>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> hybi mailing list
>>>>> hybi@ietf.org
>>>>> https://www.ietf.org/mailman/listinfo/hybi
>>>>>
>>>>>
>>>>
>>>
>>
>> _______________________________________________
>> hybi mailing list
>> hybi@ietf.org
>> https://www.ietf.org/mailman/listinfo/hybi
>>
>>
>

--001a11c3e45441340704f262b5ce
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Basically, keep-alives ties up enough network resources an=
d/or device battery life that OS providers and network providers have stron=
g incentive to consider a connection idle (and close it/blackhole it) for s=
maller and smaller time periods, eventually resulting in an arms race which=
 increases inefficiencies everywhere.<div>
<br></div><div>-=3DR</div></div><div class=3D"gmail_extra"><br><br><div cla=
ss=3D"gmail_quote">On Fri, Feb 14, 2014 at 2:07 AM, Adam Rice <span dir=3D"=
ltr">&lt;<a href=3D"mailto:ricea@google.com" target=3D"_blank">ricea@google=
.com</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div dir=3D"ltr">I don&#39;t think anyone ha=
s yet made the point that WebSocket connections cannot be assumed to be sta=
teless, so browsers cannot implement transparent reconnection without an ex=
tension with some server-side state.<div>

<br></div><div>I have not so far seen a compelling argument that reconnecti=
on logic is better implemented as a protocol extension rather than at the a=
pplication level. Hypothetically mobile networks could provide some kind of=
 WebSocket proxy with suspend/resume capabilities, but so far no such thing=
 exists.</div>

<div><br></div><div>Currently WebSocket applications can control keepalive =
behaviour by sending (or not) pings from the server.</div><div><br></div><d=
iv>I see two potential areas for improvement:</div><div>1) Client side dete=
ction of loss of connection</div>

<div>2) More efficient liveness detection</div><div><br></div><div>1) could=
 be resolved with something as simple as an &quot;expect_ping_when_idle&quo=
t; extension which told the client about whatever keepalive policy the serv=
er was attempting to enforce.</div>

<div><br></div><div>Roberto seems to be implying that 2) may in fact be cou=
nter-productive, although I&#39;m not sure I fully understand the argument.=
</div><span class=3D"HOEnZb"><font color=3D"#888888"><div><br></div><div>Ad=
am Rice</div>
</font></span></div><div class=3D"HOEnZb"><div class=3D"h5"><div class=3D"g=
mail_extra">
<br><br><div class=3D"gmail_quote">On 14 February 2014 04:36, Roberto Peon =
<span dir=3D"ltr">&lt;<a href=3D"mailto:fenix@google.com" target=3D"_blank"=
>fenix@google.com</a>&gt;</span> wrote:<br><blockquote class=3D"gmail_quote=
" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">

<div dir=3D"ltr"><br><div class=3D"gmail_extra"><br><br><div class=3D"gmail=
_quote"><div>On Wed, Feb 12, 2014 at 9:14 PM, Takeshi Yoshino <span dir=3D"=
ltr">&lt;<a href=3D"mailto:tyoshino@google.com" target=3D"_blank">tyoshino@=
google.com</a>&gt;</span> wrote:<br>


<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div dir=3D"ltr"><div class=3D"gmail_extra">=
<div class=3D"gmail_quote"><div>On Thu, Feb 13, 2014 at 2:08 AM, Roberto Pe=
on <span dir=3D"ltr">&lt;<a href=3D"mailto:fenix@google.com" target=3D"_bla=
nk">fenix@google.com</a>&gt;</span> wrote:<br>




<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div dir=3D"ltr">On should steer away from keep-alive (whi=
ch is anathema on mobile links), and use liveness detection when one needs =
to solve these problems.<div>




(I believe that was what you essentially proposed, Martin)<br>
<div><br></div></div></div></blockquote><div><br></div></div><div>I want to=
 confirm my understanding on your terms. Do you mean that</div><div>- it&#3=
9;s kinda illusion that we are able to keep an end-to-end connectivity aliv=
e</div>




<div>- but should just have a way to detect the peer is dead when wanted</d=
iv></div></div></div></blockquote><div><br></div></div><div>yup.</div><div>=
Essentially, we don&#39;t want traffic black-holed (because the connection =
is dead but we haven&#39;t detected it), and we don&#39;t want to incent &q=
uot;the network&quot; to kill connections unnecessarily (this is particular=
ly an issue on shared bandwidth mediums like mobile networks).</div>

<div>
<div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8=
ex;border-left:1px #ccc solid;padding-left:1ex"><div dir=3D"ltr"><div class=
=3D"gmail_extra"><div class=3D"gmail_quote"><div><div>=C2=A0</div><blockquo=
te class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-widt=
h:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-le=
ft:1ex">




<div dir=3D"ltr"><div><div><div>A PING (application-layer equivalent if nec=
essary) should be automatically generated when the client has been idle for=
 some time and then has any message to send. If the PONG is not received wi=
thin some reasonable timeout (preferably ~2-3RT or so, or within some other=
 human-scale constant), the connection should be assumed dead, and another =
should be created.</div>






<div><br></div><div>imho, this should be done for you by the browser (not t=
hat it necessarily *is*, but it *should*)</div></div></div></div></blockquo=
te><div><br></div></div><div>Thinking of what browser vendors can do at thi=
s point:</div>




<div>a) have built-in default-on ping-generate/timeout-close (try to do som=
ething intelligent like TCP&#39;s RTO calculation?) feature in browser. one=
 concern is that people may have skipped implementation of ping/pong</div>




<div>b) same as (a) but default-off and can be turned on by calling some me=
thod on WebSocket object</div><div>c) have one-shot ping sending method on =
WebSocket object with timeout arg</div><div>d) have one-shot ping sending m=
ethod and pong received event. Sounds overkilling to me</div>




<div>e) periodical version of (c) or (d)</div></div></div></div></blockquot=
e></div><div><br>We have implemenation experience for (a) for HTTP/2/SPDY, =
and it seems to work pretty well.</div><div>HTTP2/SPDY has the same problem=
: if the connection is dead, it is a poor user experience to attempt to kee=
p sending all traffic down it!</div>


<div><br></div><div>The design rationale behind the choice of (a) was:</div=
><div>it transparently fixed the biggest problem case without requiring any=
 user involvement or knowledge.</div><div>It was important to use ping, bec=
ause ping/pong requires no interaction with the application-layer logic, an=
d thus is most likely to be consistent and have timing similar to TCP&#39;s=
 PING.</div>

<span><font color=3D"#888888">
<div><br></div><div>-=3DR</div></font></span><div><div><div>=C2=A0</div><bl=
ockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #=
ccc solid;padding-left:1ex"><div dir=3D"ltr"><div class=3D"gmail_extra"><di=
v class=3D"gmail_quote">

<div><div>
<div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px =
0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-l=
eft-style:solid;padding-left:1ex">

<div dir=3D"ltr"><div><div><div><br></div><div>Keep-alive is fairly evil at=
 the infrastructure layer, and incents browsers, servers and carriers to do=
 things to the tear down connection quickly/prematurely.</div>
<div><br></div><div>Liveness-detection doesn&#39;t solve the server notific=
ation problem (which requires at least some level of high-level polling, th=
ough hopefully on a longer timescale), but keep in mind that in seeking per=
fection, one is likely to cause a significant less than perfect response fr=
om the system and the players within it.=C2=A0</div>





<div><br></div><div>(It is pretty annoying that we&#39;ve not developed goo=
d APIs or behaviors for doing decent things on mobile devices...)</div><spa=
n><font color=3D"#888888"><div><br></div><div>-=3DR</div></font></span></di=
v>




</div></div><div class=3D"gmail_extra"><br>
<br><div class=3D"gmail_quote"><div><div>On Tue, Feb 11, 2014 at 7:50 PM, T=
akeshi Yoshino <span dir=3D"ltr">&lt;<a href=3D"mailto:tyoshino@google.com"=
 target=3D"_blank">tyoshino@google.com</a>&gt;</span> wrote:<br></div>

</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;b=
order-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:s=
olid;padding-left:1ex"><div><div>
<div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote"><div=
>On Thu, Jan 9, 2014 at 9:13 PM, Tobias Oberstein <span dir=3D"ltr">&lt;<a =
href=3D"mailto:tobias.oberstein@tavendo.de" target=3D"_blank">tobias.oberst=
ein@tavendo.de</a>&gt;</span> wrote:<br>







<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div>&gt; The WS API is explicit about the fact that there=
 is no way from JavaScript to<br>





&gt; generate PINGs or check for PONGs. So do I just create my own, applica=
tion<br>
&gt; level equivalent, and send a ping-like message every so often and assu=
me<br>
<br>
</div>You can have a WS server that does automatic ping-pong for connection=
 keep-alive/loss-detection.<br>
<br></blockquote><div><br></div></div><div>This works for server. For clien=
t, app level work is needed.</div><div><div>=C2=A0</div><blockquote class=
=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;bo=
rder-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">






Or you can do it on app level from JS.<br>
<div><br>
&gt; the connection is dead when I don&#39;t get a reply &quot;soonish&quot=
;? Are there libraries<br>
&gt; that support such a mechanism? Or is every browser supported to regula=
rly<br>
&gt; PINGPONG with the server, and call the onclose function when there&#39=
;s no<br>
<br>
</div>As far as I know, there is no JS API where you could request the brow=
ser to do automatic ping-pong upon opening a WS connection.<br>
<div><div><br>
&gt; answer?<br>
&gt;<br></div></div></blockquote><div><br></div></div><div>Right. We&#39;ve=
 been saying that we should start discussing such a change on the API, but =
haven&#39;t.</div><div><br></div><div>// Send a ping (w/o body) every 5 sec=
</div>







<div>ws.pingInterval =3D 5;</div><div>// _Close the WebSocket connection_ w=
hen it passed 10 sec without receiving any byte (not limited to PONG)</div>=
<div>ws.timeout =3D 10;</div><div>=C2=A0</div></div></div></div>
<br></div></div><div>_______________________________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
<br></div></blockquote></div><br></div>
</blockquote></div></div></div><br></div></div>
</blockquote></div></div></div><br></div></div>
<br>_______________________________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
<br></blockquote></div><br></div>
</div></div></blockquote></div><br></div>

--001a11c3e45441340704f262b5ce--


From nobody Sun Feb 16 18:56:53 2014
Return-Path: <ricea@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id CCCD01A040E for <hybi@ietfa.amsl.com>; Sun, 16 Feb 2014 18:56:51 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.326
X-Spam-Level: 
X-Spam-Status: No, score=-1.326 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, J_CHICKENPOX_27=0.6, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id GuK_OezrcdAt for <hybi@ietfa.amsl.com>; Sun, 16 Feb 2014 18:56:49 -0800 (PST)
Received: from mail-ob0-x22e.google.com (mail-ob0-x22e.google.com [IPv6:2607:f8b0:4003:c01::22e]) by ietfa.amsl.com (Postfix) with ESMTP id 12AAB1A02FD for <hybi@ietf.org>; Sun, 16 Feb 2014 18:56:48 -0800 (PST)
Received: by mail-ob0-f174.google.com with SMTP id uy5so16345846obc.19 for <hybi@ietf.org>; Sun, 16 Feb 2014 18:56:46 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=JHQPFQ9v7ExcyrU2kOdUyJ3X4qlfh/GZ/MjaqHe1OKk=; b=LbApu1lqrg6HOEEyNjcdfuMqt7eqW5MfYqkDc90if8u2lvZkYGuzeI/1o25+h3NPzQ iG5UFsOBZtUTxj1IamwuxNX+hK0jUDV4IBrWk9Rt2HLMBISuiFOsW9LxZbRDwvfs8RMC LUVErpaahfo1eyIE/pHtCWv1EIHC46RmJmH6UoUjFdqH1m/NBmjibL9ErsJkq1454auY qbApnN+X8pF5eblMVc0W3qX/bECTc3giqNobz4sLWr2fWxUYR1HPubrzGWZJdGSXIxrR ZXHZ/2s8nmYCY93/2RqqT6OAi6dlGquhgJ/Din4GAhbZyn7wA8eAS5Hj2ZEMfGk8nYfo CoLQ==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=JHQPFQ9v7ExcyrU2kOdUyJ3X4qlfh/GZ/MjaqHe1OKk=; b=Dw8AIpvNtrtbatH+0I3JPl99b1Fc8kgCuWwRomepNE5cnfqcjKx83xAGTO4LeI/xJz sQ1tz3sA+9pS9QMT3j0LBo4zHLnUNruQvZ3HfNmmTabm2LmTXff/wPdLqxMjaAcinkhL 8nT3E3g9fplctpVOTnRmSN0xPXEVsG93Ws/hR9MljfBbxmBR8aA3aDyn86HHSi+wL9wp hx6nGF4U9zEOgbSY//vxoSSky/Dai0kTz5z+JhXAAGQ7pvYSy4o+GAW9xS/C6Y1DSgsq Dx61lgY1Mvg5BW4zBGsa2rk2AQJUJkQayDH0PZAo54fv96Egry4hnziy5oeCdNtTEjZu UBLg==
X-Gm-Message-State: ALoCoQmwyZgCYu3zr/NfcKLMQGWyvthsyuDYn9RUhYkUZrxelSOB2cdXGCckowFmk4QVw/ReQ1x4qD2DuEMZpoMuDWvnfTr6rUQ/D6iPgFii6YkDTab1p4TSvdLeOQSAWVw5RlZL0g0Ntpjzt+gSsAY/SQDIxszB61pKuz6ouaL6sQwMDOXBq+09aUEkMimda4EU269G60GK
MIME-Version: 1.0
X-Received: by 10.182.19.132 with SMTP id f4mr18489661obe.14.1392605806416; Sun, 16 Feb 2014 18:56:46 -0800 (PST)
Received: by 10.182.118.41 with HTTP; Sun, 16 Feb 2014 18:56:46 -0800 (PST)
In-Reply-To: <CAGzyod5OgpYR=pVpa5HVjhj=VJYb0uPUJEkONUrLGB7x2fe8Sw@mail.gmail.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E0@EXVMBX020-12.exch020.serverdata.net> <52CE8D59.3070802@it.aoyama.ac.jp> <634914A010D0B943A035D226786325D4446BE403EB@EXVMBX020-12.exch020.serverdata.net> <CAH9hSJZfEoYv9uxH31R-s1zbXy7PqQW7MsA475VJJdFrjDhj_w@mail.gmail.com> <CAGzyod6i6HnOETSm1j_nhbQDp-129g06ZhR1Ey=KdGzzCQ+hJw@mail.gmail.com> <CAH9hSJYaRLbXsXXx=dcTyDqC+GJ8vj=E=oDg6vSQ+549DniV3Q@mail.gmail.com> <CAGzyod4Ga0Gy9f9f+chufqhWa4XswmOdAtNU7rkXY6ZzayTa-g@mail.gmail.com> <CAHixhFpzoToMF1L+bchnLfpWd2u2mB9-1Z41vPMs=aO_13W2Hg@mail.gmail.com> <CAGzyod5OgpYR=pVpa5HVjhj=VJYb0uPUJEkONUrLGB7x2fe8Sw@mail.gmail.com>
Date: Mon, 17 Feb 2014 11:56:46 +0900
Message-ID: <CAHixhFqhsO3R+c7aM3F1YNeDb1hWpKuN4Ojop4TcnPQfKKxyLw@mail.gmail.com>
From: Adam Rice <ricea@google.com>
To: Roberto Peon <fenix@google.com>
Content-Type: multipart/alternative; boundary=001a11c2aa6a6772a504f2914eb9
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/CggZ1TWBJhkfKx24NraO_YJeVcU
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Detection of broken WS connections
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 17 Feb 2014 02:56:52 -0000

--001a11c2aa6a6772a504f2914eb9
Content-Type: text/plain; charset=UTF-8

Consider an earthquake early-warning web app. It can provide a 15-second
warning of an event which only happens a couple of times a year. If the
warning is more than 5 seconds late, it is effectively useless.

Polling would be hideously inefficient for this use case. Even if the
OS/network forces you to ping every 5 minutes to keep the connection alive,
it is still going to be several orders of magnitude more efficient than
polling.

There are legitimate uses for long-lived, idle connections. WebSocket is
currently the least-worst browser technology available to satisfy them. If
we can make it less bad, I think that would be worthwhile.

With respect for Salvatore's point, if the suspend/resume facility existed
you could in principle tunnel SSL over it (ie. within the WebSocket
payload) and still get the benefit without trusting the proxy more than you
already have to trust your network provider anyway.

Adam


On 15 February 2014 04:21, Roberto Peon <fenix@google.com> wrote:

> Basically, keep-alives ties up enough network resources and/or device
> battery life that OS providers and network providers have strong incentive
> to consider a connection idle (and close it/blackhole it) for smaller and
> smaller time periods, eventually resulting in an arms race which increases
> inefficiencies everywhere.
>
> -=R
>
>
> On Fri, Feb 14, 2014 at 2:07 AM, Adam Rice <ricea@google.com> wrote:
>
>> I don't think anyone has yet made the point that WebSocket connections
>> cannot be assumed to be stateless, so browsers cannot implement transparent
>> reconnection without an extension with some server-side state.
>>
>> I have not so far seen a compelling argument that reconnection logic is
>> better implemented as a protocol extension rather than at the application
>> level. Hypothetically mobile networks could provide some kind of WebSocket
>> proxy with suspend/resume capabilities, but so far no such thing exists.
>>
>> Currently WebSocket applications can control keepalive behaviour by
>> sending (or not) pings from the server.
>>
>> I see two potential areas for improvement:
>> 1) Client side detection of loss of connection
>> 2) More efficient liveness detection
>>
>> 1) could be resolved with something as simple as an
>> "expect_ping_when_idle" extension which told the client about whatever
>> keepalive policy the server was attempting to enforce.
>>
>> Roberto seems to be implying that 2) may in fact be counter-productive,
>> although I'm not sure I fully understand the argument.
>>
>> Adam Rice
>>
>>
>> On 14 February 2014 04:36, Roberto Peon <fenix@google.com> wrote:
>>
>>>
>>>
>>>
>>> On Wed, Feb 12, 2014 at 9:14 PM, Takeshi Yoshino <tyoshino@google.com>wrote:
>>>
>>>> On Thu, Feb 13, 2014 at 2:08 AM, Roberto Peon <fenix@google.com> wrote:
>>>>
>>>>> On should steer away from keep-alive (which is anathema on mobile
>>>>> links), and use liveness detection when one needs to solve these problems.
>>>>> (I believe that was what you essentially proposed, Martin)
>>>>>
>>>>>
>>>> I want to confirm my understanding on your terms. Do you mean that
>>>> - it's kinda illusion that we are able to keep an end-to-end
>>>> connectivity alive
>>>> - but should just have a way to detect the peer is dead when wanted
>>>>
>>>
>>> yup.
>>> Essentially, we don't want traffic black-holed (because the connection
>>> is dead but we haven't detected it), and we don't want to incent "the
>>> network" to kill connections unnecessarily (this is particularly an issue
>>> on shared bandwidth mediums like mobile networks).
>>>
>>>
>>>>
>>>>
>>>>> A PING (application-layer equivalent if necessary) should be
>>>>> automatically generated when the client has been idle for some time and
>>>>> then has any message to send. If the PONG is not received within some
>>>>> reasonable timeout (preferably ~2-3RT or so, or within some other
>>>>> human-scale constant), the connection should be assumed dead, and another
>>>>> should be created.
>>>>>
>>>>> imho, this should be done for you by the browser (not that it
>>>>> necessarily *is*, but it *should*)
>>>>>
>>>>
>>>> Thinking of what browser vendors can do at this point:
>>>> a) have built-in default-on ping-generate/timeout-close (try to do
>>>> something intelligent like TCP's RTO calculation?) feature in browser. one
>>>> concern is that people may have skipped implementation of ping/pong
>>>> b) same as (a) but default-off and can be turned on by calling some
>>>> method on WebSocket object
>>>> c) have one-shot ping sending method on WebSocket object with timeout
>>>> arg
>>>> d) have one-shot ping sending method and pong received event. Sounds
>>>> overkilling to me
>>>> e) periodical version of (c) or (d)
>>>>
>>>
>>> We have implemenation experience for (a) for HTTP/2/SPDY, and it seems
>>> to work pretty well.
>>> HTTP2/SPDY has the same problem: if the connection is dead, it is a poor
>>> user experience to attempt to keep sending all traffic down it!
>>>
>>> The design rationale behind the choice of (a) was:
>>> it transparently fixed the biggest problem case without requiring any
>>> user involvement or knowledge.
>>> It was important to use ping, because ping/pong requires no interaction
>>> with the application-layer logic, and thus is most likely to be consistent
>>> and have timing similar to TCP's PING.
>>>
>>> -=R
>>>
>>>
>>>>
>>>>
>>>>>
>>>>> Keep-alive is fairly evil at the infrastructure layer, and incents
>>>>> browsers, servers and carriers to do things to the tear down connection
>>>>> quickly/prematurely.
>>>>>
>>>>> Liveness-detection doesn't solve the server notification problem
>>>>> (which requires at least some level of high-level polling, though hopefully
>>>>> on a longer timescale), but keep in mind that in seeking perfection, one is
>>>>> likely to cause a significant less than perfect response from the system
>>>>> and the players within it.
>>>>>
>>>>> (It is pretty annoying that we've not developed good APIs or behaviors
>>>>> for doing decent things on mobile devices...)
>>>>>
>>>>> -=R
>>>>>
>>>>>
>>>>> On Tue, Feb 11, 2014 at 7:50 PM, Takeshi Yoshino <tyoshino@google.com>wrote:
>>>>>
>>>>>> On Thu, Jan 9, 2014 at 9:13 PM, Tobias Oberstein <
>>>>>> tobias.oberstein@tavendo.de> wrote:
>>>>>>
>>>>>>> > The WS API is explicit about the fact that there is no way from
>>>>>>> JavaScript to
>>>>>>> > generate PINGs or check for PONGs. So do I just create my own,
>>>>>>> application
>>>>>>> > level equivalent, and send a ping-like message every so often and
>>>>>>> assume
>>>>>>>
>>>>>>> You can have a WS server that does automatic ping-pong for
>>>>>>> connection keep-alive/loss-detection.
>>>>>>>
>>>>>>>
>>>>>> This works for server. For client, app level work is needed.
>>>>>>
>>>>>>
>>>>>>> Or you can do it on app level from JS.
>>>>>>>
>>>>>>> > the connection is dead when I don't get a reply "soonish"? Are
>>>>>>> there libraries
>>>>>>> > that support such a mechanism? Or is every browser supported to
>>>>>>> regularly
>>>>>>> > PINGPONG with the server, and call the onclose function when
>>>>>>> there's no
>>>>>>>
>>>>>>> As far as I know, there is no JS API where you could request the
>>>>>>> browser to do automatic ping-pong upon opening a WS connection.
>>>>>>>
>>>>>>> > answer?
>>>>>>> >
>>>>>>>
>>>>>>
>>>>>> Right. We've been saying that we should start discussing such a
>>>>>> change on the API, but haven't.
>>>>>>
>>>>>> // Send a ping (w/o body) every 5 sec
>>>>>> ws.pingInterval = 5;
>>>>>> // _Close the WebSocket connection_ when it passed 10 sec without
>>>>>> receiving any byte (not limited to PONG)
>>>>>> ws.timeout = 10;
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> hybi mailing list
>>>>>> hybi@ietf.org
>>>>>> https://www.ietf.org/mailman/listinfo/hybi
>>>>>>
>>>>>>
>>>>>
>>>>
>>>
>>> _______________________________________________
>>> hybi mailing list
>>> hybi@ietf.org
>>> https://www.ietf.org/mailman/listinfo/hybi
>>>
>>>
>>
>

--001a11c2aa6a6772a504f2914eb9
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Consider an earthquake early-warning web app. It can provi=
de a 15-second warning of an event which only happens a couple of times a y=
ear. If the warning is more than 5 seconds late, it is effectively useless.=
<div>
<br></div><div>Polling would be hideously inefficient for this use case. Ev=
en if the OS/network forces you to ping every 5 minutes to keep the connect=
ion alive, it is still going to be several orders of magnitude more efficie=
nt than polling.</div>
<div><br></div><div>There are legitimate uses for long-lived, idle connecti=
ons. WebSocket is currently the least-worst browser technology available to=
 satisfy them. If we can make it less bad, I think that would be worthwhile=
.</div>
<div><br></div><div>With respect for Salvatore&#39;s point, if the suspend/=
resume facility existed you could in principle tunnel SSL over it (ie. with=
in the WebSocket payload) and still get the benefit without trusting the pr=
oxy more than you already have to trust your network provider anyway.</div>
<div><br></div><div>Adam</div></div><div class=3D"gmail_extra"><br><br><div=
 class=3D"gmail_quote">On 15 February 2014 04:21, Roberto Peon <span dir=3D=
"ltr">&lt;<a href=3D"mailto:fenix@google.com" target=3D"_blank">fenix@googl=
e.com</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div dir=3D"ltr">Basically, keep-alives ties=
 up enough network resources and/or device battery life that OS providers a=
nd network providers have strong incentive to consider a connection idle (a=
nd close it/blackhole it) for smaller and smaller time periods, eventually =
resulting in an arms race which increases inefficiencies everywhere.<span c=
lass=3D"HOEnZb"><font color=3D"#888888"><div>

<br></div><div>-=3DR</div></font></span></div><div class=3D"HOEnZb"><div cl=
ass=3D"h5"><div class=3D"gmail_extra"><br><br><div class=3D"gmail_quote">On=
 Fri, Feb 14, 2014 at 2:07 AM, Adam Rice <span dir=3D"ltr">&lt;<a href=3D"m=
ailto:ricea@google.com" target=3D"_blank">ricea@google.com</a>&gt;</span> w=
rote:<br>

<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div dir=3D"ltr">I don&#39;t think anyone ha=
s yet made the point that WebSocket connections cannot be assumed to be sta=
teless, so browsers cannot implement transparent reconnection without an ex=
tension with some server-side state.<div>


<br></div><div>I have not so far seen a compelling argument that reconnecti=
on logic is better implemented as a protocol extension rather than at the a=
pplication level. Hypothetically mobile networks could provide some kind of=
 WebSocket proxy with suspend/resume capabilities, but so far no such thing=
 exists.</div>


<div><br></div><div>Currently WebSocket applications can control keepalive =
behaviour by sending (or not) pings from the server.</div><div><br></div><d=
iv>I see two potential areas for improvement:</div><div>1) Client side dete=
ction of loss of connection</div>


<div>2) More efficient liveness detection</div><div><br></div><div>1) could=
 be resolved with something as simple as an &quot;expect_ping_when_idle&quo=
t; extension which told the client about whatever keepalive policy the serv=
er was attempting to enforce.</div>


<div><br></div><div>Roberto seems to be implying that 2) may in fact be cou=
nter-productive, although I&#39;m not sure I fully understand the argument.=
</div><span><font color=3D"#888888"><div><br></div><div>Adam Rice</div>
</font></span></div><div><div><div class=3D"gmail_extra">
<br><br><div class=3D"gmail_quote">On 14 February 2014 04:36, Roberto Peon =
<span dir=3D"ltr">&lt;<a href=3D"mailto:fenix@google.com" target=3D"_blank"=
>fenix@google.com</a>&gt;</span> wrote:<br><blockquote class=3D"gmail_quote=
" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">


<div dir=3D"ltr"><br><div class=3D"gmail_extra"><br><br><div class=3D"gmail=
_quote"><div>On Wed, Feb 12, 2014 at 9:14 PM, Takeshi Yoshino <span dir=3D"=
ltr">&lt;<a href=3D"mailto:tyoshino@google.com" target=3D"_blank">tyoshino@=
google.com</a>&gt;</span> wrote:<br>



<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div dir=3D"ltr"><div class=3D"gmail_extra">=
<div class=3D"gmail_quote"><div>On Thu, Feb 13, 2014 at 2:08 AM, Roberto Pe=
on <span dir=3D"ltr">&lt;<a href=3D"mailto:fenix@google.com" target=3D"_bla=
nk">fenix@google.com</a>&gt;</span> wrote:<br>





<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div dir=3D"ltr">On should steer away from keep-alive (whi=
ch is anathema on mobile links), and use liveness detection when one needs =
to solve these problems.<div>





(I believe that was what you essentially proposed, Martin)<br>
<div><br></div></div></div></blockquote><div><br></div></div><div>I want to=
 confirm my understanding on your terms. Do you mean that</div><div>- it&#3=
9;s kinda illusion that we are able to keep an end-to-end connectivity aliv=
e</div>





<div>- but should just have a way to detect the peer is dead when wanted</d=
iv></div></div></div></blockquote><div><br></div></div><div>yup.</div><div>=
Essentially, we don&#39;t want traffic black-holed (because the connection =
is dead but we haven&#39;t detected it), and we don&#39;t want to incent &q=
uot;the network&quot; to kill connections unnecessarily (this is particular=
ly an issue on shared bandwidth mediums like mobile networks).</div>


<div>
<div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8=
ex;border-left:1px #ccc solid;padding-left:1ex"><div dir=3D"ltr"><div class=
=3D"gmail_extra"><div class=3D"gmail_quote"><div><div>=C2=A0</div><blockquo=
te class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-widt=
h:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-le=
ft:1ex">





<div dir=3D"ltr"><div><div><div>A PING (application-layer equivalent if nec=
essary) should be automatically generated when the client has been idle for=
 some time and then has any message to send. If the PONG is not received wi=
thin some reasonable timeout (preferably ~2-3RT or so, or within some other=
 human-scale constant), the connection should be assumed dead, and another =
should be created.</div>







<div><br></div><div>imho, this should be done for you by the browser (not t=
hat it necessarily *is*, but it *should*)</div></div></div></div></blockquo=
te><div><br></div></div><div>Thinking of what browser vendors can do at thi=
s point:</div>





<div>a) have built-in default-on ping-generate/timeout-close (try to do som=
ething intelligent like TCP&#39;s RTO calculation?) feature in browser. one=
 concern is that people may have skipped implementation of ping/pong</div>





<div>b) same as (a) but default-off and can be turned on by calling some me=
thod on WebSocket object</div><div>c) have one-shot ping sending method on =
WebSocket object with timeout arg</div><div>d) have one-shot ping sending m=
ethod and pong received event. Sounds overkilling to me</div>





<div>e) periodical version of (c) or (d)</div></div></div></div></blockquot=
e></div><div><br>We have implemenation experience for (a) for HTTP/2/SPDY, =
and it seems to work pretty well.</div><div>HTTP2/SPDY has the same problem=
: if the connection is dead, it is a poor user experience to attempt to kee=
p sending all traffic down it!</div>



<div><br></div><div>The design rationale behind the choice of (a) was:</div=
><div>it transparently fixed the biggest problem case without requiring any=
 user involvement or knowledge.</div><div>It was important to use ping, bec=
ause ping/pong requires no interaction with the application-layer logic, an=
d thus is most likely to be consistent and have timing similar to TCP&#39;s=
 PING.</div>


<span><font color=3D"#888888">
<div><br></div><div>-=3DR</div></font></span><div><div><div>=C2=A0</div><bl=
ockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #=
ccc solid;padding-left:1ex"><div dir=3D"ltr"><div class=3D"gmail_extra"><di=
v class=3D"gmail_quote">


<div><div>
<div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px =
0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-l=
eft-style:solid;padding-left:1ex">

<div dir=3D"ltr"><div><div><div><br></div><div>Keep-alive is fairly evil at=
 the infrastructure layer, and incents browsers, servers and carriers to do=
 things to the tear down connection quickly/prematurely.</div>
<div><br></div><div>Liveness-detection doesn&#39;t solve the server notific=
ation problem (which requires at least some level of high-level polling, th=
ough hopefully on a longer timescale), but keep in mind that in seeking per=
fection, one is likely to cause a significant less than perfect response fr=
om the system and the players within it.=C2=A0</div>






<div><br></div><div>(It is pretty annoying that we&#39;ve not developed goo=
d APIs or behaviors for doing decent things on mobile devices...)</div><spa=
n><font color=3D"#888888"><div><br></div><div>-=3DR</div></font></span></di=
v>





</div></div><div class=3D"gmail_extra"><br>
<br><div class=3D"gmail_quote"><div><div>On Tue, Feb 11, 2014 at 7:50 PM, T=
akeshi Yoshino <span dir=3D"ltr">&lt;<a href=3D"mailto:tyoshino@google.com"=
 target=3D"_blank">tyoshino@google.com</a>&gt;</span> wrote:<br></div>

</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;b=
order-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:s=
olid;padding-left:1ex"><div><div>
<div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote"><div=
>On Thu, Jan 9, 2014 at 9:13 PM, Tobias Oberstein <span dir=3D"ltr">&lt;<a =
href=3D"mailto:tobias.oberstein@tavendo.de" target=3D"_blank">tobias.oberst=
ein@tavendo.de</a>&gt;</span> wrote:<br>








<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div>&gt; The WS API is explicit about the fact that there=
 is no way from JavaScript to<br>






&gt; generate PINGs or check for PONGs. So do I just create my own, applica=
tion<br>
&gt; level equivalent, and send a ping-like message every so often and assu=
me<br>
<br>
</div>You can have a WS server that does automatic ping-pong for connection=
 keep-alive/loss-detection.<br>
<br></blockquote><div><br></div></div><div>This works for server. For clien=
t, app level work is needed.</div><div><div>=C2=A0</div><blockquote class=
=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;bo=
rder-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex">







Or you can do it on app level from JS.<br>
<div><br>
&gt; the connection is dead when I don&#39;t get a reply &quot;soonish&quot=
;? Are there libraries<br>
&gt; that support such a mechanism? Or is every browser supported to regula=
rly<br>
&gt; PINGPONG with the server, and call the onclose function when there&#39=
;s no<br>
<br>
</div>As far as I know, there is no JS API where you could request the brow=
ser to do automatic ping-pong upon opening a WS connection.<br>
<div><div><br>
&gt; answer?<br>
&gt;<br></div></div></blockquote><div><br></div></div><div>Right. We&#39;ve=
 been saying that we should start discussing such a change on the API, but =
haven&#39;t.</div><div><br></div><div>// Send a ping (w/o body) every 5 sec=
</div>








<div>ws.pingInterval =3D 5;</div><div>// _Close the WebSocket connection_ w=
hen it passed 10 sec without receiving any byte (not limited to PONG)</div>=
<div>ws.timeout =3D 10;</div><div>=C2=A0</div></div></div></div>
<br></div></div><div>_______________________________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
<br></div></blockquote></div><br></div>
</blockquote></div></div></div><br></div></div>
</blockquote></div></div></div><br></div></div>
<br>_______________________________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
<br></blockquote></div><br></div>
</div></div></blockquote></div><br></div>
</div></div></blockquote></div><br></div>

--001a11c2aa6a6772a504f2914eb9--


From nobody Mon Feb 17 02:19:51 2014
Return-Path: <tobias.oberstein@tavendo.de>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 2FE7C1A047B for <hybi@ietfa.amsl.com>; Mon, 17 Feb 2014 02:19:48 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 0.6
X-Spam-Level: 
X-Spam-Status: No, score=0.6 tagged_above=-999 required=5 tests=[BAYES_40=-0.001, HTML_MESSAGE=0.001, J_CHICKENPOX_27=0.6, RCVD_IN_DNSWL_NONE=-0.0001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id SjGGRgXFy2Ta for <hybi@ietfa.amsl.com>; Mon, 17 Feb 2014 02:19:39 -0800 (PST)
Received: from EXHUB020-3.exch020.serverdata.net (exhub020-3.exch020.serverdata.net [206.225.164.30]) by ietfa.amsl.com (Postfix) with ESMTP id E658F1A047E for <hybi@ietf.org>; Mon, 17 Feb 2014 02:19:38 -0800 (PST)
Received: from EXVMBX020-12.exch020.serverdata.net ([169.254.3.11]) by EXHUB020-3.exch020.serverdata.net ([206.225.164.30]) with mapi; Mon, 17 Feb 2014 02:19:36 -0800
From: Tobias Oberstein <tobias.oberstein@tavendo.de>
To: Adam Rice <ricea@google.com>, Roberto Peon <fenix@google.com>
Date: Mon, 17 Feb 2014 02:19:32 -0800
Thread-Topic: [hybi] Detection of broken WS connections
Thread-Index: Ac8ri+rsSCZJAXpfT1aTUyKQV3INRgAPEBGQ
Message-ID: <634914A010D0B943A035D226786325D4446C4509F2@EXVMBX020-12.exch020.serverdata.net>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E0@EXVMBX020-12.exch020.serverdata.net> <52CE8D59.3070802@it.aoyama.ac.jp> <634914A010D0B943A035D226786325D4446BE403EB@EXVMBX020-12.exch020.serverdata.net> <CAH9hSJZfEoYv9uxH31R-s1zbXy7PqQW7MsA475VJJdFrjDhj_w@mail.gmail.com> <CAGzyod6i6HnOETSm1j_nhbQDp-129g06ZhR1Ey=KdGzzCQ+hJw@mail.gmail.com> <CAH9hSJYaRLbXsXXx=dcTyDqC+GJ8vj=E=oDg6vSQ+549DniV3Q@mail.gmail.com> <CAGzyod4Ga0Gy9f9f+chufqhWa4XswmOdAtNU7rkXY6ZzayTa-g@mail.gmail.com> <CAHixhFpzoToMF1L+bchnLfpWd2u2mB9-1Z41vPMs=aO_13W2Hg@mail.gmail.com> <CAGzyod5OgpYR=pVpa5HVjhj=VJYb0uPUJEkONUrLGB7x2fe8Sw@mail.gmail.com> <CAHixhFqhsO3R+c7aM3F1YNeDb1hWpKuN4Ojop4TcnPQfKKxyLw@mail.gmail.com>
In-Reply-To: <CAHixhFqhsO3R+c7aM3F1YNeDb1hWpKuN4Ojop4TcnPQfKKxyLw@mail.gmail.com>
Accept-Language: de-DE, en-US
Content-Language: de-DE
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
acceptlanguage: de-DE, en-US
Content-Type: multipart/alternative; boundary="_000_634914A010D0B943A035D226786325D4446C4509F2EXVMBX02012ex_"
MIME-Version: 1.0
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/DCNqIF6HSMBryccBEgVCR2e2hNo
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Detection of broken WS connections
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 17 Feb 2014 10:19:48 -0000

--_000_634914A010D0B943A035D226786325D4446C4509F2EXVMBX02012ex_
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: base64
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--_000_634914A010D0B943A035D226786325D4446C4509F2EXVMBX02012ex_
Content-Type: text/html; charset="utf-8"
Content-Transfer-Encoding: base64
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--_000_634914A010D0B943A035D226786325D4446C4509F2EXVMBX02012ex_--


From nobody Mon Feb 17 12:09:46 2014
Return-Path: <salvatore.loreto@ericsson.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 7D8891A027A for <hybi@ietfa.amsl.com>; Mon, 17 Feb 2014 12:09:42 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.239
X-Spam-Level: 
X-Spam-Status: No, score=-1.239 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HELO_EQ_SE=0.35, HOST_MISMATCH_NET=0.311, HTML_MESSAGE=0.001, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 5L_XQ4vT33EW for <hybi@ietfa.amsl.com>; Mon, 17 Feb 2014 12:09:41 -0800 (PST)
Received: from sessmg20.mgmt.ericsson.se (sessmg20.ericsson.net [193.180.251.50]) by ietfa.amsl.com (Postfix) with ESMTP id C54711A0257 for <hybi@ietf.org>; Mon, 17 Feb 2014 12:09:40 -0800 (PST)
X-AuditID: c1b4fb32-b7f4c8e0000012f5-dc-53026c812b99
Received: from ESESSHC001.ericsson.se (Unknown_Domain [153.88.253.124]) by sessmg20.mgmt.ericsson.se (Symantec Mail Security) with SMTP id 20.16.04853.18C62035; Mon, 17 Feb 2014 21:09:37 +0100 (CET)
Received: from ESESSMB109.ericsson.se ([169.254.9.128]) by ESESSHC001.ericsson.se ([153.88.183.21]) with mapi id 14.02.0387.000; Mon, 17 Feb 2014 21:09:37 +0100
From: Salvatore Loreto <salvatore.loreto@ericsson.com>
To: Adam Rice <ricea@google.com>
Thread-Topic: [hybi] Detection of broken WS connections
Thread-Index: AQHPK4vlHEAGpMqP8EKcj7MJA6woXJq50IAA
Date: Mon, 17 Feb 2014 20:09:36 +0000
Message-ID: <3AA835A8-83D9-454A-8136-2A250430CEEE@ericsson.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E0@EXVMBX020-12.exch020.serverdata.net> <52CE8D59.3070802@it.aoyama.ac.jp> <634914A010D0B943A035D226786325D4446BE403EB@EXVMBX020-12.exch020.serverdata.net> <CAH9hSJZfEoYv9uxH31R-s1zbXy7PqQW7MsA475VJJdFrjDhj_w@mail.gmail.com> <CAGzyod6i6HnOETSm1j_nhbQDp-129g06ZhR1Ey=KdGzzCQ+hJw@mail.gmail.com> <CAH9hSJYaRLbXsXXx=dcTyDqC+GJ8vj=E=oDg6vSQ+549DniV3Q@mail.gmail.com> <CAGzyod4Ga0Gy9f9f+chufqhWa4XswmOdAtNU7rkXY6ZzayTa-g@mail.gmail.com> <CAHixhFpzoToMF1L+bchnLfpWd2u2mB9-1Z41vPMs=aO_13W2Hg@mail.gmail.com> <CAGzyod5OgpYR=pVpa5HVjhj=VJYb0uPUJEkONUrLGB7x2fe8Sw@mail.gmail.com> <CAHixhFqhsO3R+c7aM3F1YNeDb1hWpKuN4Ojop4TcnPQfKKxyLw@mail.gmail.com>
In-Reply-To: <CAHixhFqhsO3R+c7aM3F1YNeDb1hWpKuN4Ojop4TcnPQfKKxyLw@mail.gmail.com>
Accept-Language: en-US
Content-Language: en-US
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
x-originating-ip: [153.88.183.154]
Content-Type: multipart/alternative; boundary="_000_3AA835A883D9454A81362A250430CEEEericssoncom_"
MIME-Version: 1.0
X-Brightmail-Tracker: H4sIAAAAAAAAA+NgFmphkeLIzCtJLcpLzFFi42KZGfG3RrcxhynY4PYNJYv/PQYW719uY7I4 +93DgdljwaZSjyVLfjIFMEVx2aSk5mSWpRbp2yVwZWz7cJml4JVgxdlXV1gbGNfxdzFyckgI mEjc2rCNBcIWk7hwbz1bFyMXh5DACUaJy7M+MUM4Sxgltp+bwApSxSZgJvH84RZmEFtEQEHi 0OFTTCA2s4CLxIGpX8BsYaCaxpY+dogac4kZdw6zQthGEoualoLFWQRUJSZcWQYW5xWwl7h1 6Ac7xLIeNonD66+BDeIUCJTYuXMzWBEj0HnfT62BWiYucevJfCaIswUkluw5zwxhi0q8fPyP FcJWklh0+zNQDQdQfbLEq7uiELsEJU7OfMIygVF0FpJJsxCqZiGpgijRkViw+xMbhK0tsWzh a2YY+8yBx0wQtrXErE3XUdQsYORYxShZnFpcnJtuZKCXm55bopdalJlcXJyfp1ecuokRGJMH t/w22sF4co/9IUZpDhYlcd7rrDVBQgLpiSWp2ampBalF8UWlOanFhxiZODilGhjZlO+sPnmk L393vq9d2/mQ8IB9D9ys7q5b8jjd5KtZ/gauHLfTW1b+rJ7O3t9UfXD+op9CjrPmnfE7csPj 50T958Lq99pOPIz+FzTtdRXvVJcTq0xZ1HvYVabxH4kSS7x8ZiE/D/cGz9WfpULiz//Vvvia a8bOQjWx/QHbppUF6QX98J9adjVJiaU4I9FQi7moOBEAS8TipJcCAAA=
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/hdXinUL9uPOCJnki8MoWKsaJC60
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Detection of broken WS connections
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 17 Feb 2014 20:09:42 -0000

--_000_3AA835A883D9454A81362A250430CEEEericssoncom_
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: quoted-printable


On Feb 17, 2014, at 4:56 AM, Adam Rice <ricea@google.com<mailto:ricea@googl=
e.com>> wrote:

With respect for Salvatore's point, if the suspend/resume facility existed =
you could in principle tunnel SSL over it (ie. within the WebSocket payload=
) and still get the benefit without trusting the proxy more than you alread=
y have to trust your network provider anyway.

sorry then I don't understand how the proxy can provide support if the susp=
end/resume facility are tunnelled over SSL

/Sal


--_000_3AA835A883D9454A81362A250430CEEEericssoncom_
Content-Type: text/html; charset="us-ascii"
Content-ID: <C26086E283D4864E840E33B09BA0C2D0@ericsson.com>
Content-Transfer-Encoding: quoted-printable

<html>
<head>
<meta http-equiv=3D"Content-Type" content=3D"text/html; charset=3Dus-ascii"=
>
</head>
<body style=3D"word-wrap: break-word; -webkit-nbsp-mode: space; -webkit-lin=
e-break: after-white-space; ">
<br>
<div>
<div>On Feb 17, 2014, at 4:56 AM, Adam Rice &lt;<a href=3D"mailto:ricea@goo=
gle.com">ricea@google.com</a>&gt; wrote:</div>
<br class=3D"Apple-interchange-newline">
<blockquote type=3D"cite"><span style=3D"font-family: Helvetica; font-size:=
 medium; font-style: normal; font-variant: normal; font-weight: normal; let=
ter-spacing: normal; line-height: normal; orphans: 2; text-align: -webkit-a=
uto; text-indent: 0px; text-transform: none; white-space: normal; widows: 2=
; word-spacing: 0px; -webkit-text-size-adjust: auto; -webkit-text-stroke-wi=
dth: 0px; display: inline !important; float: none; ">With
 respect for Salvatore's point, if the suspend/resume facility existed you =
could in principle tunnel SSL over it (ie. within the WebSocket payload) an=
d still get the benefit without trusting the proxy more than you already ha=
ve to trust your network provider
 anyway.</span></blockquote>
<br>
</div>
<div>sorry then I don't understand how the proxy can provide support if the=
 suspend/resume facility are tunnelled over SSL&nbsp;</div>
<div><br>
</div>
<div>/Sal</div>
<br>
</body>
</html>

--_000_3AA835A883D9454A81362A250430CEEEericssoncom_--


From nobody Mon Feb 17 18:32:58 2014
Return-Path: <ricea@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id D5A511A02F3 for <hybi@ietfa.amsl.com>; Mon, 17 Feb 2014 18:32:56 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.925
X-Spam-Level: 
X-Spam-Status: No, score=-1.925 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001, WEIRD_PORT=0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id rrpIA0Z0nfVw for <hybi@ietfa.amsl.com>; Mon, 17 Feb 2014 18:32:55 -0800 (PST)
Received: from mail-ob0-x234.google.com (mail-ob0-x234.google.com [IPv6:2607:f8b0:4003:c01::234]) by ietfa.amsl.com (Postfix) with ESMTP id C7EF91A0319 for <hybi@ietf.org>; Mon, 17 Feb 2014 18:32:53 -0800 (PST)
Received: by mail-ob0-f180.google.com with SMTP id wp4so17788657obc.39 for <hybi@ietf.org>; Mon, 17 Feb 2014 18:32:51 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=gaiGNH60vuNkwLPK+6sqIFxqteGi1i69wPMExxgXXVM=; b=X49LOGDTk4L99lwH7chKkA++wLy/gue8/jfKNLsPz20JlzwIhGqtP2Jynk1kuamraK Qo77Ze4Rb4EGEqnbmDbp8lQyTT33UmjAyJZEawT/TnZ9Hr6sNEa94gyO5lGrcuFuv59W SaJZepODRTXJ0RppiH/J1oMF30r1OOGFqGDHkj+qhIyYXZnzaS7RKTAfbRBPqtI7ZZmN 3NUiZIn3lUNFVmL9RL8t6I+vN9mdqb4AL8RfKVtVtR4KJ/iNDD7E1+1TarXxXDebQvWV fZ4XFZQM142Iw3TUQ8ieKIBQ7dHt3nUsriDeG1sz/wMrM7xlvLW/gJMT5HJ/XYOQX4sJ JicQ==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=gaiGNH60vuNkwLPK+6sqIFxqteGi1i69wPMExxgXXVM=; b=eau24fBzCIN/4mNjzUFIUksD/QIHwwvk1NX/HWQaZ5EYuvkYTZY0YZpicStYqFDmbv 8sl+Fn5RGwVlAXvEi2RodlbpQ+EzlxITsEItACY1JTRfK887B1ASV9hd8e3T7Z645QPs /m/TVZGUWKClx+rcOOze+1BijZAKNbFUWWBgvTWfrmQ+ImPYNiFOPnFi+kRR4AVGb41A qBgftbnugjffYvU7A8qpxrsUgAhL12Yi+udUFbvlbPq4SCfXUhMXgyJpxXV2w92cAry+ qPKDLMuAm2OKER5jcs6tUjqLBHm4BRBoqwJ1z8xcQlBiyG4ygamZilhqKGQObMNmswuk wIAg==
X-Gm-Message-State: ALoCoQmtQ6AtHVW0RBIV7+tVqUel+eoADqQlxUWgZMb/iE25ARzhalExbCTr5YThE18Ts1HPvVYO4vKngWfqlY/cN7dmMtziHdpjAviBQwUGgmGLbjjfhf012yHU2gsdYF2XIY4lfj7rZjykYe8bhb0JN589GQ/IdEUf2gF5SnPAcqv+eN2m4dtKFiBIh1LKgV7qVxG/L59X
MIME-Version: 1.0
X-Received: by 10.60.54.138 with SMTP id j10mr4497538oep.51.1392690770895; Mon, 17 Feb 2014 18:32:50 -0800 (PST)
Received: by 10.182.118.41 with HTTP; Mon, 17 Feb 2014 18:32:50 -0800 (PST)
In-Reply-To: <3AA835A8-83D9-454A-8136-2A250430CEEE@ericsson.com>
References: <CAH9hSJah=4Z4-NufVrUrP965epc9eMdEbC+3LmucN4DhR=bWiQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BE403E0@EXVMBX020-12.exch020.serverdata.net> <52CE8D59.3070802@it.aoyama.ac.jp> <634914A010D0B943A035D226786325D4446BE403EB@EXVMBX020-12.exch020.serverdata.net> <CAH9hSJZfEoYv9uxH31R-s1zbXy7PqQW7MsA475VJJdFrjDhj_w@mail.gmail.com> <CAGzyod6i6HnOETSm1j_nhbQDp-129g06ZhR1Ey=KdGzzCQ+hJw@mail.gmail.com> <CAH9hSJYaRLbXsXXx=dcTyDqC+GJ8vj=E=oDg6vSQ+549DniV3Q@mail.gmail.com> <CAGzyod4Ga0Gy9f9f+chufqhWa4XswmOdAtNU7rkXY6ZzayTa-g@mail.gmail.com> <CAHixhFpzoToMF1L+bchnLfpWd2u2mB9-1Z41vPMs=aO_13W2Hg@mail.gmail.com> <CAGzyod5OgpYR=pVpa5HVjhj=VJYb0uPUJEkONUrLGB7x2fe8Sw@mail.gmail.com> <CAHixhFqhsO3R+c7aM3F1YNeDb1hWpKuN4Ojop4TcnPQfKKxyLw@mail.gmail.com> <3AA835A8-83D9-454A-8136-2A250430CEEE@ericsson.com>
Date: Tue, 18 Feb 2014 11:32:50 +0900
Message-ID: <CAHixhFrDRPmfM7FOC0utkeCQ6p5RAt8SaCBEdEZny2zNJtGjAQ@mail.gmail.com>
From: Adam Rice <ricea@google.com>
To: Salvatore Loreto <salvatore.loreto@ericsson.com>
Content-Type: multipart/alternative; boundary=089e0112cd32ae7b0904f2a516c2
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/JJ8QEwEbQsjug78ONyV7jtYCh8A
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Detection of broken WS connections
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 18 Feb 2014 02:32:57 -0000

--089e0112cd32ae7b0904f2a516c2
Content-Type: text/plain; charset=UTF-8

For suspend/resume, the existence of data is sufficient to make a decision
about whether a connection is needed or not. The contents of the data need
not be understandable.

Concretely, the client starts by sending a request to the proxy with
headers like

Sec-WebSocket-Extensions: suspend-resume
Sec-WebSocket-Protocol: mux-proxy

The proxy responds, accepting these extensions and protocol and
establishing the connection.

Then for the browser to connect to a wss site it sends a message like:

CONNECT wss.example.com:433

After which all messages on that multiplexed channel are encrypted with the
key of wss.example.com, and the proxy cannot see their contents.

When the channel is idle, the browser sends a frame with opcode Suspend,
the proxy confirms with opcode Suspend, and the connection between the
browser and the proxy is closed. The connection between the proxy and
wss.example.com stays open.

If the browser has a message to send, it reconnects to the proxy and sends
a Resume frame.

If the proxy receives data from wss.example.com while suspended, it does
whatever tranport-specific magic is needed to notify the browser, and then
the browser does the "Resume" handshake as above, and receives the pending
data.


On 18 February 2014 05:09, Salvatore Loreto
<salvatore.loreto@ericsson.com>wrote:

>
>  On Feb 17, 2014, at 4:56 AM, Adam Rice <ricea@google.com> wrote:
>
> With respect for Salvatore's point, if the suspend/resume facility existed
> you could in principle tunnel SSL over it (ie. within the WebSocket
> payload) and still get the benefit without trusting the proxy more than you
> already have to trust your network provider anyway.
>
>
>  sorry then I don't understand how the proxy can provide support if the
> suspend/resume facility are tunnelled over SSL
>
>  /Sal
>
>

--089e0112cd32ae7b0904f2a516c2
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div>For suspend/resume, the existence of data is sufficie=
nt to make a decision about whether a connection is needed or not. The cont=
ents of the data need not be understandable.</div><div><br></div>Concretely=
, the client starts by sending a request to the proxy with headers like<div=
>
<br></div><div>Sec-WebSocket-Extensions: suspend-resume</div><div>Sec-WebSo=
cket-Protocol: mux-proxy</div><div><br></div><div>The proxy responds, accep=
ting these extensions and protocol and establishing the connection.</div>
<div><br></div><div>Then for the browser to connect to a wss site it sends =
a message like:</div><div><br></div><div>CONNECT <a href=3D"http://wss.exam=
ple.com:433">wss.example.com:433</a></div><div><br></div><div>After which a=
ll messages on that multiplexed channel are encrypted with the key of <a hr=
ef=3D"http://wss.example.com">wss.example.com</a>, and the proxy cannot see=
 their contents.</div>
<div><br></div><div>When the channel is idle, the browser sends a frame wit=
h opcode Suspend, the proxy confirms with opcode Suspend, and the connectio=
n between the browser and the proxy is closed. The connection between the p=
roxy and <a href=3D"http://wss.example.com">wss.example.com</a> stays open.=
</div>
<div><br></div><div>If the browser has a message to send, it reconnects to =
the proxy and sends a Resume frame.</div><div><br></div><div>If the proxy r=
eceives data from <a href=3D"http://wss.example.com">wss.example.com</a> wh=
ile suspended, it does whatever tranport-specific magic is needed to notify=
 the browser, and then the browser does the &quot;Resume&quot; handshake as=
 above, and receives the pending data.</div>
</div><div class=3D"gmail_extra"><br><br><div class=3D"gmail_quote">On 18 F=
ebruary 2014 05:09, Salvatore Loreto <span dir=3D"ltr">&lt;<a href=3D"mailt=
o:salvatore.loreto@ericsson.com" target=3D"_blank">salvatore.loreto@ericsso=
n.com</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">



<div style=3D"word-wrap:break-word"><div class=3D"">
<br>
<div>
<div>On Feb 17, 2014, at 4:56 AM, Adam Rice &lt;<a href=3D"mailto:ricea@goo=
gle.com" target=3D"_blank">ricea@google.com</a>&gt; wrote:</div>
<br>
<blockquote type=3D"cite"><span style=3D"font-family:Helvetica;font-size:me=
dium;font-style:normal;font-variant:normal;font-weight:normal;letter-spacin=
g:normal;line-height:normal;text-align:-webkit-auto;text-indent:0px;text-tr=
ansform:none;white-space:normal;word-spacing:0px;display:inline!important;f=
loat:none">With
 respect for Salvatore&#39;s point, if the suspend/resume facility existed =
you could in principle tunnel SSL over it (ie. within the WebSocket payload=
) and still get the benefit without trusting the proxy more than you alread=
y have to trust your network provider
 anyway.</span></blockquote>
<br>
</div>
</div><div>sorry then I don&#39;t understand how the proxy can provide supp=
ort if the suspend/resume facility are tunnelled over SSL=C2=A0</div>
<div><br>
</div>
<div>/Sal</div>
<br>
</div>

</blockquote></div><br></div>

--089e0112cd32ae7b0904f2a516c2--


From nobody Mon Feb 17 21:38:16 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 9D6AC1A02EE for <hybi@ietfa.amsl.com>; Mon, 17 Feb 2014 21:38:15 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.926
X-Spam-Level: 
X-Spam-Status: No, score=-1.926 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 9nqqpM4NmCC6 for <hybi@ietfa.amsl.com>; Mon, 17 Feb 2014 21:38:13 -0800 (PST)
Received: from mail-we0-x22f.google.com (mail-we0-x22f.google.com [IPv6:2a00:1450:400c:c03::22f]) by ietfa.amsl.com (Postfix) with ESMTP id 538B71A00D3 for <hybi@ietf.org>; Mon, 17 Feb 2014 21:38:13 -0800 (PST)
Received: by mail-we0-f175.google.com with SMTP id q59so11407289wes.34 for <hybi@ietf.org>; Mon, 17 Feb 2014 21:38:10 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc:content-type; bh=VVg5d0ZjWln0QEZJkQoaqzqZkWTumWTJd8NWwi+OPZI=; b=PwE3b2zmx7KjbyufyFMTb/qZ0NdQo+X1C+9mLXBLY6HxAZFjOEunNC566pXDFCPLeN XdAjEEYckCUCJ3qkb3OjIiy7s58wl24mmpFa+81EIYgqF3PKoWxaXz5Om2dgslEH/1K4 po7m4cDyNDyfzZ2tcE7GEMS4qJBpMLT3mUFMLcVdMyGjZw4Ryvi3bgiPiQHE6/CY9y+p QbTFrXwUSPWricVDIK+5TrWOQjv+im9s3LaucBLL4hAWlf/jpggp2TEA5fn/C0jIvQAz V4hVbOGlZILiCykRi5eofIZGHBh3YciM+Ddr/0vFFKwux+seqS4jNpTWlRUakbAwQiaN nDcg==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc:content-type; bh=VVg5d0ZjWln0QEZJkQoaqzqZkWTumWTJd8NWwi+OPZI=; b=B2Ehx132iCTl+94vYGC/5LVeiR9/vYbX5Z3dvg1EDrKwwoWrdfAD5pP+1X+oF1HNhK xg4j78VBLHy1MZZzXybGPhCOzevUaEkEuVzr3leR6hcUMTqbu6rFURnuWQqVRYrQvf0q NN2cHiDKv6y9ndlSksfY9BLJqgx85WkTdTf/4M24Hyog+gYgKcN2s9eAKI39bA4XIn8S UMYhGsAam4WsYFQFc9e65jQhc35SXLGwZdYDB+Q8snMDAgK6shMWSEk3+HSVgK5ZYYNf uyJsIZQMIV49bkthg9PLvGYR5sweULjCvgtBlBksjhBIc+jwwg6TeiFU9EOJ64tfSZvH 6+CQ==
X-Gm-Message-State: ALoCoQlOWpP6CUc5HLCfKP63KnM++dBFwkYxoDEGEGArgPFjD7Q7WV7YhIIKdvEIuDUSdS65tgpdO2ZghhTnpiX6LsO2wh1F50wcKEc3wKwceHPQo8xiAYhXgJgyJ/+X2WbmU56E9NbjUeFdM5px/gGzi63fp3/BWkUz5asI+OacpcIZRg5BPx9agPgI6iORiESju87jpVj2
X-Received: by 10.181.13.11 with SMTP id eu11mr15802442wid.30.1392701890086; Mon, 17 Feb 2014 21:38:10 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Mon, 17 Feb 2014 21:37:49 -0800 (PST)
In-Reply-To: <CAH9hSJYv5VXGjS7AfG12-ArEvE6Uj_eE6pPxAiQcchcgV8vHcg@mail.gmail.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <CACuKZqEcA1Pv8RpWfmThMjTzi2BbVMMKXqujs6BxVfxRPZJ9NQ@mail.gmail.com> <CAG4zZZCr4aTfVpw2coX2g0qw++4kdgNCFVze6tHKZ+fJNqb0aQ@mail.gmail.com> <CAG4zZZA4WR50Ea2e36XEgTgpJURwiMFn+DXeYuxHFfZatYzDOQ@mail.gmail.com> <CAH9hSJYv5VXGjS7AfG12-ArEvE6Uj_eE6pPxAiQcchcgV8vHcg@mail.gmail.com>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Tue, 18 Feb 2014 14:37:49 +0900
Message-ID: <CAH9hSJbzra7uz7yfKQwfZaP_jhnxwdZyx8JnwCmBGhiMk6rbtg@mail.gmail.com>
To: Joakim Erdfelt <joakim@intalio.com>
Content-Type: multipart/alternative; boundary=f46d043c7de66ff11404f2a7ad17
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/zHEPAxwRXjZzJrPJzSRRwXOgl3Q
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 18 Feb 2014 05:38:15 -0000

--f46d043c7de66ff11404f2a7ad17
Content-Type: text/plain; charset=UTF-8

So, to be clear, Joakim, you're not objecting to make it unable to know the
size of each frame, but objecting to make it completely unable to know the
size of big messages on receiving the head of it, right?

Your use case (fail-fast) seems to require the sender to send messages
unfragmented. So, it sounds like application (or kind of application)
specific needs. You could do some application specific hack such as
introducing "complete-message-size" header in the payload than depending on
protocol level length header. If not, clients in the wild which are out of
your control may do arbitrary fragmentation. FYI, Chromium's WebSocket
stack is being refactored and after that it starts sending big messages as
fragmented into small frames.

--f46d043c7de66ff11404f2a7ad17
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_extra">So, to be clear, Joakim, you&#3=
9;re not objecting to make it unable to know the size of each frame, but ob=
jecting to make it completely unable to know the size of big messages on re=
ceiving the head of it, right?</div>

<div class=3D"gmail_extra"><br></div><div class=3D"gmail_extra">Your use ca=
se (fail-fast) seems to require the sender to send messages unfragmented. S=
o, it sounds like application (or kind of application) specific needs. You =
could do some application specific hack such as introducing &quot;complete-=
message-size&quot; header in the payload than depending on protocol level l=
ength header. If not, clients in the wild which are out of your control may=
 do arbitrary fragmentation. FYI, Chromium&#39;s WebSocket stack is being r=
efactored and after that it starts sending big messages as fragmented into =
small frames.</div>

</div>

--f46d043c7de66ff11404f2a7ad17--


From nobody Mon Feb 17 22:02:43 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 9913C1A0358 for <hybi@ietfa.amsl.com>; Mon, 17 Feb 2014 22:02:42 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.926
X-Spam-Level: 
X-Spam-Status: No, score=-1.926 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 3Twd1P2ZSiSK for <hybi@ietfa.amsl.com>; Mon, 17 Feb 2014 22:02:40 -0800 (PST)
Received: from mail-wi0-x22a.google.com (mail-wi0-x22a.google.com [IPv6:2a00:1450:400c:c05::22a]) by ietfa.amsl.com (Postfix) with ESMTP id E4C5D1A0357 for <hybi@ietf.org>; Mon, 17 Feb 2014 22:02:39 -0800 (PST)
Received: by mail-wi0-f170.google.com with SMTP id hi5so3043271wib.3 for <hybi@ietf.org>; Mon, 17 Feb 2014 22:02:36 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc:content-type; bh=sY1P6MtHUne5Cov1yU1c0OKFWbQTUVjErwcrVouFzcg=; b=Im9XC6TlSzIuK5JE/4YZZaapz0D0c6FseCed5RV0/cYRIm+sryYpvMZBfac9cGrd2z NFNNNEfZVBiv4MctTF7QFcMDSDbTwEJOX/F0qdUG7T5RF1KkIP0mzT8klB3pX9pasbAO VI+6K1QnhykwwFUGaa3t+OREKRlRwBUokkr60VQ+qLW1mqoEhlEHhbsA32JNC65FwyCj i8Yf3AjXM7Q3c+Ci1aHWmM9OBAzZvmm34ERwd1mH9KQbhCSiulCs9EJkqmPRSxBTbM9X zcmH6G1ZeT7sg0kW6P7BSBdoTn3kg2EEM0T/e/tVgJ6HYuDlhpm8fpZjC8xVJzddSi60 2nVQ==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc:content-type; bh=sY1P6MtHUne5Cov1yU1c0OKFWbQTUVjErwcrVouFzcg=; b=CUDYGPHY8+0ap6DlJKPLMwKY2fPKcliZZ5GS6JjgrjOzWw/lAFpTSzjOYNuVyHIbDC YvMLi9jzktL5852qnYAz859qpXSTMTnOWQn5saBAPskgFiEW0/pwk4EPykI5TTGooaHO 7F1g53Py/udYoT4A7CAaw65OyXT1ytXuDBDeQQ3TrVK3uia0v9ihbkr9uA1nfxE2IvzS GyfRANXFzSTrXrqzpvkkiZpX5CBR7bKNNIQUcknpvUkX9eqITEJZAFOByLa7LrwKhSyI OYCn0bLa8k8BTS7bqq1wRAgWRiHqEslVe62o/KytIALzXNVMWGNxDssziWGMU0tDbDvQ nuZQ==
X-Gm-Message-State: ALoCoQlr6OlX0Nc9uFSueoByiYuqxaQxFnfcw1QmDtTHisgM4IbjKEQhbFXbjPih2cxbccs7uLmpv0FDxXQTM3sefQS7VVRahBk+ND7IE+CD3REVHCgBCmgs6P5xWfagExuHXzSGpXkkI/gqGtX2+HUEUqv9QE15aT8YphstauH7cUWCuR0LXXYpz0jndWhh2JJDHHITOJe+
X-Received: by 10.194.60.37 with SMTP id e5mr20337700wjr.32.1392703356533; Mon, 17 Feb 2014 22:02:36 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Mon, 17 Feb 2014 22:02:15 -0800 (PST)
In-Reply-To: <634914A010D0B943A035D226786325D4446BBDA3EC@EXVMBX020-12.exch020.serverdata.net>
References: <DC5D7086-B93B-4880-A202-BD0EEA89240B@zaphoyd.com> <CAG4zZZBuOX7U_JN5W-PQCzMwNr09be9=+T2vYkcW=_k9p18XvQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BBDA3EC@EXVMBX020-12.exch020.serverdata.net>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Tue, 18 Feb 2014 15:02:15 +0900
Message-ID: <CAH9hSJYviS5m1zAQs76KsyN_w6an5hWvLBT79o5jSYrem+kRig@mail.gmail.com>
To: Tobias Oberstein <tobias.oberstein@tavendo.de>
Content-Type: multipart/alternative; boundary=047d7ba979dad821e104f2a8043e
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/mcFdZM-TvcfzQxhy7glkAcxPCtQ
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Client behavior with blank server Sec-WebSocket-Protocol header
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 18 Feb 2014 06:02:42 -0000

--047d7ba979dad821e104f2a8043e
Content-Type: text/plain; charset=UTF-8

Sorry that none of us hasn't responded to this thread for long time, Peter.

The change which banned this usage is this one:
http://src.chromium.org/viewvc/blink?view=revision&revision=155603

As Joonas pointed, the WebSocket API spec is saying
req:non-empty/resp:empty combination is invalid for long time since this
change in 2010: http://html5.org/tools/web-apps-tracker?from=5172&to=5173

It was non-normative text, but now it has normative one:
http://lists.w3.org/Archives/Public/public-webapps-bugzilla/2013Oct/0215.html

In 2011, I started a thread to discuss what UA and server library should do
with subprotocol values:
http://www.ietf.org/mail-archive/web/hybi/current/msg07241.html. This
question corresponds to (c) in the post. I suggested either we do nothing
or define strict handling algorithm after giving clear semantics for
"absence of Sec-WebSocket-Protocol".

Takeshi


On Sat, Dec 28, 2013 at 8:23 PM, Tobias Oberstein <
tobias.oberstein@tavendo.de> wrote:

> >A quick autobahn server test, it returns no Sec-WebSocket-Protocol if
> there is no match.
> >A quick autobahn client test, the connection still occurs on no match.
>
> Yes, this is the default behavior of Autobahn: if the server does not
> speak any of the subprotocols announced
> by the client, the server won't return a Sec-WebSocket-Protocol header at
> all.
>
> And the client will only automatically fail the handshake if the server
> _did_ return a Sec-WebSocket-Protocol
> _and_ the header value is different from any of the subprotocols initially
> announced.
>
> This behavior can be tuned using the onConnect() hooks on server and
> client side.
>
> The default behavior is designed to follow the spec:
>
> "
>    6.  If the response includes a |Sec-WebSocket-Protocol| header field
>        and this header field indicates the use of a subprotocol that was
>        not present in the client's handshake (the server has indicated a
>        subprotocol not requested by the client), the client MUST _Fail
>        the WebSocket Connection_.
> "
>
> /Tobias
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi
>

--047d7ba979dad821e104f2a8043e
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Sorry that none of us hasn&#39;t responded to this thread =
for long time, Peter.<div><br></div><div>The change which banned this usage=
 is this one:=C2=A0<a href=3D"http://src.chromium.org/viewvc/blink?view=3Dr=
evision&amp;revision=3D155603">http://src.chromium.org/viewvc/blink?view=3D=
revision&amp;revision=3D155603</a></div>

<div><br></div><div>As Joonas pointed, the WebSocket API spec is saying req=
:non-empty/resp:empty combination is invalid for long time since this chang=
e in 2010:=C2=A0<a href=3D"http://html5.org/tools/web-apps-tracker?from=3D5=
172&amp;to=3D5173">http://html5.org/tools/web-apps-tracker?from=3D5172&amp;=
to=3D5173</a></div>

<div><br></div><div>It was non-normative text, but now it has normative one=
:=C2=A0<a href=3D"http://lists.w3.org/Archives/Public/public-webapps-bugzil=
la/2013Oct/0215.html">http://lists.w3.org/Archives/Public/public-webapps-bu=
gzilla/2013Oct/0215.html</a></div>

<div><br></div><div>In 2011, I started a thread to discuss what UA and serv=
er library should do with subprotocol values:=C2=A0<a href=3D"http://www.ie=
tf.org/mail-archive/web/hybi/current/msg07241.html">http://www.ietf.org/mai=
l-archive/web/hybi/current/msg07241.html</a>. This question corresponds to =
(c) in the post. I suggested either we do nothing or define strict handling=
 algorithm after giving clear semantics for &quot;absence of Sec-WebSocket-=
Protocol&quot;.</div>

</div><div class=3D"gmail_extra"><br clear=3D"all"><div>Takeshi</div>
<br><br><div class=3D"gmail_quote">On Sat, Dec 28, 2013 at 8:23 PM, Tobias =
Oberstein <span dir=3D"ltr">&lt;<a href=3D"mailto:tobias.oberstein@tavendo.=
de" target=3D"_blank">tobias.oberstein@tavendo.de</a>&gt;</span> wrote:<br>=
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">

<div class=3D"">&gt;A quick autobahn server test, it returns no Sec-WebSock=
et-Protocol if there is no match.<br>
&gt;A quick autobahn client test, the connection still occurs on no match.<=
br>
<br>
</div>Yes, this is the default behavior of Autobahn: if the server does not=
 speak any of the subprotocols announced<br>
by the client, the server won&#39;t return a Sec-WebSocket-Protocol header =
at all.<br>
<br>
And the client will only automatically fail the handshake if the server _di=
d_ return a Sec-WebSocket-Protocol<br>
_and_ the header value is different from any of the subprotocols initially =
announced.<br>
<br>
This behavior can be tuned using the onConnect() hooks on server and client=
 side.<br>
<br>
The default behavior is designed to follow the spec:<br>
<br>
&quot;<br>
=C2=A0 =C2=A06. =C2=A0If the response includes a |Sec-WebSocket-Protocol| h=
eader field<br>
=C2=A0 =C2=A0 =C2=A0 =C2=A0and this header field indicates the use of a sub=
protocol that was<br>
=C2=A0 =C2=A0 =C2=A0 =C2=A0not present in the client&#39;s handshake (the s=
erver has indicated a<br>
=C2=A0 =C2=A0 =C2=A0 =C2=A0subprotocol not requested by the client), the cl=
ient MUST _Fail<br>
=C2=A0 =C2=A0 =C2=A0 =C2=A0the WebSocket Connection_.<br>
&quot;<br>
<span class=3D"HOEnZb"><font color=3D"#888888"><br>
/Tobias<br>
</font></span><div class=3D"HOEnZb"><div class=3D"h5">_____________________=
__________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
</div></div></blockquote></div><br></div>

--047d7ba979dad821e104f2a8043e--


From nobody Tue Feb 18 10:08:48 2014
Return-Path: <joakim@intalio.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id A27B01A06E8 for <hybi@ietfa.amsl.com>; Tue, 18 Feb 2014 10:08:42 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -0.777
X-Spam-Level: 
X-Spam-Status: No, score=-0.777 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, J_CHICKENPOX_64=0.6, J_CHICKENPOX_75=0.6, RCVD_IN_DNSWL_LOW=-0.7] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id YwpiD9rWWc0i for <hybi@ietfa.amsl.com>; Tue, 18 Feb 2014 10:08:39 -0800 (PST)
Received: from mail-ee0-f44.google.com (mail-ee0-f44.google.com [74.125.83.44]) by ietfa.amsl.com (Postfix) with ESMTP id 7B2C71A0524 for <hybi@ietf.org>; Tue, 18 Feb 2014 10:08:37 -0800 (PST)
Received: by mail-ee0-f44.google.com with SMTP id c13so7938479eek.17 for <hybi@ietf.org>; Tue, 18 Feb 2014 10:08:33 -0800 (PST)
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=WbRqLQVS8G6jgUWsz6nHCelSkN/u46EAMz1BrJnOHtA=; b=eWW78xMl8Idz/3hDP/o+Ifd88Jf/Ia5eDtwD494Ld0Eu6OIkPwl6MhKWtNo0XhO3MW MQcStgRMxIKdgaofaSsTRRuAEpuAqOHtDgCbjxvcIAm60W5p0xGkRdNNNZKFCb9gN5Y5 OsNwyTeOlgQlvLRv8nd/0uoVwo/SkyzPFT/FvspwniqSSZ5rUxB/ZJguHEzHmI3GRadn uWOfab6KD2RgPAJkjao2SXKATckxvBn7+Hoog9HHm87Z+UKAtpdidqU8LFn+zgdaR8yA T2IK6Lk68nfaQ5p2WPZV6ydi+zGY1/+Dt4oNtvtXNLCPyujK6zIHyFHdvLJ9t3th7e61 L4qg==
X-Gm-Message-State: ALoCoQleG9offtXragmwx9KE0gEil1poOMTRD28PXHGJbGdztVZzqQ6SPwprSI+iXV3P+1KvAg/n
MIME-Version: 1.0
X-Received: by 10.14.103.67 with SMTP id e43mr205786eeg.94.1392746913896; Tue, 18 Feb 2014 10:08:33 -0800 (PST)
Received: by 10.14.152.202 with HTTP; Tue, 18 Feb 2014 10:08:33 -0800 (PST)
In-Reply-To: <CAH9hSJbzra7uz7yfKQwfZaP_jhnxwdZyx8JnwCmBGhiMk6rbtg@mail.gmail.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <CACuKZqEcA1Pv8RpWfmThMjTzi2BbVMMKXqujs6BxVfxRPZJ9NQ@mail.gmail.com> <CAG4zZZCr4aTfVpw2coX2g0qw++4kdgNCFVze6tHKZ+fJNqb0aQ@mail.gmail.com> <CAG4zZZA4WR50Ea2e36XEgTgpJURwiMFn+DXeYuxHFfZatYzDOQ@mail.gmail.com> <CAH9hSJYv5VXGjS7AfG12-ArEvE6Uj_eE6pPxAiQcchcgV8vHcg@mail.gmail.com> <CAH9hSJbzra7uz7yfKQwfZaP_jhnxwdZyx8JnwCmBGhiMk6rbtg@mail.gmail.com>
Date: Tue, 18 Feb 2014 11:08:33 -0700
Message-ID: <CAG4zZZC1cmaH36znAvVLZE_MyJ+ThNk1Ky7tQ-QnKf_qwiQNDw@mail.gmail.com>
From: Joakim Erdfelt <joakim@intalio.com>
To: Takeshi Yoshino <tyoshino@google.com>
Content-Type: multipart/alternative; boundary=001a11c289ac10e8b704f2b229ba
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/ZZAQyHGTETsAGPTIYHolq3vKJEc
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 18 Feb 2014 18:08:42 -0000

--001a11c289ac10e8b704f2b229ba
Content-Type: text/plain; charset=ISO-8859-1

I'm not sure I was clear, as you seem to still have questions and are
hinting at alternatives.

The most important thing I said:
   The size is important at the start of any payload about to be received.

All other statements are based off of that.
The nuanced questions you are asking seems to indicate that you treat a
frame differently than a message is irrelevant.
If the frame happens to represent a complete message (eg:
opcode=(text|binary) + fin=true) vs being only 1 frame belonging to a
series of frames that make up a whole message makes no difference to the
statement.

Since websocket is message based, not stream based, the above core
statement remains true.

The fail-fast example applies for fragmented or whole messages.

some pseudo code:

   header = frame.parse_header(network_bytes)
   if (header.size > system_memory_constraints) then
      fail_protocol(1009)
      return
   end
   // is message still being collected and not yet delivered to local
endpoint?
   if (message.delivered == false) then
      if (message.size + header.size > message.max_constraints) then
        fail_protocol(1009)
        return
      else
        payload = frame.read_payload(network_bytes)
        message.append(payload)
      end
   end
   if (header.fin == true) then
      message.deliver_to(local_endpoint)
      message.reset
   end


--
Joakim Erdfelt <joakim@intalio.com>
webtide.com <http://www.webtide.com/> - intalio.com/jetty
Expert advice, services and support from from the Jetty & CometD experts
eclipse.org/jetty - cometd.org


On Mon, Feb 17, 2014 at 10:37 PM, Takeshi Yoshino <tyoshino@google.com>wrote:

> So, to be clear, Joakim, you're not objecting to make it unable to know
> the size of each frame, but objecting to make it completely unable to know
> the size of big messages on receiving the head of it, right?
>
> Your use case (fail-fast) seems to require the sender to send messages
> unfragmented. So, it sounds like application (or kind of application)
> specific needs. You could do some application specific hack such as
> introducing "complete-message-size" header in the payload than depending on
> protocol level length header. If not, clients in the wild which are out of
> your control may do arbitrary fragmentation. FYI, Chromium's WebSocket
> stack is being refactored and after that it starts sending big messages as
> fragmented into small frames.
>

--001a11c289ac10e8b704f2b229ba
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div>I&#39;m not sure I was clear, as you seem to still ha=
ve questions and are hinting at alternatives.</div><div><br></div><div>The =
most important thing I said:</div>=A0 =A0The size is important at the start=
 of any payload about to be received.<div>
<br></div><div>All other statements are based off of that.</div><div>The nu=
anced questions you are asking seems to indicate that you treat a frame dif=
ferently than a message is irrelevant.</div><div>If the frame happens to re=
present a complete message (eg: opcode=3D(text|binary) + fin=3Dtrue) vs bei=
ng only 1 frame belonging to a series of frames that make up a whole messag=
e makes no difference to the statement.</div>
<div><br></div><div>Since websocket is message based, not stream based, the=
 above core statement remains true.</div><div><br></div><div>The fail-fast =
example applies for fragmented or whole messages.</div><div><br></div><div>
some pseudo code:</div><div><br></div><div>=A0 =A0header =3D frame.parse_he=
ader(network_bytes)</div><div>=A0 =A0if (header.size &gt; system_memory_con=
straints) then</div><div>=A0 =A0 =A0 fail_protocol(1009)</div><div>=A0 =A0 =
=A0 return</div><div>
=A0 =A0end</div><div>=A0 =A0// is message still being collected and not yet=
 delivered to local endpoint?</div><div>=A0 =A0if (message.delivered =3D=3D=
 false) then</div><div>=A0 =A0 =A0 if (message.size + header.size &gt; mess=
age.max_constraints) then</div>
<div>=A0 =A0 =A0 =A0 fail_protocol(1009)</div><div>=A0 =A0 =A0 =A0 return</=
div><div>=A0 =A0 =A0 else</div><div>=A0 =A0 =A0 =A0 payload =3D frame.read_=
payload(network_bytes)</div><div>=A0 =A0 =A0 =A0 message.append(payload)</d=
iv><div>=A0 =A0 =A0 end</div><div>=A0 =A0end</div>
<div>=A0 =A0if (header.fin =3D=3D true) then</div><div>=A0 =A0 =A0 message.=
deliver_to(local_endpoint)</div><div>=A0 =A0 =A0 message.reset</div><div>=
=A0 =A0end</div><div><br></div></div><div class=3D"gmail_extra"><br clear=
=3D"all"><div><div dir=3D"ltr">
<div>--</div><div>Joakim Erdfelt &lt;<a href=3D"mailto:joakim@intalio.com" =
target=3D"_blank">joakim@intalio.com</a>&gt;</div><div><a href=3D"http://ww=
w.webtide.com/" target=3D"_blank">webtide.com</a>=A0- <a href=3D"http://int=
alio.com/jetty" target=3D"_blank">intalio.com/jetty</a></div>
<div><span>Expert advice, services and support from=A0</span><span>from the=
 Jetty &amp; CometD experts</span></div><div><a href=3D"http://eclipse.org/=
jetty/" target=3D"_blank">eclipse.org/jetty</a>=A0-=A0<a href=3D"http://com=
etd.org/" target=3D"_blank">cometd.org</a></div>
</div></div>
<br><br><div class=3D"gmail_quote">On Mon, Feb 17, 2014 at 10:37 PM, Takesh=
i Yoshino <span dir=3D"ltr">&lt;<a href=3D"mailto:tyoshino@google.com" targ=
et=3D"_blank">tyoshino@google.com</a>&gt;</span> wrote:<br><blockquote clas=
s=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;pad=
ding-left:1ex">
<div dir=3D"ltr"><div class=3D"gmail_extra">So, to be clear, Joakim, you&#3=
9;re not objecting to make it unable to know the size of each frame, but ob=
jecting to make it completely unable to know the size of big messages on re=
ceiving the head of it, right?</div>


<div class=3D"gmail_extra"><br></div><div class=3D"gmail_extra">Your use ca=
se (fail-fast) seems to require the sender to send messages unfragmented. S=
o, it sounds like application (or kind of application) specific needs. You =
could do some application specific hack such as introducing &quot;complete-=
message-size&quot; header in the payload than depending on protocol level l=
ength header. If not, clients in the wild which are out of your control may=
 do arbitrary fragmentation. FYI, Chromium&#39;s WebSocket stack is being r=
efactored and after that it starts sending big messages as fragmented into =
small frames.</div>


</div>
</blockquote></div><br></div>

--001a11c289ac10e8b704f2b229ba--


From nobody Tue Feb 18 10:12:12 2014
Return-Path: <joakim@intalio.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 549DC1A0512 for <hybi@ietfa.amsl.com>; Tue, 18 Feb 2014 10:12:11 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -0.777
X-Spam-Level: 
X-Spam-Status: No, score=-0.777 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, J_CHICKENPOX_64=0.6, J_CHICKENPOX_75=0.6, RCVD_IN_DNSWL_LOW=-0.7] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Uz2HvgHGbHli for <hybi@ietfa.amsl.com>; Tue, 18 Feb 2014 10:12:09 -0800 (PST)
Received: from mail-ee0-f51.google.com (mail-ee0-f51.google.com [74.125.83.51]) by ietfa.amsl.com (Postfix) with ESMTP id E158B1A010A for <hybi@ietf.org>; Tue, 18 Feb 2014 10:12:08 -0800 (PST)
Received: by mail-ee0-f51.google.com with SMTP id b57so7911955eek.24 for <hybi@ietf.org>; Tue, 18 Feb 2014 10:12:05 -0800 (PST)
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=YOe6xTNWLP2dBaA7PZU9J0FUsCT8jDc7FTbgZQfQTi4=; b=mQoQR3yuHjQWpwrGK7STilPDfXgAXKb/QUZh0ERyNQ+rh3IpfakaB2q2pqjav/ncsz ydOiRpUL5HK+Y3vfqf8POrOuSKlcfYx4P+6EIrJQQf+CFnkGKInTliMvuy9f5UQT0UY4 CVqFVS1QF9qgmtZQ3ojTM9XYvQWLJjjj8D3vwBhynchA+zj8XPJpv+ndv8JfIkEZAt5g SeqJvLkIiUjCaCQ2FKIxrd4Ok9BxJFtbujhDX0SkbPNYvrm0yHoaao6f/VT0D01hVSKy ENo59rNGJeOOey7AqO96ADFR/GCG/VA0w4U4ADNGROXCEMbIfFQgC4sYp2x/rVWqtSH4 peVQ==
X-Gm-Message-State: ALoCoQkAcZA42MJx8N0NnBjWciJsp32LCng09Sy2639nLyQI4TK4cnfAcWWUS7QOKtn8lrAQ/Cs7
MIME-Version: 1.0
X-Received: by 10.14.119.197 with SMTP id n45mr225572eeh.93.1392747125485; Tue, 18 Feb 2014 10:12:05 -0800 (PST)
Received: by 10.14.152.202 with HTTP; Tue, 18 Feb 2014 10:12:05 -0800 (PST)
In-Reply-To: <CAG4zZZC1cmaH36znAvVLZE_MyJ+ThNk1Ky7tQ-QnKf_qwiQNDw@mail.gmail.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <CACuKZqEcA1Pv8RpWfmThMjTzi2BbVMMKXqujs6BxVfxRPZJ9NQ@mail.gmail.com> <CAG4zZZCr4aTfVpw2coX2g0qw++4kdgNCFVze6tHKZ+fJNqb0aQ@mail.gmail.com> <CAG4zZZA4WR50Ea2e36XEgTgpJURwiMFn+DXeYuxHFfZatYzDOQ@mail.gmail.com> <CAH9hSJYv5VXGjS7AfG12-ArEvE6Uj_eE6pPxAiQcchcgV8vHcg@mail.gmail.com> <CAH9hSJbzra7uz7yfKQwfZaP_jhnxwdZyx8JnwCmBGhiMk6rbtg@mail.gmail.com> <CAG4zZZC1cmaH36znAvVLZE_MyJ+ThNk1Ky7tQ-QnKf_qwiQNDw@mail.gmail.com>
Date: Tue, 18 Feb 2014 11:12:05 -0700
Message-ID: <CAG4zZZB3h6TTFLUp-ucYQP7tMf-8=h5fh1UqufMh3w-0JN+hSw@mail.gmail.com>
From: Joakim Erdfelt <joakim@intalio.com>
To: Takeshi Yoshino <tyoshino@google.com>
Content-Type: multipart/alternative; boundary=001a1132f7f8ad47f604f2b23516
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/TT5IoAC35d4Y6GN3kIler_LSSPg
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 18 Feb 2014 18:12:11 -0000

--001a1132f7f8ad47f604f2b23516
Content-Type: text/plain; charset=ISO-8859-1

Clicked too soon on that pseudo code.
This is more clear. (i hope)

   header = frame.parse_header(network_bytes)
   if (header.size > system_memory_constraints) then
      fail_protocol(1009)
      // do not read payload
      return
   end
   if (message.size + header.size > message.max_constraints) then
      fail_protocol(1009)
      // do not read payload
      return
   else
      payload = frame.read_payload(network_bytes)
      message.append(payload)
   end
   if (header.fin == true) then
      message.deliver_to(local_endpoint)
      // free up any message resources, reset message size to 0
      message.reset
   end


--
Joakim Erdfelt <joakim@intalio.com>
webtide.com <http://www.webtide.com/> - intalio.com/jetty
Expert advice, services and support from from the Jetty & CometD experts
eclipse.org/jetty - cometd.org


On Tue, Feb 18, 2014 at 11:08 AM, Joakim Erdfelt <joakim@intalio.com> wrote:

> I'm not sure I was clear, as you seem to still have questions and are
> hinting at alternatives.
>
> The most important thing I said:
>    The size is important at the start of any payload about to be received.
>
> All other statements are based off of that.
> The nuanced questions you are asking seems to indicate that you treat a
> frame differently than a message is irrelevant.
> If the frame happens to represent a complete message (eg:
> opcode=(text|binary) + fin=true) vs being only 1 frame belonging to a
> series of frames that make up a whole message makes no difference to the
> statement.
>
> Since websocket is message based, not stream based, the above core
> statement remains true.
>
> The fail-fast example applies for fragmented or whole messages.
>
> some pseudo code:
>
>    header = frame.parse_header(network_bytes)
>    if (header.size > system_memory_constraints) then
>       fail_protocol(1009)
>       return
>    end
>    // is message still being collected and not yet delivered to local
> endpoint?
>    if (message.delivered == false) then
>       if (message.size + header.size > message.max_constraints) then
>         fail_protocol(1009)
>         return
>       else
>         payload = frame.read_payload(network_bytes)
>         message.append(payload)
>       end
>    end
>    if (header.fin == true) then
>       message.deliver_to(local_endpoint)
>       message.reset
>    end
>
>
> --
> Joakim Erdfelt <joakim@intalio.com>
> webtide.com <http://www.webtide.com/> - intalio.com/jetty
> Expert advice, services and support from from the Jetty & CometD experts
> eclipse.org/jetty - cometd.org
>
>
> On Mon, Feb 17, 2014 at 10:37 PM, Takeshi Yoshino <tyoshino@google.com>wrote:
>
>> So, to be clear, Joakim, you're not objecting to make it unable to know
>> the size of each frame, but objecting to make it completely unable to know
>> the size of big messages on receiving the head of it, right?
>>
>> Your use case (fail-fast) seems to require the sender to send messages
>> unfragmented. So, it sounds like application (or kind of application)
>> specific needs. You could do some application specific hack such as
>> introducing "complete-message-size" header in the payload than depending on
>> protocol level length header. If not, clients in the wild which are out of
>> your control may do arbitrary fragmentation. FYI, Chromium's WebSocket
>> stack is being refactored and after that it starts sending big messages as
>> fragmented into small frames.
>>
>
>

--001a1132f7f8ad47f604f2b23516
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Clicked too soon on that pseudo code.<div>This is more cle=
ar. (i hope)</div><div><br></div><div><div>=A0 =A0header =3D frame.parse_he=
ader(network_bytes)</div><div>=A0 =A0if (header.size &gt; system_memory_con=
straints) then</div>
<div>=A0 =A0 =A0 fail_protocol(1009)</div><div>=A0 =A0 =A0 // do not read p=
ayload</div><div>=A0 =A0 =A0 return</div><div>=A0 =A0end</div><div>=A0 =A0i=
f (message.size + header.size &gt; message.max_constraints) then</div><div>=
=A0 =A0 =A0 fail_protocol(1009)</div>
<div>=A0 =A0 =A0 // do not read payload</div><div>=A0 =A0 =A0 return</div><=
div>=A0 =A0else</div><div>=A0 =A0 =A0 payload =3D frame.read_payload(networ=
k_bytes)</div><div>=A0 =A0 =A0 message.append(payload)</div><div>=A0 =A0end=
</div><div>=A0 =A0if (header.fin =3D=3D true) then</div>
<div>=A0 =A0 =A0 message.deliver_to(local_endpoint)</div><div>=A0 =A0 =A0 /=
/ free up any message resources, reset message size to 0</div><div>=A0 =A0 =
=A0 message.reset</div><div>=A0 =A0end</div></div><div><br></div></div><div=
 class=3D"gmail_extra">
<br clear=3D"all"><div><div dir=3D"ltr"><div>--</div><div>Joakim Erdfelt &l=
t;<a href=3D"mailto:joakim@intalio.com" target=3D"_blank">joakim@intalio.co=
m</a>&gt;</div><div><a href=3D"http://www.webtide.com/" target=3D"_blank">w=
ebtide.com</a>=A0- <a href=3D"http://intalio.com/jetty" target=3D"_blank">i=
ntalio.com/jetty</a></div>
<div><span>Expert advice, services and support from=A0</span><span>from the=
 Jetty &amp; CometD experts</span></div><div><a href=3D"http://eclipse.org/=
jetty/" target=3D"_blank">eclipse.org/jetty</a>=A0-=A0<a href=3D"http://com=
etd.org/" target=3D"_blank">cometd.org</a></div>
</div></div>
<br><br><div class=3D"gmail_quote">On Tue, Feb 18, 2014 at 11:08 AM, Joakim=
 Erdfelt <span dir=3D"ltr">&lt;<a href=3D"mailto:joakim@intalio.com" target=
=3D"_blank">joakim@intalio.com</a>&gt;</span> wrote:<br><blockquote class=
=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padd=
ing-left:1ex">
<div dir=3D"ltr"><div>I&#39;m not sure I was clear, as you seem to still ha=
ve questions and are hinting at alternatives.</div><div><br></div><div>The =
most important thing I said:</div>=A0 =A0The size is important at the start=
 of any payload about to be received.<div>

<br></div><div>All other statements are based off of that.</div><div>The nu=
anced questions you are asking seems to indicate that you treat a frame dif=
ferently than a message is irrelevant.</div><div>If the frame happens to re=
present a complete message (eg: opcode=3D(text|binary) + fin=3Dtrue) vs bei=
ng only 1 frame belonging to a series of frames that make up a whole messag=
e makes no difference to the statement.</div>

<div><br></div><div>Since websocket is message based, not stream based, the=
 above core statement remains true.</div><div><br></div><div>The fail-fast =
example applies for fragmented or whole messages.</div><div><br></div>
<div>
some pseudo code:</div><div><br></div><div>=A0 =A0header =3D frame.parse_he=
ader(network_bytes)</div><div>=A0 =A0if (header.size &gt; system_memory_con=
straints) then</div><div>=A0 =A0 =A0 fail_protocol(1009)</div><div>=A0 =A0 =
=A0 return</div><div>

=A0 =A0end</div><div>=A0 =A0// is message still being collected and not yet=
 delivered to local endpoint?</div><div>=A0 =A0if (message.delivered =3D=3D=
 false) then</div><div>=A0 =A0 =A0 if (message.size + header.size &gt; mess=
age.max_constraints) then</div>

<div>=A0 =A0 =A0 =A0 fail_protocol(1009)</div><div>=A0 =A0 =A0 =A0 return</=
div><div>=A0 =A0 =A0 else</div><div>=A0 =A0 =A0 =A0 payload =3D frame.read_=
payload(network_bytes)</div><div>=A0 =A0 =A0 =A0 message.append(payload)</d=
iv><div>=A0 =A0 =A0 end</div><div>=A0 =A0end</div>

<div>=A0 =A0if (header.fin =3D=3D true) then</div><div>=A0 =A0 =A0 message.=
deliver_to(local_endpoint)</div><div>=A0 =A0 =A0 message.reset</div><div>=
=A0 =A0end</div><div><br></div></div><div class=3D"gmail_extra"><div class=
=3D""><br clear=3D"all"><div>
<div dir=3D"ltr">
<div>--</div><div>Joakim Erdfelt &lt;<a href=3D"mailto:joakim@intalio.com" =
target=3D"_blank">joakim@intalio.com</a>&gt;</div><div><a href=3D"http://ww=
w.webtide.com/" target=3D"_blank">webtide.com</a>=A0- <a href=3D"http://int=
alio.com/jetty" target=3D"_blank">intalio.com/jetty</a></div>

<div><span>Expert advice, services and support from=A0</span><span>from the=
 Jetty &amp; CometD experts</span></div><div><a href=3D"http://eclipse.org/=
jetty/" target=3D"_blank">eclipse.org/jetty</a>=A0-=A0<a href=3D"http://com=
etd.org/" target=3D"_blank">cometd.org</a></div>

</div></div>
<br><br></div><div class=3D""><div class=3D"gmail_quote">On Mon, Feb 17, 20=
14 at 10:37 PM, Takeshi Yoshino <span dir=3D"ltr">&lt;<a href=3D"mailto:tyo=
shino@google.com" target=3D"_blank">tyoshino@google.com</a>&gt;</span> wrot=
e:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
<div dir=3D"ltr"><div class=3D"gmail_extra">So, to be clear, Joakim, you&#3=
9;re not objecting to make it unable to know the size of each frame, but ob=
jecting to make it completely unable to know the size of big messages on re=
ceiving the head of it, right?</div>



<div class=3D"gmail_extra"><br></div><div class=3D"gmail_extra">Your use ca=
se (fail-fast) seems to require the sender to send messages unfragmented. S=
o, it sounds like application (or kind of application) specific needs. You =
could do some application specific hack such as introducing &quot;complete-=
message-size&quot; header in the payload than depending on protocol level l=
ength header. If not, clients in the wild which are out of your control may=
 do arbitrary fragmentation. FYI, Chromium&#39;s WebSocket stack is being r=
efactored and after that it starts sending big messages as fragmented into =
small frames.</div>



</div>
</blockquote></div><br></div></div>
</blockquote></div><br></div>

--001a1132f7f8ad47f604f2b23516--


From nobody Tue Feb 18 13:33:19 2014
Return-Path: <webmaster@zaphoyd.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 123CD1A0081 for <hybi@ietfa.amsl.com>; Tue, 18 Feb 2014 13:33:13 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.899
X-Spam-Level: 
X-Spam-Status: No, score=-1.899 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_NONE=-0.0001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id yEHI-mo6gKPH for <hybi@ietfa.amsl.com>; Tue, 18 Feb 2014 13:33:09 -0800 (PST)
Received: from authsmtp01.uchicago.edu (authsmtp01.uchicago.edu [128.135.12.121]) by ietfa.amsl.com (Postfix) with ESMTP id C6E751A0470 for <hybi@ietf.org>; Tue, 18 Feb 2014 13:33:08 -0800 (PST)
Received: from ranna.uchicago.edu (ranna.uchicago.edu [128.135.45.206]) (Authenticated sender: zaphoyd) by authsmtp01.uchicago.edu (Postfix) with ESMTP id 5037549802B; Tue, 18 Feb 2014 15:33:05 -0600 (CST)
Content-Type: multipart/alternative; boundary="Apple-Mail=_33DF1344-C92E-4326-99CC-EF5F18600338"
Mime-Version: 1.0 (Mac OS X Mail 7.1 \(1827\))
From: Peter Thorson <webmaster@zaphoyd.com>
In-Reply-To: <CAH9hSJYviS5m1zAQs76KsyN_w6an5hWvLBT79o5jSYrem+kRig@mail.gmail.com>
Date: Tue, 18 Feb 2014 15:33:04 -0600
Message-Id: <EEE7DF44-6633-4474-95ED-8CFF3C714AB7@zaphoyd.com>
References: <DC5D7086-B93B-4880-A202-BD0EEA89240B@zaphoyd.com> <CAG4zZZBuOX7U_JN5W-PQCzMwNr09be9=+T2vYkcW=_k9p18XvQ@mail.gmail.com> <634914A010D0B943A035D226786325D4446BBDA3EC@EXVMBX020-12.exch020.serverdata.net> <CAH9hSJYviS5m1zAQs76KsyN_w6an5hWvLBT79o5jSYrem+kRig@mail.gmail.com>
To: Takeshi Yoshino <tyoshino@google.com>
X-Mailer: Apple Mail (2.1827)
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/Mut-LLWh6NhDU75S3V9dy93th0s
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Client behavior with blank server Sec-WebSocket-Protocol header
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 18 Feb 2014 21:33:13 -0000

--Apple-Mail=_33DF1344-C92E-4326-99CC-EF5F18600338
Content-Transfer-Encoding: quoted-printable
Content-Type: text/plain;
	charset=windows-1252

Takeshi,

Thanks for the links here. I now see the chain of events that lead to =
this being clarified in the W3C spec and the resulting change in Chrome =
that now enforces this. I don=92t agree with the decision that was made =
but it is good to have it clarified one way or the other. I can write =
some more definitive documentation now to answer all of the =93why does =
your library not work in Chrome anymore?=94 questions I=92ve been =
getting and offer sane workarounds.

Best,

Peter

On Feb 18, 2014, at 24:02 , Takeshi Yoshino <tyoshino@google.com> wrote:

> Sorry that none of us hasn't responded to this thread for long time, =
Peter.
>=20
> The change which banned this usage is this one: =
http://src.chromium.org/viewvc/blink?view=3Drevision&revision=3D155603
>=20
> As Joonas pointed, the WebSocket API spec is saying =
req:non-empty/resp:empty combination is invalid for long time since this =
change in 2010: =
http://html5.org/tools/web-apps-tracker?from=3D5172&to=3D5173
>=20
> It was non-normative text, but now it has normative one: =
http://lists.w3.org/Archives/Public/public-webapps-bugzilla/2013Oct/0215.h=
tml
>=20
> In 2011, I started a thread to discuss what UA and server library =
should do with subprotocol values: =
http://www.ietf.org/mail-archive/web/hybi/current/msg07241.html. This =
question corresponds to (c) in the post. I suggested either we do =
nothing or define strict handling algorithm after giving clear semantics =
for "absence of Sec-WebSocket-Protocol".
>=20
> Takeshi
>=20
>=20
> On Sat, Dec 28, 2013 at 8:23 PM, Tobias Oberstein =
<tobias.oberstein@tavendo.de> wrote:
> >A quick autobahn server test, it returns no Sec-WebSocket-Protocol if =
there is no match.
> >A quick autobahn client test, the connection still occurs on no =
match.
>=20
> Yes, this is the default behavior of Autobahn: if the server does not =
speak any of the subprotocols announced
> by the client, the server won't return a Sec-WebSocket-Protocol header =
at all.
>=20
> And the client will only automatically fail the handshake if the =
server _did_ return a Sec-WebSocket-Protocol
> _and_ the header value is different from any of the subprotocols =
initially announced.
>=20
> This behavior can be tuned using the onConnect() hooks on server and =
client side.
>=20
> The default behavior is designed to follow the spec:
>=20
> "
>    6.  If the response includes a |Sec-WebSocket-Protocol| header =
field
>        and this header field indicates the use of a subprotocol that =
was
>        not present in the client's handshake (the server has indicated =
a
>        subprotocol not requested by the client), the client MUST _Fail
>        the WebSocket Connection_.
> "
>=20
> /Tobias
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi
>=20
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi


--Apple-Mail=_33DF1344-C92E-4326-99CC-EF5F18600338
Content-Transfer-Encoding: quoted-printable
Content-Type: text/html;
	charset=windows-1252

<html><head><meta http-equiv=3D"Content-Type" content=3D"text/html =
charset=3Dwindows-1252"></head><body style=3D"word-wrap: break-word; =
-webkit-nbsp-mode: space; -webkit-line-break: =
after-white-space;">Takeshi,<div><br></div><div>Thanks for the links =
here. I now see the chain of events that lead to this being clarified in =
the W3C spec and the resulting change in Chrome that now enforces this. =
I don=92t agree with the decision that was made but it is good to have =
it clarified one way or the other. I can write some more definitive =
documentation now to answer all of the =93why does your library not work =
in Chrome anymore?=94 questions I=92ve been getting and offer sane =
workarounds.</div><div><br></div><div>Best,</div><div><br></div><div>Peter=
</div><div><br><div><div>On Feb 18, 2014, at 24:02 , Takeshi Yoshino =
&lt;<a href=3D"mailto:tyoshino@google.com">tyoshino@google.com</a>&gt; =
wrote:</div><br class=3D"Apple-interchange-newline"><blockquote =
type=3D"cite"><div dir=3D"ltr">Sorry that none of us hasn't responded to =
this thread for long time, Peter.<div><br></div><div>The change which =
banned this usage is this one:&nbsp;<a =
href=3D"http://src.chromium.org/viewvc/blink?view=3Drevision&amp;revision=3D=
155603">http://src.chromium.org/viewvc/blink?view=3Drevision&amp;revision=3D=
155603</a></div>

<div><br></div><div>As Joonas pointed, the WebSocket API spec is saying =
req:non-empty/resp:empty combination is invalid for long time since this =
change in 2010:&nbsp;<a =
href=3D"http://html5.org/tools/web-apps-tracker?from=3D5172&amp;to=3D5173"=
>http://html5.org/tools/web-apps-tracker?from=3D5172&amp;to=3D5173</a></di=
v>

<div><br></div><div>It was non-normative text, but now it has normative =
one:&nbsp;<a =
href=3D"http://lists.w3.org/Archives/Public/public-webapps-bugzilla/2013Oc=
t/0215.html">http://lists.w3.org/Archives/Public/public-webapps-bugzilla/2=
013Oct/0215.html</a></div>

<div><br></div><div>In 2011, I started a thread to discuss what UA and =
server library should do with subprotocol values:&nbsp;<a =
href=3D"http://www.ietf.org/mail-archive/web/hybi/current/msg07241.html">h=
ttp://www.ietf.org/mail-archive/web/hybi/current/msg07241.html</a>. This =
question corresponds to (c) in the post. I suggested either we do =
nothing or define strict handling algorithm after giving clear semantics =
for "absence of Sec-WebSocket-Protocol".</div>

</div><div class=3D"gmail_extra"><br clear=3D"all"><div>Takeshi</div>
<br><br><div class=3D"gmail_quote">On Sat, Dec 28, 2013 at 8:23 PM, =
Tobias Oberstein <span dir=3D"ltr">&lt;<a =
href=3D"mailto:tobias.oberstein@tavendo.de" =
target=3D"_blank">tobias.oberstein@tavendo.de</a>&gt;</span> =
wrote:<br><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 =
.8ex;border-left:1px #ccc solid;padding-left:1ex">

<div class=3D"">&gt;A quick autobahn server test, it returns no =
Sec-WebSocket-Protocol if there is no match.<br>
&gt;A quick autobahn client test, the connection still occurs on no =
match.<br>
<br>
</div>Yes, this is the default behavior of Autobahn: if the server does =
not speak any of the subprotocols announced<br>
by the client, the server won't return a Sec-WebSocket-Protocol header =
at all.<br>
<br>
And the client will only automatically fail the handshake if the server =
_did_ return a Sec-WebSocket-Protocol<br>
_and_ the header value is different from any of the subprotocols =
initially announced.<br>
<br>
This behavior can be tuned using the onConnect() hooks on server and =
client side.<br>
<br>
The default behavior is designed to follow the spec:<br>
<br>
"<br>
&nbsp; &nbsp;6. &nbsp;If the response includes a =
|Sec-WebSocket-Protocol| header field<br>
&nbsp; &nbsp; &nbsp; &nbsp;and this header field indicates the use of a =
subprotocol that was<br>
&nbsp; &nbsp; &nbsp; &nbsp;not present in the client's handshake (the =
server has indicated a<br>
&nbsp; &nbsp; &nbsp; &nbsp;subprotocol not requested by the client), the =
client MUST _Fail<br>
&nbsp; &nbsp; &nbsp; &nbsp;the WebSocket Connection_.<br>
"<br>
<span class=3D"HOEnZb"><font color=3D"#888888"><br>
/Tobias<br>
</font></span><div class=3D"HOEnZb"><div =
class=3D"h5">_______________________________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" =
target=3D"_blank">https://www.ietf.org/mailman/listinfo/hybi</a><br>
</div></div></blockquote></div><br></div>
_______________________________________________<br>hybi mailing =
list<br><a =
href=3D"mailto:hybi@ietf.org">hybi@ietf.org</a><br>https://www.ietf.org/ma=
ilman/listinfo/hybi<br></blockquote></div><br></div></body></html>=

--Apple-Mail=_33DF1344-C92E-4326-99CC-EF5F18600338--


From nobody Tue Feb 18 19:59:56 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 22B9E1A0253 for <hybi@ietfa.amsl.com>; Tue, 18 Feb 2014 19:59:48 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -0.726
X-Spam-Level: 
X-Spam-Status: No, score=-0.726 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, J_CHICKENPOX_64=0.6, J_CHICKENPOX_75=0.6, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id IjDWadVRoTx3 for <hybi@ietfa.amsl.com>; Tue, 18 Feb 2014 19:59:45 -0800 (PST)
Received: from mail-wg0-x233.google.com (mail-wg0-x233.google.com [IPv6:2a00:1450:400c:c00::233]) by ietfa.amsl.com (Postfix) with ESMTP id 92F721A0329 for <hybi@ietf.org>; Tue, 18 Feb 2014 19:59:35 -0800 (PST)
Received: by mail-wg0-f51.google.com with SMTP id n12so4010627wgh.30 for <hybi@ietf.org>; Tue, 18 Feb 2014 19:59:32 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc:content-type; bh=Pe9lPJnjWfj7yOwg+LOP5ipu0Zids8lgVT7959hAEjY=; b=U9cg3qNcudAXDspsIiAule0k0bYa/JZfqFYRyeL+7ACm0M7H+LYj5xxVrnB2ObOYpo EGKT/I/TrGXc7fU9G8yeayexwvRUQgbGIjmbC/6RWmX4I9RPWAslZCYlXiO5K140L/cW KN7Q6Vy4r5T91tntLrzCRKpXb6bdi7RD8RvgZ50RdIXA5CpmCyAGKB5xG5JogTYejyXq 9iK7gOFOUQnx1L1GjLvQpWxrmo/TpGh7cPYXUVbd97oYmCAbMgwVOlTHBcXTEY7ykwB4 eTzPBNVxMP5Q46EEPbbhIe90+ApvYPeCxiOFvh+2ezNVnYAftMBc7sLY9I3l7A7EIZsZ ot5w==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc:content-type; bh=Pe9lPJnjWfj7yOwg+LOP5ipu0Zids8lgVT7959hAEjY=; b=ghyuAuiZEimGBS1CzaPzZufLHI5zj3VW6Ud7ih25HFN5KlqeecLzDqvFXA7oT35DpY RxLgbAWIm/nEz7GLXJtEjbGwA5CXc2Vx1NxDJZhO9rDUS74wQ27uvX4dMd4LsheP5VkE yaUFPzOfN0t3HYdufsL5MPbro1unxa1zO+Fr8WuDLooJCYvTd1XaKjjdqLcL8AvITDft WKaTzIT4xKJzk71wEnK4uLRi9c/ZDmbECiRdqlTBD5x0lc3zKuIpuEpNHPG9Jhg48jJs P/WCKhyxyJXZ6CZOnTPEMI6p8Y8vvO8Q7DWOtlzZMJYgNsW93T03b5A8TtfagRM2+XQm HGjA==
X-Gm-Message-State: ALoCoQkDHuKQBvanQAJyy/uzi3z8x4KgBeJFzPE5NeGht3o89vnYiZi6iUPLDM7E3EaK4krBD2JT2st7wPSVPsrFgHmse5eHiT1gR8/U99a3kl6lOn56UgvJ5vFgy+qjtvOc5nV6UykR6Z5yDOK7KAgigc+Gs4MWA0nfGSa08uNjwAPtAbJ3t8viLpwrjuln9jqRoGvZJZVJ
X-Received: by 10.180.107.136 with SMTP id hc8mr21228003wib.11.1392782372006;  Tue, 18 Feb 2014 19:59:32 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Tue, 18 Feb 2014 19:59:10 -0800 (PST)
In-Reply-To: <CAG4zZZB3h6TTFLUp-ucYQP7tMf-8=h5fh1UqufMh3w-0JN+hSw@mail.gmail.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <CACuKZqEcA1Pv8RpWfmThMjTzi2BbVMMKXqujs6BxVfxRPZJ9NQ@mail.gmail.com> <CAG4zZZCr4aTfVpw2coX2g0qw++4kdgNCFVze6tHKZ+fJNqb0aQ@mail.gmail.com> <CAG4zZZA4WR50Ea2e36XEgTgpJURwiMFn+DXeYuxHFfZatYzDOQ@mail.gmail.com> <CAH9hSJYv5VXGjS7AfG12-ArEvE6Uj_eE6pPxAiQcchcgV8vHcg@mail.gmail.com> <CAH9hSJbzra7uz7yfKQwfZaP_jhnxwdZyx8JnwCmBGhiMk6rbtg@mail.gmail.com> <CAG4zZZC1cmaH36znAvVLZE_MyJ+ThNk1Ky7tQ-QnKf_qwiQNDw@mail.gmail.com> <CAG4zZZB3h6TTFLUp-ucYQP7tMf-8=h5fh1UqufMh3w-0JN+hSw@mail.gmail.com>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Wed, 19 Feb 2014 12:59:10 +0900
Message-ID: <CAH9hSJaeCrLjkhHUzaaGDw-apiSv-eaPZeYGEHBuwRoF3yPCOA@mail.gmail.com>
To: Joakim Erdfelt <joakim@intalio.com>
Content-Type: multipart/alternative; boundary=e89a8f3bad4588840e04f2ba6aa7
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/ESJb1B_7tDmqsl-HqWv18phK31A
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 19 Feb 2014 03:59:48 -0000

--e89a8f3bad4588840e04f2ba6aa7
Content-Type: text/plain; charset=UTF-8

Thanks, Joakim. So you're still be able to do the fail-fast without
WebSocket frame length header if we adopt HEADERS+DATA+END_SEGMENT approach
but with WebSocket frame length header eliminated. HTTP/2.0 DATA frame has
its own payload length field though it's small compared to the range
WebSocket header can express.

Takeshi


On Wed, Feb 19, 2014 at 3:12 AM, Joakim Erdfelt <joakim@intalio.com> wrote:

> Clicked too soon on that pseudo code.
> This is more clear. (i hope)
>
>    header = frame.parse_header(network_bytes)
>    if (header.size > system_memory_constraints) then
>       fail_protocol(1009)
>       // do not read payload
>       return
>    end
>    if (message.size + header.size > message.max_constraints) then
>       fail_protocol(1009)
>       // do not read payload
>       return
>    else
>       payload = frame.read_payload(network_bytes)
>       message.append(payload)
>    end
>    if (header.fin == true) then
>       message.deliver_to(local_endpoint)
>       // free up any message resources, reset message size to 0
>       message.reset
>    end
>
>
> --
> Joakim Erdfelt <joakim@intalio.com>
> webtide.com <http://www.webtide.com/> - intalio.com/jetty
> Expert advice, services and support from from the Jetty & CometD experts
> eclipse.org/jetty - cometd.org
>
>
> On Tue, Feb 18, 2014 at 11:08 AM, Joakim Erdfelt <joakim@intalio.com>wrote:
>
>> I'm not sure I was clear, as you seem to still have questions and are
>> hinting at alternatives.
>>
>> The most important thing I said:
>>    The size is important at the start of any payload about to be received.
>>
>> All other statements are based off of that.
>> The nuanced questions you are asking seems to indicate that you treat a
>> frame differently than a message is irrelevant.
>> If the frame happens to represent a complete message (eg:
>> opcode=(text|binary) + fin=true) vs being only 1 frame belonging to a
>> series of frames that make up a whole message makes no difference to the
>> statement.
>>
>> Since websocket is message based, not stream based, the above core
>> statement remains true.
>>
>> The fail-fast example applies for fragmented or whole messages.
>>
>>  some pseudo code:
>>
>>    header = frame.parse_header(network_bytes)
>>    if (header.size > system_memory_constraints) then
>>       fail_protocol(1009)
>>       return
>>    end
>>    // is message still being collected and not yet delivered to local
>> endpoint?
>>    if (message.delivered == false) then
>>       if (message.size + header.size > message.max_constraints) then
>>         fail_protocol(1009)
>>         return
>>       else
>>         payload = frame.read_payload(network_bytes)
>>         message.append(payload)
>>       end
>>    end
>>    if (header.fin == true) then
>>       message.deliver_to(local_endpoint)
>>       message.reset
>>    end
>>
>>
>>  --
>> Joakim Erdfelt <joakim@intalio.com>
>> webtide.com <http://www.webtide.com/> - intalio.com/jetty
>> Expert advice, services and support from from the Jetty & CometD experts
>> eclipse.org/jetty - cometd.org
>>
>>
>> On Mon, Feb 17, 2014 at 10:37 PM, Takeshi Yoshino <tyoshino@google.com>wrote:
>>
>>> So, to be clear, Joakim, you're not objecting to make it unable to know
>>> the size of each frame, but objecting to make it completely unable to know
>>> the size of big messages on receiving the head of it, right?
>>>
>>> Your use case (fail-fast) seems to require the sender to send messages
>>> unfragmented. So, it sounds like application (or kind of application)
>>> specific needs. You could do some application specific hack such as
>>> introducing "complete-message-size" header in the payload than depending on
>>> protocol level length header. If not, clients in the wild which are out of
>>> your control may do arbitrary fragmentation. FYI, Chromium's WebSocket
>>> stack is being refactored and after that it starts sending big messages as
>>> fragmented into small frames.
>>>
>>
>>
>

--e89a8f3bad4588840e04f2ba6aa7
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Thanks, Joakim. So you&#39;re still be able to do the fail=
-fast without WebSocket frame length header if we adopt HEADERS+DATA+END_SE=
GMENT approach but with WebSocket frame length header=C2=A0eliminated. HTTP=
/2.0 DATA frame has its own payload length field though it&#39;s small comp=
ared to the range WebSocket header can=C2=A0express.</div>

<div class=3D"gmail_extra"><br clear=3D"all"><div>Takeshi</div>
<br><br><div class=3D"gmail_quote">On Wed, Feb 19, 2014 at 3:12 AM, Joakim =
Erdfelt <span dir=3D"ltr">&lt;<a href=3D"mailto:joakim@intalio.com" target=
=3D"_blank">joakim@intalio.com</a>&gt;</span> wrote:<br><blockquote class=
=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padd=
ing-left:1ex">

<div dir=3D"ltr">Clicked too soon on that pseudo code.<div>This is more cle=
ar. (i hope)</div><div><br></div><div><div class=3D""><div>=C2=A0 =C2=A0hea=
der =3D frame.parse_header(network_bytes)</div><div>=C2=A0 =C2=A0if (header=
.size &gt; system_memory_constraints) then</div>


<div>=C2=A0 =C2=A0 =C2=A0 fail_protocol(1009)</div></div><div>=C2=A0 =C2=A0=
 =C2=A0 // do not read payload</div><div>=C2=A0 =C2=A0 =C2=A0 return</div><=
div>=C2=A0 =C2=A0end</div><div class=3D""><div>=C2=A0 =C2=A0if (message.siz=
e + header.size &gt; message.max_constraints) then</div><div>=C2=A0 =C2=A0 =
=C2=A0 fail_protocol(1009)</div>


</div><div>=C2=A0 =C2=A0 =C2=A0 // do not read payload</div><div class=3D""=
><div>=C2=A0 =C2=A0 =C2=A0 return</div><div>=C2=A0 =C2=A0else</div><div>=C2=
=A0 =C2=A0 =C2=A0 payload =3D frame.read_payload(network_bytes)</div><div>=
=C2=A0 =C2=A0 =C2=A0 message.append(payload)</div><div>=C2=A0 =C2=A0end</di=
v></div>

<div class=3D""><div>=C2=A0 =C2=A0if (header.fin =3D=3D true) then</div>
<div>=C2=A0 =C2=A0 =C2=A0 message.deliver_to(local_endpoint)</div></div><di=
v>=C2=A0 =C2=A0 =C2=A0 // free up any message resources, reset message size=
 to 0</div><div>=C2=A0 =C2=A0 =C2=A0 message.reset</div><div>=C2=A0 =C2=A0e=
nd</div></div><div><br></div></div><div class=3D"gmail_extra">

<div class=3D"">
<br clear=3D"all"><div><div dir=3D"ltr"><div>--</div><div>Joakim Erdfelt &l=
t;<a href=3D"mailto:joakim@intalio.com" target=3D"_blank">joakim@intalio.co=
m</a>&gt;</div><div><a href=3D"http://www.webtide.com/" target=3D"_blank">w=
ebtide.com</a>=C2=A0- <a href=3D"http://intalio.com/jetty" target=3D"_blank=
">intalio.com/jetty</a></div>


<div><span>Expert advice, services and support from=C2=A0</span><span>from =
the Jetty &amp; CometD experts</span></div><div><a href=3D"http://eclipse.o=
rg/jetty/" target=3D"_blank">eclipse.org/jetty</a>=C2=A0-=C2=A0<a href=3D"h=
ttp://cometd.org/" target=3D"_blank">cometd.org</a></div>


</div></div>
<br><br></div><div><div class=3D"h5"><div class=3D"gmail_quote">On Tue, Feb=
 18, 2014 at 11:08 AM, Joakim Erdfelt <span dir=3D"ltr">&lt;<a href=3D"mail=
to:joakim@intalio.com" target=3D"_blank">joakim@intalio.com</a>&gt;</span> =
wrote:<br>

<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
<div dir=3D"ltr"><div>I&#39;m not sure I was clear, as you seem to still ha=
ve questions and are hinting at alternatives.</div><div><br></div><div>The =
most important thing I said:</div>=C2=A0 =C2=A0The size is important at the=
 start of any payload about to be received.<div>



<br></div><div>All other statements are based off of that.</div><div>The nu=
anced questions you are asking seems to indicate that you treat a frame dif=
ferently than a message is irrelevant.</div><div>If the frame happens to re=
present a complete message (eg: opcode=3D(text|binary) + fin=3Dtrue) vs bei=
ng only 1 frame belonging to a series of frames that make up a whole messag=
e makes no difference to the statement.</div>



<div><br></div><div>Since websocket is message based, not stream based, the=
 above core statement remains true.</div><div><br></div><div>The fail-fast =
example applies for fragmented or whole messages.</div><div><br></div>


<div>
some pseudo code:</div><div><br></div><div>=C2=A0 =C2=A0header =3D frame.pa=
rse_header(network_bytes)</div><div>=C2=A0 =C2=A0if (header.size &gt; syste=
m_memory_constraints) then</div><div>=C2=A0 =C2=A0 =C2=A0 fail_protocol(100=
9)</div><div>=C2=A0 =C2=A0 =C2=A0 return</div><div>



=C2=A0 =C2=A0end</div><div>=C2=A0 =C2=A0// is message still being collected=
 and not yet delivered to local endpoint?</div><div>=C2=A0 =C2=A0if (messag=
e.delivered =3D=3D false) then</div><div>=C2=A0 =C2=A0 =C2=A0 if (message.s=
ize + header.size &gt; message.max_constraints) then</div>



<div>=C2=A0 =C2=A0 =C2=A0 =C2=A0 fail_protocol(1009)</div><div>=C2=A0 =C2=
=A0 =C2=A0 =C2=A0 return</div><div>=C2=A0 =C2=A0 =C2=A0 else</div><div>=C2=
=A0 =C2=A0 =C2=A0 =C2=A0 payload =3D frame.read_payload(network_bytes)</div=
><div>=C2=A0 =C2=A0 =C2=A0 =C2=A0 message.append(payload)</div><div>=C2=A0 =
=C2=A0 =C2=A0 end</div><div>=C2=A0 =C2=A0end</div>



<div>=C2=A0 =C2=A0if (header.fin =3D=3D true) then</div><div>=C2=A0 =C2=A0 =
=C2=A0 message.deliver_to(local_endpoint)</div><div>=C2=A0 =C2=A0 =C2=A0 me=
ssage.reset</div><div>=C2=A0 =C2=A0end</div><div><br></div></div><div class=
=3D"gmail_extra"><div><br clear=3D"all"><div>
<div dir=3D"ltr">
<div>--</div><div>Joakim Erdfelt &lt;<a href=3D"mailto:joakim@intalio.com" =
target=3D"_blank">joakim@intalio.com</a>&gt;</div><div><a href=3D"http://ww=
w.webtide.com/" target=3D"_blank">webtide.com</a>=C2=A0- <a href=3D"http://=
intalio.com/jetty" target=3D"_blank">intalio.com/jetty</a></div>



<div><span>Expert advice, services and support from=C2=A0</span><span>from =
the Jetty &amp; CometD experts</span></div><div><a href=3D"http://eclipse.o=
rg/jetty/" target=3D"_blank">eclipse.org/jetty</a>=C2=A0-=C2=A0<a href=3D"h=
ttp://cometd.org/" target=3D"_blank">cometd.org</a></div>



</div></div>
<br><br></div><div><div class=3D"gmail_quote">On Mon, Feb 17, 2014 at 10:37=
 PM, Takeshi Yoshino <span dir=3D"ltr">&lt;<a href=3D"mailto:tyoshino@googl=
e.com" target=3D"_blank">tyoshino@google.com</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
<div dir=3D"ltr"><div class=3D"gmail_extra">So, to be clear, Joakim, you&#3=
9;re not objecting to make it unable to know the size of each frame, but ob=
jecting to make it completely unable to know the size of big messages on re=
ceiving the head of it, right?</div>





<div class=3D"gmail_extra"><br></div><div class=3D"gmail_extra">Your use ca=
se (fail-fast) seems to require the sender to send messages unfragmented. S=
o, it sounds like application (or kind of application) specific needs. You =
could do some application specific hack such as introducing &quot;complete-=
message-size&quot; header in the payload than depending on protocol level l=
ength header. If not, clients in the wild which are out of your control may=
 do arbitrary fragmentation. FYI, Chromium&#39;s WebSocket stack is being r=
efactored and after that it starts sending big messages as fragmented into =
small frames.</div>





</div>
</blockquote></div><br></div></div>
</blockquote></div><br></div></div></div>
</blockquote></div><br></div>

--e89a8f3bad4588840e04f2ba6aa7--


From nobody Tue Feb 18 20:09:12 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 8AE361A0551 for <hybi@ietfa.amsl.com>; Tue, 18 Feb 2014 20:09:10 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.926
X-Spam-Level: 
X-Spam-Status: No, score=-1.926 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 4Xi2V2kGcJwT for <hybi@ietfa.amsl.com>; Tue, 18 Feb 2014 20:09:08 -0800 (PST)
Received: from mail-wg0-x235.google.com (mail-wg0-x235.google.com [IPv6:2a00:1450:400c:c00::235]) by ietfa.amsl.com (Postfix) with ESMTP id 224341A0309 for <hybi@ietf.org>; Tue, 18 Feb 2014 20:09:07 -0800 (PST)
Received: by mail-wg0-f53.google.com with SMTP id x12so3907229wgg.8 for <hybi@ietf.org>; Tue, 18 Feb 2014 20:09:04 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:from:date:message-id:subject:to:cc:content-type; bh=AKeivn1zdSFEQsWFo5sdB/beUw6KsujNvayCeQwTm84=; b=C47SmfXhe7HMAoHIjjQ0C1NOObbIF1IJ55gtg/V0KLd+MXX2aEz93NIvw4sEjYgy6v iGL8KTlm7+ArJao7YVbsOU5x5rIKib6mSNOsOqh8b27CgxwfgbikshI+2mGHN4kq9Xof ET6qDSvoqtxDuRPjBhmHHZGFa8WcJX8T5yGl7/rOJR9t7fSa23s1dN3pHJ9UdVdbg1KR 7ULkEi7mYtQvioMJ3tob5tzsLViMubDgdKY9lxLpZmdcZb49aFmw6K6n1omut02E53h/ mSIB4WVZFwviGSKdjpDA4yoiSDGfXgDADcx56OIt4LPYbgGAfh53u3+rBnWflO/6Nq2e +JNQ==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:from:date:message-id:subject:to:cc :content-type; bh=AKeivn1zdSFEQsWFo5sdB/beUw6KsujNvayCeQwTm84=; b=kIl4wW1xBdzC8UHDY0gLLcvoVI+kK0rOaAlcPbz4GArRQggClov7KuDrsm+Pu5Q76S PpUhm+eL5T2eRZCnf5Idbi6XT9UObonf8uNhzJEtttYCtXo/08wlEKPsOEMlYoFOBR4G awzYXy8ai7tmmI9VdLYVdslkFU5whYqk0t3lo8MxwMp/hu5yl7itWWZ5WtHntc3Oh+34 YK7r/CWkDLbT99z79Kv01B5UEBEc6T22sBQLd7uxBQKFqc8jlGZffqRbon0d8D8IgeN/ MRgR8lKwmvNC7RcyuoxeHjq6hJx6mJzbiC+eF6Z25JcZ0wZwS8fh4DlG0cKrii1YeU+N MoEw==
X-Gm-Message-State: ALoCoQkDuabvAIiDEi0m6sfrfceAaHmBX6X3nc1iMP9pxWqXuLbbtq0u7CwzK2MUXBxtZPPBySsGHDIuJRSjIqnPpXuEoh5Papd+Jo+2xxYQlye4puAL+3y7a2IYp6SsKMGyxIpxw77P0VvpnWZzWWmR+mMlDW6DfRW50B2MvjqssVuoFKkNmmjui4gk1kWuZsfPfPxY82Fe
X-Received: by 10.180.107.136 with SMTP id hc8mr21256075wib.11.1392782944511;  Tue, 18 Feb 2014 20:09:04 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Tue, 18 Feb 2014 20:08:44 -0800 (PST)
From: Takeshi Yoshino <tyoshino@google.com>
Date: Wed, 19 Feb 2014 13:08:44 +0900
Message-ID: <CAH9hSJa0XvzmS1MmmPmqKjQs0P8EUP8n=x+b0im1S2CO5wL1NA@mail.gmail.com>
To: Yutaka Hirano <yhirano@chromium.org>, Tobias Oberstein <tobias.oberstein@tavendo.de>
Content-Type: multipart/alternative; boundary=e89a8f3bad45a83def04f2ba8cef
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/Hqqye7xYHqmqvykPbuqEcTBKFBg
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: [hybi] Straw poll: Do you miss per-frame extension? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 19 Feb 2014 04:09:10 -0000

--e89a8f3bad45a83def04f2ba8cef
Content-Type: text/plain; charset=UTF-8

It seems we're using "RFC 6455 Semantics" with a bit different nuance. I'd
like to divide the discussion into more primitive questions.

Now, I want to ask you, "do you miss per-frame extensions?".

More generally, "do you want there to be boundaries inside a message that
will never be modified by multiplexing mechanism?"

Please respond to this thread with your answer and rationale.

FYI, the expired WebSocket dedicated mux spec
draft-ietf-hybi-websocket-multiplexing-11 DOES MODIFY frame boundaries.

--e89a8f3bad45a83def04f2ba8cef
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">It seems we&#39;re using &quot;RFC 6455 Semantics&quot; wi=
th a bit different nuance. I&#39;d like to divide the discussion into more =
primitive questions.<div><br></div><div>Now, I want to ask you, &quot;do yo=
u miss per-frame extensions?&quot;.</div>

<div><br></div><div>More generally, &quot;do you want there to be boundarie=
s inside a message that will never be modified by multiplexing mechanism?&q=
uot;</div><div><br></div><div>Please respond to this thread with your answe=
r and rationale.</div>

<div><br></div><div>FYI, the expired WebSocket dedicated mux spec draft-iet=
f-hybi-websocket-multiplexing-11 DOES MODIFY frame boundaries.</div></div>

--e89a8f3bad45a83def04f2ba8cef--


From nobody Tue Feb 18 20:11:56 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 71B521A02FD for <hybi@ietfa.amsl.com>; Tue, 18 Feb 2014 20:11:54 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.926
X-Spam-Level: 
X-Spam-Status: No, score=-1.926 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id KUB65eQDDEOo for <hybi@ietfa.amsl.com>; Tue, 18 Feb 2014 20:11:53 -0800 (PST)
Received: from mail-we0-x235.google.com (mail-we0-x235.google.com [IPv6:2a00:1450:400c:c03::235]) by ietfa.amsl.com (Postfix) with ESMTP id D38351A0309 for <hybi@ietf.org>; Tue, 18 Feb 2014 20:11:52 -0800 (PST)
Received: by mail-we0-f181.google.com with SMTP id w61so11904598wes.26 for <hybi@ietf.org>; Tue, 18 Feb 2014 20:11:49 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc:content-type; bh=yzT1VpLdFoJxWkOprGqPDiVhgZDghOtM69Jlze/DHjE=; b=Zd3fwWEoSyqqOYSuDOJ/N1eh1bv5cF1IG7ZSbm/sxfJspkDSs1VUlSZBRj/FZ0zG0R kfA7Xfsto1Aub1vQTC5vSJOt4uSCDic7bgvaP2BXInta3C3dohV6lFkQoMN8dfQ/Gu8+ nWsQs81F6lq3AckS+9np4mkCONPJ7O4kuGlTQsAu63cZxSMYLMgF++VfN4wFWRcJdrQQ 57p73tOmxV/t7fP/W2GutmPDKuOx+aT7a0Om+JyUiVz5z66A9A0sd1ohO1ku0jml2Y6L GlKcg6/xVxgdupXUF81slEIbhf3hmXeuGIDv4cACPGzfJJxyH1UUfCv9mkK21o8v9UxI MNdQ==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc:content-type; bh=yzT1VpLdFoJxWkOprGqPDiVhgZDghOtM69Jlze/DHjE=; b=Ir8e7fA/+tIMMLqiuajdRQx2aaCQFlAkI7TcgRrXoXS8AaHoHVWZdbAxtiDVtIAyRh +Iveu55BXb9Q/JwzkwRmaMRcbG3Utikc7iXHiVWNcY1qK2sjdVxpLaT8mw6KeQXJK4Bt IcROThk7ZwKJUH+V1jicLMJ5bC+mwgdd7lRkF6Sgo3ukn0mDVBkrcIMYcAYF3BObIgei 5t6bd6FMeLkiO/+yv5S31991fwP/b0ijo6QuralDn5kZBQ81CDmC1PeBeSfxyavMk5fm NH+gzthXplkLXg2bEtGL061lwhDV3LIDo+jXkV95XCCPrBVjU8xmlT/nHv4mAAa3o69t Hpqg==
X-Gm-Message-State: ALoCoQl5G24bj4oKzdKtdn6MH6E1sa0KeUqCMNo5BY8MZQCokby5LNX4hr3K0C24FOesMBP8AV901irvB0qWU+quLTw5j3IFehbkxfJ85xdnVkti4+rxTfGksXZ37aUupTObnKG4+DwSq23WnnZhGIjsRzj13FF7eSgPNHSU+rzpEYLzcANaPfsRjMjYGbrF0QwGO545spsn
X-Received: by 10.180.37.193 with SMTP id a1mr21093102wik.52.1392783109319; Tue, 18 Feb 2014 20:11:49 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Tue, 18 Feb 2014 20:11:29 -0800 (PST)
In-Reply-To: <CAH9hSJa0XvzmS1MmmPmqKjQs0P8EUP8n=x+b0im1S2CO5wL1NA@mail.gmail.com>
References: <CAH9hSJa0XvzmS1MmmPmqKjQs0P8EUP8n=x+b0im1S2CO5wL1NA@mail.gmail.com>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Wed, 19 Feb 2014 13:11:29 +0900
Message-ID: <CAH9hSJbQ9kWjtYCd6UM9hAGA=gr0oE7M5BVPtwJq1K8-95hAkQ@mail.gmail.com>
To: Yutaka Hirano <yhirano@chromium.org>, Tobias Oberstein <tobias.oberstein@tavendo.de>
Content-Type: multipart/alternative; boundary=e89a8f502fa07b09ec04f2ba9699
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/FLIwpJ1sV9eOYZKGS7jpVuGtj_E
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss per-frame extension? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 19 Feb 2014 04:11:54 -0000

--e89a8f502fa07b09ec04f2ba9699
Content-Type: text/plain; charset=UTF-8

My answer is "no, I don't". I think we should drop per-frame extension
capability. Even RFC 6455 should be updated to drop it when we get chance
to do so.

--e89a8f502fa07b09ec04f2ba9699
Content-Type: text/html; charset=UTF-8

<div dir="ltr"><div class="gmail_extra">My answer is &quot;no, I don&#39;t&quot;. I think we should drop per-frame extension capability. Even RFC 6455 should be updated to drop it when we get chance to do so.</div></div>

--e89a8f502fa07b09ec04f2ba9699--


From nobody Tue Feb 18 20:19:47 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 19B9E1A032C for <hybi@ietfa.amsl.com>; Tue, 18 Feb 2014 20:19:45 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.926
X-Spam-Level: 
X-Spam-Status: No, score=-1.926 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id X5p1wwqEAKDO for <hybi@ietfa.amsl.com>; Tue, 18 Feb 2014 20:19:42 -0800 (PST)
Received: from mail-wi0-x22f.google.com (mail-wi0-x22f.google.com [IPv6:2a00:1450:400c:c05::22f]) by ietfa.amsl.com (Postfix) with ESMTP id 6C4371A0325 for <hybi@ietf.org>; Tue, 18 Feb 2014 20:19:42 -0800 (PST)
Received: by mail-wi0-f175.google.com with SMTP id hm4so4233578wib.8 for <hybi@ietf.org>; Tue, 18 Feb 2014 20:19:38 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:from:date:message-id:subject:to:content-type; bh=+rIMY+BpcazRF+EAPEXzZXUtXECHYMEe3rmyNJDqwLE=; b=cja5WpxjWECs+ON74UfMjCVqEZtaqv/JU+RZ4mhd7nPw8LmLs/Je8uAG1a/DwFAyAI UPB3LIMVfJ3grAVIO39/Tfa6QL/obS9W+b48pADj6OCbMeVPRhf4kz4FDy9JtmKXFGa2 xAtgGVii4LZLENnzTYMNF5qf9kM+N8Ue4mwvFmJWOmisSQGMPBnoOKnanxUsJRBxvaI2 WVtI98K9LZzytRvc2yPUkbHzuD83T96r9skubGOAd+9LmvY8Zwy3PudZY/i0JoIKQeNh 72AvYWrbMJZq9AvUp8T0GTqejMM6ZVHlUZwjjiwzGF3np6gqVGP7nuQJPLTYIT+UkqiC awIA==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:from:date:message-id:subject:to :content-type; bh=+rIMY+BpcazRF+EAPEXzZXUtXECHYMEe3rmyNJDqwLE=; b=QDTd89N8gebeCq9hfgZlOM2hxK1IBz+Hj0w3VStIZVoYCEko3nsjadhurVOCNKF60d Mmxved53AIzd1c5EcCGA2+PipFIA3Tp3YKkKT9dQ58rqBa0JAm3AcAWx7+RAS+oPoMS3 VmtrXYZ1tplSrSMKtKHLV/KGx1yEf3dLChkow39G8t+JGbYmARihCaAgCJN//mAh4DWb TQwfk9HGt25AjXwYNJxU8nQK4dltLXybZ9Os9pz58Wi5rimZ1DM0FfJo0aHKU2fc9W9J 1Xao3zvXwYkWvdGJROq8qKWA+QtztsTSFnfGAAoDvkjyyvNEGZRm68f+MNXjJhQqKOAF JuUw==
X-Gm-Message-State: ALoCoQnRYpD9/e6fLmQyABe33nXeG0yYw5hU120HlXZT4+S4OHp8AJL0Ahc2RbChpmsvzlPCgMMtyOz/WtKLKl1HCN0TbqHk4yA6E/5meM3M0AAXBjV5SdeTlYgcpGGa0/8oOx9FRL1ikAI3QJT72Gs0cmSiO8DHK3bHGZswWUx7KTT9Bp87TmXcLT6nf11OXnO3Po4hSBbV
X-Received: by 10.194.61.114 with SMTP id o18mr21451326wjr.6.1392783578867; Tue, 18 Feb 2014 20:19:38 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Tue, 18 Feb 2014 20:19:18 -0800 (PST)
From: Takeshi Yoshino <tyoshino@google.com>
Date: Wed, 19 Feb 2014 13:19:18 +0900
Message-ID: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com>
To: Yutaka Hirano <yhirano@google.com>, Tobias Oberstein <tobias.oberstein@tavendo.de>,  Peter Thorson <webmaster@zaphoyd.com>, Roberto Peon <fenix@google.com>,  "hybi@ietf.org" <hybi@ietf.org>
Content-Type: multipart/alternative; boundary=047d7bacc8a677c19804f2bab20a
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/A7HExBApOY-_Gndwi7nbyJ7lPrk
Subject: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 19 Feb 2014 04:19:45 -0000

--047d7bacc8a677c19804f2bab20a
Content-Type: text/plain; charset=UTF-8

It seems we're using "RFC 6455 Semantics" with a bit different nuance. I'd
like to divide the discussion into more primitive questions.

The second point I want to hear your opinion is: "do you miss interjectable
WebSocket level control frame?"

Since RFC 6455's control frames can be interjected in the middle of a data
message, to keep this as-is, we need to prevent FRAME boundary from being
modified by multiplexing mechanism. Possible solutions are:
- encapsulate WebSocket frame
- prepare some mechanism (HTTP/2.0 frame type) in HTTP/2.0 layer similar to
HTTP/2.0 PING frame that provides a box to which we can map WebSocket level
control (PING/PONG/CLOSE/and other reserved control opcodes)

Please respond to this thread with your answer and rationale.

--047d7bacc8a677c19804f2bab20a
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div>It seems we&#39;re using &quot;RFC 6455 Semantics&quo=
t; with a bit different nuance. I&#39;d like to divide the discussion into =
more primitive questions.</div><div><br></div><div>The second point I want =
to hear your opinion is: &quot;do you miss interjectable WebSocket level co=
ntrol frame?&quot;</div>

<div><br></div><div>Since RFC 6455&#39;s control frames can be interjected =
in the middle of a data message, to keep this as-is, we need to prevent FRA=
ME boundary from being modified by multiplexing mechanism. Possible solutio=
ns are:</div>

<div>- encapsulate WebSocket frame</div><div>- prepare some mechanism (HTTP=
/2.0 frame type) in HTTP/2.0 layer similar to HTTP/2.0 PING frame that prov=
ides a box to which we can map WebSocket level control (PING/PONG/CLOSE/and=
 other reserved control opcodes)</div>

<div><br></div><div>Please respond to this thread with your answer and rati=
onale.</div>
</div>

--047d7bacc8a677c19804f2bab20a--


From nobody Tue Feb 18 20:24:33 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id D2C261A030F for <hybi@ietfa.amsl.com>; Tue, 18 Feb 2014 20:24:31 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.926
X-Spam-Level: 
X-Spam-Status: No, score=-1.926 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 2h8cgrHagmF6 for <hybi@ietfa.amsl.com>; Tue, 18 Feb 2014 20:24:30 -0800 (PST)
Received: from mail-we0-x22d.google.com (mail-we0-x22d.google.com [IPv6:2a00:1450:400c:c03::22d]) by ietfa.amsl.com (Postfix) with ESMTP id E0E721A0129 for <hybi@ietf.org>; Tue, 18 Feb 2014 20:24:29 -0800 (PST)
Received: by mail-we0-f173.google.com with SMTP id x48so6742476wes.32 for <hybi@ietf.org>; Tue, 18 Feb 2014 20:24:26 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :content-type; bh=Uz+YZMbetcAwp1TyV9GsWhsW8z7nrop/POE7NwyHxSA=; b=SSQr8zSzlUSGiZwS8TY/hSOAewBd7Ffx/fUbgCuMoW+Rs37psxy/BFBxgr0/lE1eWb RWRERcWDAfLIKCmYghkRg4CIUznVjTC7P01v8HoyN2zn8A+OFSK0Qpish0YxKx7irSaE 5NPAg3op2qZjWCR7V0DeukgPotBYL5vwIyt7pqhDEAUoMUM1mLz+IdSyphpMd1MkBHdF qqHjCG0aKnJtajbo4gYWh12N2nKSHKXXICZrD8DikzHiSEUDlWeoUSvPym0KVtEyggLQ QvLFfqklwlKdTuiNSK86WZNdpfIEzfD40V1f/BIGeTxx/zUIEgcQq3NLSKA49uwJVXAg o/rA==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:content-type; bh=Uz+YZMbetcAwp1TyV9GsWhsW8z7nrop/POE7NwyHxSA=; b=FHBdjuoPkI2Dyo56BtO/DTBScziVN/PUtTospnNudh2I5x3rZYkc5/n+BmJ0J+y9Ub /pIyX0jBNx7ZDUc03SYYU/emTMmmQckWVwlf8D9gzS4udWKsoy8fcRSWft295IxJAixU 4BWKyMjR26lLxK+FujYXrWtp8BZoVkiG1JUezYIPUpRcbx5nNSlxpT0JFLc7NFfYZq8c c6HawxJkTJ5zL8klswaBnS/RW0L+dnE+EEgRysUQT00RoSuEhdRBu5wVfS4avhpknpym +Ll4oK2i/GEwlidvn/3Eel8zkxRC2gRGleeFLm8OkYTOR5aS0s0KjxweNWZjnHI7HVXs c9aQ==
X-Gm-Message-State: ALoCoQnKiOKGUjL88hipqkbqVbb6sXJ1zrSM0rtS2PJOxox1yUdZhoQG0h+FTmhiHglPX4z88LSxlO2f/er/SXl6+Eo64Fp8nnnuZ/CMeiAxyyolm37UWYGqOAB6AC1LfqWotF2doFWXjiEHxgzH10ne+VQM77gkxiDL+85LEY3DopCTxmikX+wDRBweRcHMMCZW4CT21w7I
X-Received: by 10.194.190.8 with SMTP id gm8mr26030962wjc.36.1392783866375; Tue, 18 Feb 2014 20:24:26 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Tue, 18 Feb 2014 20:24:06 -0800 (PST)
In-Reply-To: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Wed, 19 Feb 2014 13:24:06 +0900
Message-ID: <CAH9hSJbBmvNPBSSAk-khdWXgWw0GTt0FG3KsdzYeJcfiAPDk0A@mail.gmail.com>
To: Yutaka Hirano <yhirano@google.com>, Tobias Oberstein <tobias.oberstein@tavendo.de>,  Peter Thorson <webmaster@zaphoyd.com>, Roberto Peon <fenix@google.com>,  "hybi@ietf.org" <hybi@ietf.org>
Content-Type: multipart/alternative; boundary=047d7bea33f89ad62604f2bac379
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/eG_WcVizhlh0nQdSQlTUrV9fLdI
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 19 Feb 2014 04:24:32 -0000

--047d7bea33f89ad62604f2bac379
Content-Type: text/plain; charset=UTF-8

I'm inclined to say "Yes" for this question so said yes for #2 of Yutaka's
original question [1], but I could also cope with if we decide to drop this.

[1] http://www.ietf.org/mail-archive/web/hybi/current/msg10403.html

--047d7bea33f89ad62604f2bac379
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_extra">I&#39;m inclined to say &quot;Y=
es&quot; for this question so said yes for #2 of Yutaka&#39;s original ques=
tion [1], but I could also cope with if we decide to drop this.</div><div c=
lass=3D"gmail_extra">

<br></div><div class=3D"gmail_extra">[1]=C2=A0<a href=3D"http://www.ietf.or=
g/mail-archive/web/hybi/current/msg10403.html">http://www.ietf.org/mail-arc=
hive/web/hybi/current/msg10403.html</a></div></div>

--047d7bea33f89ad62604f2bac379--


From nobody Tue Feb 18 22:46:45 2014
Return-Path: <ricea@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id D8AE41A033F for <hybi@ietfa.amsl.com>; Tue, 18 Feb 2014 22:46:42 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.926
X-Spam-Level: 
X-Spam-Status: No, score=-1.926 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 6V3RABvAD5KP for <hybi@ietfa.amsl.com>; Tue, 18 Feb 2014 22:46:40 -0800 (PST)
Received: from mail-oa0-x236.google.com (mail-oa0-x236.google.com [IPv6:2607:f8b0:4003:c02::236]) by ietfa.amsl.com (Postfix) with ESMTP id 5310F1A030F for <hybi@ietf.org>; Tue, 18 Feb 2014 22:46:40 -0800 (PST)
Received: by mail-oa0-f54.google.com with SMTP id j17so354851oag.13 for <hybi@ietf.org>; Tue, 18 Feb 2014 22:46:37 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=KiQUnJaanrXmS/lm7Go8z9z4A4l6QVxvJpyv9vPDWME=; b=WRrLZIov++AKBGaO898Wd/cokilIAwBhG4ARQs9MEtQciy2ilvQ0UhHbnMwMvjOEas yzBdROfLK8Z+9ltAtiRTqFhcDmfmTd0deO6bb9gW+Nnsep8IszIwlmJoaFgpjNsTUOCY NwKfaAx1tkBbNWQ2MGxJgYpBTPfMK5JG0YxALS7Gg/tHyB9Svd/itgWTeFlqbfI7LJ9Y Fem4yDZP4+1+ASPti3/Iesoeu0n7U1mCSbehGe9dGmw8OSRQqPGlpaLMarvi8dvj8+pv 2bP3JimRA7EJlkaOOIdTVKsmPQuwQXhfsR3vS1BLW8pnWqPCue9wuTX4KfxoCjeWywCZ 2GxA==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=KiQUnJaanrXmS/lm7Go8z9z4A4l6QVxvJpyv9vPDWME=; b=iR4Bbg9Mv7OjTRIcLzxCgI3U4lMq4NNbljXBD3+d2/qb6dQFfzPUAhq5+QeHrHGrjY /E3FLjCuBVbHGTEw6zDOc1Zc+5jAeexEHQm3nr/t9r8gHfJTyZXrCNy1XpKToKfx3O/A VyW4zJAeWgfxqcFdNKSVwwGNkiaSbIYUPAMrLUOuHRAVdpgMiKNUSJRiBDCcQ3lLiKqt QDDCI1bn/xftUqq/4My98p588wGm5asfnHcWU2Y9QEJ23YIpckD03u9hWllZ0kL+Fg2w 09Iw9tz2Tlv7U8Mnl+cK/JxIx5JiENrM5drYxwKZS39MMlOqs972NH5GW4iE0148dTsj muVw==
X-Gm-Message-State: ALoCoQlhfUdKiGXqkkkXrkfhA5XdR2uicuhCiichFlZ2YT9CTq8L2gK/9LdSrJTBdcbhXQYvN2dZmOU+g4kaEHBvjhFo7rWB7N2XtOtWXkyJMi3OvyCxsZ0arzAKExFQXXqc7gOutAE424b/Iqd+XkMr725Vi+5a10jn3AVbvUssueqgTuTUvF8Ai605YW7W2iqEeZNO7RGX
MIME-Version: 1.0
X-Received: by 10.60.228.135 with SMTP id si7mr29461955oec.4.1392792397095; Tue, 18 Feb 2014 22:46:37 -0800 (PST)
Received: by 10.182.118.41 with HTTP; Tue, 18 Feb 2014 22:46:37 -0800 (PST)
In-Reply-To: <CAH9hSJbQ9kWjtYCd6UM9hAGA=gr0oE7M5BVPtwJq1K8-95hAkQ@mail.gmail.com>
References: <CAH9hSJa0XvzmS1MmmPmqKjQs0P8EUP8n=x+b0im1S2CO5wL1NA@mail.gmail.com> <CAH9hSJbQ9kWjtYCd6UM9hAGA=gr0oE7M5BVPtwJq1K8-95hAkQ@mail.gmail.com>
Date: Wed, 19 Feb 2014 15:46:37 +0900
Message-ID: <CAHixhFqUcmfs3jiJBv3Wvmm8VhHgLsutSvoDOQ=05PUaQ=m+ng@mail.gmail.com>
From: Adam Rice <ricea@google.com>
To: Takeshi Yoshino <tyoshino@google.com>
Content-Type: multipart/alternative; boundary=001a113639101342e604f2bcc030
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/lC3-e05BSVle8ul1jnMU4irmVJU
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss per-frame extension? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 19 Feb 2014 06:46:43 -0000

--001a113639101342e604f2bcc030
Content-Type: text/plain; charset=UTF-8

My answer is also "no, I would not miss per-frame extensions." Adding and
removing frame boundaries is useful when implementing intermediaries, and
needing a separate code path which preserved frame boundaries would be a
nuisance.

I also support the removal of this capability in the successor to RFC6455.


On 19 February 2014 13:11, Takeshi Yoshino <tyoshino@google.com> wrote:

> My answer is "no, I don't". I think we should drop per-frame extension
> capability. Even RFC 6455 should be updated to drop it when we get chance
> to do so.
>
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi
>
>

--001a113639101342e604f2bcc030
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">My answer is also &quot;no, I would not miss per-frame ext=
ensions.&quot; Adding and removing frame boundaries is useful when implemen=
ting intermediaries, and needing a separate code path which preserved frame=
 boundaries would be a nuisance.<div>
<br></div><div>I also support the removal of this capability in the success=
or to RFC6455.</div></div><div class=3D"gmail_extra"><br><br><div class=3D"=
gmail_quote">On 19 February 2014 13:11, Takeshi Yoshino <span dir=3D"ltr">&=
lt;<a href=3D"mailto:tyoshino@google.com" target=3D"_blank">tyoshino@google=
.com</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div dir=3D"ltr"><div class=3D"gmail_extra">=
My answer is &quot;no, I don&#39;t&quot;. I think we should drop per-frame =
extension capability. Even RFC 6455 should be updated to drop it when we ge=
t chance to do so.</div>
</div>
<br>_______________________________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
<br></blockquote></div><br></div>

--001a113639101342e604f2bcc030--


From nobody Tue Feb 18 23:09:24 2014
Return-Path: <ricea@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 9921B1A055F for <hybi@ietfa.amsl.com>; Tue, 18 Feb 2014 23:09:22 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.926
X-Spam-Level: 
X-Spam-Status: No, score=-1.926 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id vdR1INkbPTYP for <hybi@ietfa.amsl.com>; Tue, 18 Feb 2014 23:09:20 -0800 (PST)
Received: from mail-oa0-x231.google.com (mail-oa0-x231.google.com [IPv6:2607:f8b0:4003:c02::231]) by ietfa.amsl.com (Postfix) with ESMTP id 3D0651A0363 for <hybi@ietf.org>; Tue, 18 Feb 2014 23:09:20 -0800 (PST)
Received: by mail-oa0-f49.google.com with SMTP id i7so20207475oag.8 for <hybi@ietf.org>; Tue, 18 Feb 2014 23:09:17 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=aVtvKSY1c6BwJQOU+Y3GalGiZCgooUpPz8UjReSj1d8=; b=WAnLQvSiZma5+7c8gSO2Pxep9TF6FD4Q3+z65efx2DSN9VFknNmfDJPwXGZFJ0QnOV J8E8NCRX9bLGvaX4sQkF9YvZxiyzMlruUs8/lsdjc1ouVXPx3BOxkZ3e12Ofwkrx6yhN 1On5gY0icm4rkZspNqFmAWtMd+efU2+hjueE+aZyi2izDDhvblYqySpfDKJP0uyrLv4t yK6hHpH2KFVvIsHt9P8BnwChlYn9IyCn6n374ggspjW17zapxIbtxuTh/fIN+C7UKk7Z faef75XTD5tZFOphJbLjwjLXHO0WQYdKtVnYq0y8kb3QaS9NS8n4eBsLcOrvOWpm0Sv6 dHGw==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=aVtvKSY1c6BwJQOU+Y3GalGiZCgooUpPz8UjReSj1d8=; b=lo6Jc1forFXIrPgVE61fcXLwQgXiuj3NKsqiAGDJClbskg8Zya7shJeP+vjHe9Gcab cDc7DfqGAqAsxwNubaM0B345XRsfAInLr6bfmCqJSAzYkwFkp2u8OIFxJKvmUw0+pLOd FJxtm83Z/H61ZsBxc0lk61BZNkcrwLaJuwaMxBLpI58oOwYO0HPBaBBZa3Tsqo8yWFtc ZNHgV32q2nKiZKT64Pi08x+JzlcA0U5UlXSCWb9lJpzS33LaXdjAt+tX1DOLcwJyKowJ U5YByPJUD+gZ/PVWEsoFZ/XUHTDjEiCHdj1sdfmFnCKvieBSCbKzM8eGbumFH2cnDgvG I0qQ==
X-Gm-Message-State: ALoCoQn8wahvHFe4w02BcsofYxQsqvlhHHVWHirH7RIOdwtX5ML6OchaHrN8Pq5tifF8P0vdlIYWWmoafjraKG9QjrBF4kgkpubVtCVp7mfG8KNv4elZSpTM8e2/G3YLaLFWQFvTH4HyuTyOkG70hHuUYBRKGsa8KHmputKZEAJBXQBFnsTbl88Bux/Cwl0scyE2SvKcX8ey
MIME-Version: 1.0
X-Received: by 10.60.228.135 with SMTP id si7mr29539168oec.4.1392793756887; Tue, 18 Feb 2014 23:09:16 -0800 (PST)
Received: by 10.182.118.41 with HTTP; Tue, 18 Feb 2014 23:09:16 -0800 (PST)
In-Reply-To: <CAH9hSJbBmvNPBSSAk-khdWXgWw0GTt0FG3KsdzYeJcfiAPDk0A@mail.gmail.com>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <CAH9hSJbBmvNPBSSAk-khdWXgWw0GTt0FG3KsdzYeJcfiAPDk0A@mail.gmail.com>
Date: Wed, 19 Feb 2014 16:09:16 +0900
Message-ID: <CAHixhFq=wfmYH8-ij_WtsQLN=NUTJwRQ=k8jCPepQDM8V8ZZYA@mail.gmail.com>
From: Adam Rice <ricea@google.com>
To: Takeshi Yoshino <tyoshino@google.com>
Content-Type: multipart/alternative; boundary=001a113639101ffa9704f2bd113e
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/NtmPtHPV3oi6Vf7o2_8d43ew3Qg
Cc: "hybi@ietf.org" <hybi@ietf.org>, Yutaka Hirano <yhirano@google.com>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 19 Feb 2014 07:09:22 -0000

--001a113639101ffa9704f2bd113e
Content-Type: text/plain; charset=UTF-8

With the current set of control frames, I would not miss the ability to
interject them in the middle of a data message.

Breaking them down one by one:

* Pong: useless in the middle of a data message (the peer already knows
that it is receiving data, and useless for RTT measurement because it has
an unknown amount of OS queueing and buffer bloat in front of it).
* Ping: almost always useless in the middle of a data message. It could be
useful if, for example, you had a middle box which closed "idle"
connections, and considered any connection "idle" if there was no TCP/IP
payload traffic originating from the client, and the server was in the
middle of sending a huge message. However, such a middle box would also
break HTTP downloads so seems unlikely to exist in practice. Even if it
existed, how would the server detect it to know that it needed to send a
Ping?
* Close: Can be used to perform a "clean" close in the middle of sending a
message. However, I cannot think of any applications that would consider a
situation in which a in-transit message was aborted to close the connection
to be a "clean close" scenario.


On 19 February 2014 13:24, Takeshi Yoshino <tyoshino@google.com> wrote:

> I'm inclined to say "Yes" for this question so said yes for #2 of Yutaka's
> original question [1], but I could also cope with if we decide to drop this.
>
> [1] http://www.ietf.org/mail-archive/web/hybi/current/msg10403.html
>
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi
>
>

--001a113639101ffa9704f2bd113e
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">With the current set of control frames, I would not miss t=
he ability to interject them in the middle of a data message.<div><br></div=
><div>Breaking them down one by one:</div><div><br></div><div>* Pong: usele=
ss in the middle of a data message (the peer already knows that it is recei=
ving data, and useless for RTT measurement because it has an unknown amount=
 of OS queueing and buffer bloat in front of it).</div>
<div>* Ping: almost always useless in the middle of a data message. It coul=
d be useful if, for example, you had a middle box which closed &quot;idle&q=
uot; connections, and considered any connection &quot;idle&quot; if there w=
as no TCP/IP payload traffic originating from the client, and the server wa=
s in the middle of sending a huge message. However, such a middle box would=
 also break HTTP downloads so seems unlikely to exist in practice. Even if =
it existed, how would the server detect it to know that it needed to send a=
 Ping?</div>
<div>* Close: Can be used to perform a &quot;clean&quot; close in the middl=
e of sending a message. However, I cannot think of any applications that wo=
uld consider a situation in which a in-transit message was aborted to close=
 the connection to be a &quot;clean close&quot; scenario.</div>
</div><div class=3D"gmail_extra"><br><br><div class=3D"gmail_quote">On 19 F=
ebruary 2014 13:24, Takeshi Yoshino <span dir=3D"ltr">&lt;<a href=3D"mailto=
:tyoshino@google.com" target=3D"_blank">tyoshino@google.com</a>&gt;</span> =
wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div dir=3D"ltr"><div class=3D"gmail_extra">=
I&#39;m inclined to say &quot;Yes&quot; for this question so said yes for #=
2 of Yutaka&#39;s original question [1], but I could also cope with if we d=
ecide to drop this.</div>
<div class=3D"gmail_extra">

<br></div><div class=3D"gmail_extra">[1]=C2=A0<a href=3D"http://www.ietf.or=
g/mail-archive/web/hybi/current/msg10403.html" target=3D"_blank">http://www=
.ietf.org/mail-archive/web/hybi/current/msg10403.html</a></div></div>
<br>_______________________________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
<br></blockquote></div><br></div>

--001a113639101ffa9704f2bd113e--


From nobody Wed Feb 19 02:42:13 2014
Return-Path: <yhirano@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 305571A046F for <hybi@ietfa.amsl.com>; Wed, 19 Feb 2014 02:42:11 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.926
X-Spam-Level: 
X-Spam-Status: No, score=-1.926 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id hRIw-r0TKZ7z for <hybi@ietfa.amsl.com>; Wed, 19 Feb 2014 02:42:09 -0800 (PST)
Received: from mail-yh0-x233.google.com (mail-yh0-x233.google.com [IPv6:2607:f8b0:4002:c01::233]) by ietfa.amsl.com (Postfix) with ESMTP id 484361A0466 for <hybi@ietf.org>; Wed, 19 Feb 2014 02:42:08 -0800 (PST)
Received: by mail-yh0-f51.google.com with SMTP id t59so164066yho.24 for <hybi@ietf.org>; Wed, 19 Feb 2014 02:42:04 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=iYU0117alcAKXCfzBfWWgf4+gqfHUjFAAk9Ucabcddg=; b=k+6gWipvlSZJbrokZKNuHJ/Jkg6jeo4FPLJYglt4t/0TGukdBl6qEPUK/as/2hshG6 lUHDiz6ZHEoxU9NOaCjtVR9zuYO3ROa881iFTzY9jVV/N5BH2iPb90eUDVdTzB7ueSjW ebhAUB4d/jwhW1B+D88odgHSCAWQd0i4zTC5oqauhIb9vmSQn6kidurGgeCWAwxOwpkx fbMA43zwKPYS77Z7FhUtcenDBWrfUTYXZAsqD55AQnm1NK/VQAU/IONbJQDU+VnbUl/w 4c9oa2TG0hq+XDxn5wRcFTMYLB139x86oSdCwDTek5GS/6g9zuf7wJ3qBy5/Y9LIS3Hj f2Fw==
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=chromium.org; s=google; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=iYU0117alcAKXCfzBfWWgf4+gqfHUjFAAk9Ucabcddg=; b=D4JQv6TdcQ/9O893+0UpeQzBfCFsJE+znG65SmrmclXlaek8exGqmVbZj4k93FKmRn zWPWULb5CgO70oXbKBYWsH0j3uNGbzmkUYfKdWXdi9NjO74EwKqqsDsng9L6jE7UTbvf YQf3cnFG5m37+n2EfKVaT5Xc/0/c1pXeou06I=
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:sender:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=iYU0117alcAKXCfzBfWWgf4+gqfHUjFAAk9Ucabcddg=; b=PfHN54Kt3EJinN0RirCQFtMefxIdI+f5/4mbNWMB0SGI5z/Cj/7JmkqWLVqJFBOA2F TO8ewHIs6cPtFqjDHQFFe5fg9YkqV8aobic8AJTYrJ2kh/WfXAgsqq8Wbo2J+COztA+p zw1uTMXhOEXPHpot2NHr6it7reVJcoiqbfaHuPSHUrVVp6Vrgf00pvG5Qqz0k8Q3/6zq oTxLpnmhqmscwroQrGYa34UAgWx4O8eFHyWXZJ1qUd2RmLE7ETmMqsv0msGscPGvCJ/w 1cUDKdzviS+p6yBrcfrozIkA8OHVzlAbm+/Own7Xf4QWPbxL+NAXsTMmG0psJBdtvThL HAfQ==
X-Gm-Message-State: ALoCoQloabiKWlkRzr7x3uj3dgi5iNC7RB6XHfC9biow25WjEiS3aCdAvDE26MxmwJbuMTG2A7EK++9vfRrZFinb3fl8Isf/bigJC5Or0ySleIAnXDvZeUzy9Ko/0rFyYUI0WV0K7rJOHB5YaSVi4yrrtdaN+8F1lEIGuWxFM04q79UGVsgLVa+Nn+m/5ujZ6j/Z3w6QbODu
MIME-Version: 1.0
X-Received: by 10.236.131.19 with SMTP id l19mr42323984yhi.0.1392806524794; Wed, 19 Feb 2014 02:42:04 -0800 (PST)
Sender: yhirano@google.com
Received: by 10.170.164.87 with HTTP; Wed, 19 Feb 2014 02:42:04 -0800 (PST)
In-Reply-To: <CAH9hSJbQ9kWjtYCd6UM9hAGA=gr0oE7M5BVPtwJq1K8-95hAkQ@mail.gmail.com>
References: <CAH9hSJa0XvzmS1MmmPmqKjQs0P8EUP8n=x+b0im1S2CO5wL1NA@mail.gmail.com> <CAH9hSJbQ9kWjtYCd6UM9hAGA=gr0oE7M5BVPtwJq1K8-95hAkQ@mail.gmail.com>
Date: Wed, 19 Feb 2014 19:42:04 +0900
X-Google-Sender-Auth: CzsKEkU28sFv5HUFEOW93H7cF8A
Message-ID: <CABihn6E4x+bV6c_564RgbvRFivihGyF_CyFH3e7nZ2D+8FkbXg@mail.gmail.com>
From: Yutaka Hirano <yhirano@chromium.org>
To: Takeshi Yoshino <tyoshino@google.com>
Content-Type: multipart/alternative; boundary=20cf3010e96d270c7f04f2c00a86
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/WovxBwV8aUrZW2Ctvutn3wYrowc
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss per-frame extension? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 19 Feb 2014 10:42:11 -0000

--20cf3010e96d270c7f04f2c00a86
Content-Type: text/plain; charset=UTF-8

>
> Now, I want to ask you, "do you miss per-frame extensions?".

No.
I totally agree with Takeshi.


On Wed, Feb 19, 2014 at 1:11 PM, Takeshi Yoshino <tyoshino@google.com>wrote:

> My answer is "no, I don't". I think we should drop per-frame extension
> capability. Even RFC 6455 should be updated to drop it when we get chance
> to do so.
>

--20cf3010e96d270c7f04f2c00a86
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px =
0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-l=
eft-style:solid;padding-left:1ex">Now, I want to ask you, &quot;do you miss=
 per-frame extensions?&quot;.</blockquote>
No.<div>I totally agree with Takeshi.</div></div><div class=3D"gmail_extra"=
><br><br><div class=3D"gmail_quote">On Wed, Feb 19, 2014 at 1:11 PM, Takesh=
i Yoshino <span dir=3D"ltr">&lt;<a href=3D"mailto:tyoshino@google.com" targ=
et=3D"_blank">tyoshino@google.com</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div dir=3D"ltr"><div class=3D"gmail_extra">=
My answer is &quot;no, I don&#39;t&quot;. I think we should drop per-frame =
extension capability. Even RFC 6455 should be updated to drop it when we ge=
t chance to do so.</div>
</div>
</blockquote></div><br></div>

--20cf3010e96d270c7f04f2c00a86--


From nobody Wed Feb 19 03:46:04 2014
Return-Path: <tobias.oberstein@tavendo.de>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 44AB51A0482 for <hybi@ietfa.amsl.com>; Wed, 19 Feb 2014 03:46:03 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.9
X-Spam-Level: 
X-Spam-Status: No, score=-1.9 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, RCVD_IN_DNSWL_NONE=-0.0001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id SKLnCpXHwedJ for <hybi@ietfa.amsl.com>; Wed, 19 Feb 2014 03:46:01 -0800 (PST)
Received: from EXHUB020-2.exch020.serverdata.net (exhub020-2.exch020.serverdata.net [206.225.164.29]) by ietfa.amsl.com (Postfix) with ESMTP id CF2C41A03D0 for <hybi@ietf.org>; Wed, 19 Feb 2014 03:46:00 -0800 (PST)
Received: from EXVMBX020-12.exch020.serverdata.net ([169.254.3.11]) by EXHUB020-2.exch020.serverdata.net ([206.225.164.29]) with mapi; Wed, 19 Feb 2014 03:45:56 -0800
From: Tobias Oberstein <tobias.oberstein@tavendo.de>
To: Takeshi Yoshino <tyoshino@google.com>, Yutaka Hirano <yhirano@google.com>,  Peter Thorson <webmaster@zaphoyd.com>, Roberto Peon <fenix@google.com>, "hybi@ietf.org" <hybi@ietf.org>
Date: Wed, 19 Feb 2014 03:45:52 -0800
Thread-Topic: Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: [hybi] Discontinuation of mux ...)
Thread-Index: Ac8tKc9VFZCVubpLRiC/+L/v0gLoSwAPBnAw
Message-ID: <634914A010D0B943A035D226786325D4446C537686@EXVMBX020-12.exch020.serverdata.net>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com>
In-Reply-To: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com>
Accept-Language: de-DE, en-US
Content-Language: de-DE
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
acceptlanguage: de-DE, en-US
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: base64
MIME-Version: 1.0
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/Xz_gTZGtVKQHh_dPYnAvFC3-loA
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 19 Feb 2014 11:46:03 -0000
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From nobody Wed Feb 19 08:59:10 2014
Return-Path: <francis@aspl.es>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id BFCB41A025D for <hybi@ietfa.amsl.com>; Wed, 19 Feb 2014 08:59:08 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 2.983
X-Spam-Level: **
X-Spam-Status: No, score=2.983 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, FH_HOST_ALMOST_IP=1.889, HOST_EQ_STATIC=1.172, HOST_EQ_STATICIP=1.511, HOST_MISMATCH_NET=0.311, HTML_MESSAGE=0.001, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id netFKGpEnv-F for <hybi@ietfa.amsl.com>; Wed, 19 Feb 2014 08:59:06 -0800 (PST)
Received: from mail.aspl.es (196.Red-212-170-101.staticIP.rima-tde.net [212.170.101.196]) by ietfa.amsl.com (Postfix) with ESMTP id 641DB1A032A for <hybi@ietf.org>; Wed, 19 Feb 2014 08:59:06 -0800 (PST)
Received: from localhost (localhost [127.0.0.1]) by mail.aspl.es (Postfix) with ESMTP id D2F77312003; Wed, 19 Feb 2014 17:59:01 +0100 (CET)
X-Virus-Scanned: amavisd-new at aspl.es
Received: from mail.aspl.es ([127.0.0.1]) by localhost (dolphin.aspl.es [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Us+a65ycOTbv; Wed, 19 Feb 2014 17:59:01 +0100 (CET)
Received: from [192.168.0.132] (unknown [192.168.0.132]) by mail.aspl.es (Postfix) with ESMTP id 4A859312001; Wed, 19 Feb 2014 17:59:01 +0100 (CET)
From: Francis Brosnan Blazquez <francis@aspl.es>
To: Tobias Oberstein <tobias.oberstein@tavendo.de>
In-Reply-To: <634914A010D0B943A035D226786325D4446C537686@EXVMBX020-12.exch020.serverdata.net>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <634914A010D0B943A035D226786325D4446C537686@EXVMBX020-12.exch020.serverdata.net>
Content-Type: multipart/alternative; boundary="=-LXhdOrrV06hVFElMWg+I"
Organization: ASPL
Date: Wed, 19 Feb 2014 17:59:10 +0100
Message-ID: <1392829150.29641.44.camel@vulcan.aspl.local>
Mime-Version: 1.0
X-Mailer: Evolution 2.30.3 
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/bJFfRrNGOQA5WhjiDPf-ZumJvuc
Cc: "hybi@ietf.org" <hybi@ietf.org>, Yutaka Hirano <yhirano@google.com>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 19 Feb 2014 16:59:08 -0000

--=-LXhdOrrV06hVFElMWg+I
Content-Type: text/plain; charset="ISO-8859-15"
Content-Transfer-Encoding: 7bit

Hi Tobias,

> I don't think this should be removed from RFC6455.
> 
> The extension mechanisms of RFC6455 allow for flexible design of
> WebSocket extensions. Being able to interject existing and new control
> messages, and to interleave multiple data messages is part of that
> flexibility.
>  
> Also: there seems to be an outburst of "lets remove X" from RFC6455
> initiatives lately.
> 
> Not sure, but it's my impression that these are all related to the
> desire for some kind of WebSocket-over-HTTP2 protocol. And more so: to
> mold RFC6455 so that it "fits into HTTP2".
> 
> I don't agree with that approach. I think HTTP2 should take WebSocket
> "as is", if it wants to act as a transport instead of attempting to
> redefine it. Or define it's own messaging (not bound by WS). 

+1

I couldn't express it better, plus the fact it took a long road to reach
RFC6455....*and*
there's a lot of software out there running on production env assuming
current WS
definition.

Best Regards,


--=-LXhdOrrV06hVFElMWg+I
Content-Type: text/html; charset="utf-8"
Content-Transfer-Encoding: 7bit

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 TRANSITIONAL//EN">
<HTML>
<HEAD>
  <META HTTP-EQUIV="Content-Type" CONTENT="text/html; CHARSET=UTF-8">
  <META NAME="GENERATOR" CONTENT="GtkHTML/3.30.3">
</HEAD>
<BODY>
Hi Tobias,<BR>
<BLOCKQUOTE TYPE=CITE>
    <TT><FONT COLOR="#1a1a1a">I don't think this should be removed from RFC6455.</FONT></TT><BR>
    <BR>
    <TT><FONT COLOR="#1a1a1a">The extension mechanisms of RFC6455 allow for flexible design of WebSocket extensions. Being able to interject existing and new control messages, and to interleave multiple data messages is part of that flexibility.</FONT></TT><BR>
    <TT><FONT COLOR="#1a1a1a">&nbsp;</FONT></TT><BR>
    <TT><FONT COLOR="#1a1a1a">Also: there seems to be an outburst of &quot;lets remove X&quot; from RFC6455 initiatives lately.</FONT></TT><BR>
    <BR>
    <TT><FONT COLOR="#1a1a1a">Not sure, but it's my impression that these are all related to the desire for some kind of WebSocket-over-HTTP2 protocol. And more so: to mold RFC6455 so that it &quot;fits into HTTP2&quot;.</FONT></TT><BR>
    <BR>
    <TT><FONT COLOR="#1a1a1a">I don't agree with that approach. I think HTTP2 should take WebSocket &quot;as is&quot;, if it wants to act as a transport instead of attempting to redefine it. Or define it's own messaging (not bound by WS).</FONT></TT> <BR>
</BLOCKQUOTE>
+1<BR>
<BR>
I couldn't express it better, plus the fact it took a long road to reach RFC6455....*and*<BR>
there's a lot of software out there running on production env assuming current WS<BR>
definition.<BR>
<BR>
Best Regards,<BR>
<BR>
</BODY>
</HTML>

--=-LXhdOrrV06hVFElMWg+I--


From nobody Wed Feb 19 11:33:21 2014
Return-Path: <fenix@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 809B71A01F0 for <hybi@ietfa.amsl.com>; Wed, 19 Feb 2014 11:33:19 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.926
X-Spam-Level: 
X-Spam-Status: No, score=-1.926 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 0rd9ewBQgwSr for <hybi@ietfa.amsl.com>; Wed, 19 Feb 2014 11:33:17 -0800 (PST)
Received: from mail-ob0-x235.google.com (mail-ob0-x235.google.com [IPv6:2607:f8b0:4003:c01::235]) by ietfa.amsl.com (Postfix) with ESMTP id 86CCA1A015E for <hybi@ietf.org>; Wed, 19 Feb 2014 11:33:17 -0800 (PST)
Received: by mail-ob0-f181.google.com with SMTP id va2so976691obc.40 for <hybi@ietf.org>; Wed, 19 Feb 2014 11:33:14 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=1lKNzQCF3MhzROMFTNn6nOC23bzuApDhtWI3FfWqhdk=; b=IvxtS+0JpmLkYaVK0ZiYsWBHAz+rNpYrlrlr2NNRtAFdRMk7JoGEb7E0t/Rld+xvtz yBYVsa1/LJxgiQd/9h36Uz6GdiTXhJtFdzQ/UgL8c3JZM62uTnVv0LPGmGqiNgbKCyEQ E67cbtZMs+cMntOmSWmUhiR2tF61aJIVBd4wkL8hYm4z22RC5IhVrKkTzZEQcJt+3gb5 HxemFZVx03QE5vN7+HHHdIzsI+GrBZcRzUsVNJWH852M1TvSkQrl6TTbIiWwGCQm1/FV 87+lXKWgc8FhTYTfGhwvIAasGYUQLDX6XbbNrLRl4IReooWPjIok7wuiKehSPY0nQ+Zg HHHg==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=1lKNzQCF3MhzROMFTNn6nOC23bzuApDhtWI3FfWqhdk=; b=X5NnuSjDjVOb33Yf3RGwnj5E99PBrkNgbLlXE0pxxHxxu0qE9e+3HG8NpmlxfU4raz Mvu1blKXCuY5/NyJuupszdi2Ghv7vqfQxeU1J6z/0yuu+XI+j8oFPxhm2r7vAddkLRvK U8Iu8OV7C9vT14CSrvNOKPVQUD32R17Rb1DFXrdGMTF61FuO7msgUytlYuzIqNuBF1o8 IRylEB+66cqE1VCeGf2CdD3NTG59IjtBYgImH9SBqcRx6+tcTKXiCirdmtgj6EtCeWM2 f2CoYLsvvQ7pSTw7U1+1LASyDjPu6jQ/TJwgv3w39+5DZVDfgzI8jn51Jg0sZFwd7rtt 5b5g==
X-Gm-Message-State: ALoCoQmvMNiiJAKtsQZgiKBJvKgmG9vT71er8hHRVKmTMSnuQ3hi2c1FUiJ2G2dqFHaAtYuoEAPj74wRvM+iCNXXeEXb0RiEIBDiWse2dWxiQoC/1plzIPWaI/+oHpSx9u+vR9HsbBHFPJecaDh8YML6ApklrqDh5Hro/wv8Fl5MRIsxdtoL3hiIMohi7ToVJYKDsVqwxJ6P
MIME-Version: 1.0
X-Received: by 10.182.250.200 with SMTP id ze8mr2739492obc.72.1392838394059; Wed, 19 Feb 2014 11:33:14 -0800 (PST)
Received: by 10.182.221.231 with HTTP; Wed, 19 Feb 2014 11:33:13 -0800 (PST)
In-Reply-To: <CABihn6E4x+bV6c_564RgbvRFivihGyF_CyFH3e7nZ2D+8FkbXg@mail.gmail.com>
References: <CAH9hSJa0XvzmS1MmmPmqKjQs0P8EUP8n=x+b0im1S2CO5wL1NA@mail.gmail.com> <CAH9hSJbQ9kWjtYCd6UM9hAGA=gr0oE7M5BVPtwJq1K8-95hAkQ@mail.gmail.com> <CABihn6E4x+bV6c_564RgbvRFivihGyF_CyFH3e7nZ2D+8FkbXg@mail.gmail.com>
Date: Wed, 19 Feb 2014 11:33:13 -0800
Message-ID: <CAGzyod63WAFcEkdtN+bmb6oVM-X7L8h8e44o9YZKDypo_ZQrMA@mail.gmail.com>
From: Roberto Peon <fenix@google.com>
To: Yutaka Hirano <yhirano@chromium.org>
Content-Type: multipart/alternative; boundary=089e0160c660b5341704f2c775b4
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/JWQ9HoHnWrWhayYr7N9QfCCI1S8
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss per-frame extension? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 19 Feb 2014 19:33:19 -0000

--089e0160c660b5341704f2c775b4
Content-Type: text/plain; charset=UTF-8

I agree with Yutaka and Takeshi-- The world would be better without frame
extensions.

-=R


On Wed, Feb 19, 2014 at 2:42 AM, Yutaka Hirano <yhirano@chromium.org> wrote:

> Now, I want to ask you, "do you miss per-frame extensions?".
>
> No.
> I totally agree with Takeshi.
>
>
> On Wed, Feb 19, 2014 at 1:11 PM, Takeshi Yoshino <tyoshino@google.com>wrote:
>
>> My answer is "no, I don't". I think we should drop per-frame extension
>> capability. Even RFC 6455 should be updated to drop it when we get chance
>> to do so.
>>
>
>

--089e0160c660b5341704f2c775b4
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">I agree with Yutaka and Takeshi-- The world would be bette=
r without frame extensions.<div><br></div><div>-=3DR</div></div><div class=
=3D"gmail_extra"><br><br><div class=3D"gmail_quote">On Wed, Feb 19, 2014 at=
 2:42 AM, Yutaka Hirano <span dir=3D"ltr">&lt;<a href=3D"mailto:yhirano@chr=
omium.org" target=3D"_blank">yhirano@chromium.org</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div dir=3D"ltr"><div class=3D""><blockquote=
 class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:=
1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding-left=
:1ex">
Now, I want to ask you, &quot;do you miss per-frame extensions?&quot;.</blo=
ckquote></div>
No.<div>I totally agree with Takeshi.</div></div><div class=3D"HOEnZb"><div=
 class=3D"h5"><div class=3D"gmail_extra"><br><br><div class=3D"gmail_quote"=
>On Wed, Feb 19, 2014 at 1:11 PM, Takeshi Yoshino <span dir=3D"ltr">&lt;<a =
href=3D"mailto:tyoshino@google.com" target=3D"_blank">tyoshino@google.com</=
a>&gt;</span> wrote:<br>

<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div dir=3D"ltr"><div class=3D"gmail_extra">=
My answer is &quot;no, I don&#39;t&quot;. I think we should drop per-frame =
extension capability. Even RFC 6455 should be updated to drop it when we ge=
t chance to do so.</div>

</div>
</blockquote></div><br></div>
</div></div></blockquote></div><br></div>

--089e0160c660b5341704f2c775b4--


From nobody Wed Feb 19 11:36:05 2014
Return-Path: <fenix@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 1BECD1A03F7 for <hybi@ietfa.amsl.com>; Wed, 19 Feb 2014 11:36:04 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.926
X-Spam-Level: 
X-Spam-Status: No, score=-1.926 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Mvt43VdunApZ for <hybi@ietfa.amsl.com>; Wed, 19 Feb 2014 11:36:01 -0800 (PST)
Received: from mail-oa0-x232.google.com (mail-oa0-x232.google.com [IPv6:2607:f8b0:4003:c02::232]) by ietfa.amsl.com (Postfix) with ESMTP id CC3981A03DE for <hybi@ietf.org>; Wed, 19 Feb 2014 11:36:01 -0800 (PST)
Received: by mail-oa0-f50.google.com with SMTP id n16so997534oag.23 for <hybi@ietf.org>; Wed, 19 Feb 2014 11:35:58 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=5VedZIO6KhWK0xZdN9bjdo+qFDsunXMaKwqYPUVVmAU=; b=ieseSQ0FpEET72lS2ilJjxUY0CvcxXABWpEWOhKWAsZNY+3SAuezC3cFS3xFn7CG27 bEk4744msi+cQnljPA6LRlJ5ONgOnIDQahoWCQkzJ6ay4KnO1TVYp4y3Nsj4FRZMMM/g VRmQd5jvg13xVyVijz+8Wn5sDZQGXJTqKcPKZ1jupgm6iDJXVe0+Qs6EF35K8CKruWu/ uwQLqt4ZUWcQLB1N5hmUvO9L1mrmfNDC3WI5LaewdVkZzHA0yHuWCON5o/OM2nb15eOe uFGHZRQUER8DjFtXa0hngLYWohUvUMI66rwcoPA9JGlOMjIha+nFc95jVzqPt0rxIVsy kFjw==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=5VedZIO6KhWK0xZdN9bjdo+qFDsunXMaKwqYPUVVmAU=; b=ASUr4q8k7GbIAnbqYglQbsZuuh6gYtqDWyqsmKpHAp2jMfVNPWmdh8Qzg1YY6ODOMp aulShl7TSxTMe0KBFPaCAl9Q3Iq8dy9FasrDTrHiWlGbzHq42QimBnvJutkhxPo9k2UE C8aecpmbcZMwDYQZIGM/xlyAIgFw+GsEVHBi3PYIrEZzByyI4JqXehX6h3TPJzsyNBxr mJ3dnESWAvmz2T7yH8ciAjJa0POe/iwF4cZcx7DmAqCdJahLORmBcdNj9o3G7p/MiLOz U1KKxoOfxICgVGu5QkOCQCJJZ7E73R0b7DgoYXRdsxLmge1hRcyhHCKK/TelAT6+T8ko ay/w==
X-Gm-Message-State: ALoCoQlZlcuBuO7L0zII5o1e25fzwsXJWxIQUqdAQmM7A3Y+qfflJ50/wIVbZP5HlBpRu5NxrtUv3evdcHtnBuJCYrtm/3yrFrq4GcjEJGKJXlrwgHnKVfeONniKTTQg6X3fDfiyWhwDUAYW288NI9wVHITQDrlOwrg7XmzTeLn9PXEVJFk1FV3gKDJIhYUCsigqPywotNLn
MIME-Version: 1.0
X-Received: by 10.60.93.102 with SMTP id ct6mr33186601oeb.12.1392838558416; Wed, 19 Feb 2014 11:35:58 -0800 (PST)
Received: by 10.182.221.231 with HTTP; Wed, 19 Feb 2014 11:35:58 -0800 (PST)
In-Reply-To: <634914A010D0B943A035D226786325D4446C537686@EXVMBX020-12.exch020.serverdata.net>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <634914A010D0B943A035D226786325D4446C537686@EXVMBX020-12.exch020.serverdata.net>
Date: Wed, 19 Feb 2014 11:35:58 -0800
Message-ID: <CAGzyod7JQZJpk_bWP0i9Pj9hOM6+om5RvXVE4j32c5hm2yVRGg@mail.gmail.com>
From: Roberto Peon <fenix@google.com>
To: Tobias Oberstein <tobias.oberstein@tavendo.de>
Content-Type: multipart/alternative; boundary=047d7b33c87e81154104f2c77f6b
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/LeuMyWGby8NYW5TPwKCryvtrujE
Cc: "hybi@ietf.org" <hybi@ietf.org>, Yutaka Hirano <yhirano@google.com>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 19 Feb 2014 19:36:04 -0000

--047d7b33c87e81154104f2c77f6b
Content-Type: text/plain; charset=UTF-8

If they're not used, and implementors don't have plans to use them however,
they should be removed.
As we learned from HTTP/1.1, parts of the spec which are not exercised are
likely to cause problems later.
-=R


On Wed, Feb 19, 2014 at 3:45 AM, Tobias Oberstein <
tobias.oberstein@tavendo.de> wrote:

> >The second point I want to hear your opinion is: "do you miss
> interjectable WebSocket level control frame?"
>
> >Since RFC 6455's control frames can be interjected in the middle of a
> data message, to keep this as-is, we need to prevent FRAME boundary from
> being modified by multiplexing mechanism. Possible solutions are:
> >- encapsulate WebSocket frame
> >- prepare some mechanism (HTTP/2.0 frame type) in HTTP/2.0 layer similar
> to HTTP/2.0 PING frame that provides a box to which we can map WebSocket
> level control (PING/PONG/CLOSE/and other reserved control opcodes)
>
> >Please respond to this thread with your answer and rationale.
>
> I don't think this should be removed from RFC6455.
>
> The extension mechanisms of RFC6455 allow for flexible design of WebSocket
> extensions. Being able to interject existing and new control messages, and
> to interleave multiple data messages is part of that flexibility.
>
> Also: there seems to be an outburst of "lets remove X" from RFC6455
> initiatives lately.
>
> Not sure, but it's my impression that these are all related to the desire
> for some kind of WebSocket-over-HTTP2 protocol. And more so: to mold
> RFC6455 so that it "fits into HTTP2".
>
> I don't agree with that approach. I think HTTP2 should take WebSocket "as
> is", if it wants to act as a transport instead of attempting to redefine
> it. Or define it's own messaging (not bound by WS).
>
> /Tobias
>

--047d7b33c87e81154104f2c77f6b
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div>If they&#39;re not used, and implementors don&#39;t h=
ave plans to use them however, they should be removed.</div><div>As we lear=
ned from HTTP/1.1, parts of the spec which are not exercised are likely to =
cause problems later.</div>
<div>-=3DR<br></div></div><div class=3D"gmail_extra"><br><br><div class=3D"=
gmail_quote">On Wed, Feb 19, 2014 at 3:45 AM, Tobias Oberstein <span dir=3D=
"ltr">&lt;<a href=3D"mailto:tobias.oberstein@tavendo.de" target=3D"_blank">=
tobias.oberstein@tavendo.de</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div class=3D"">&gt;The second point I want =
to hear your opinion is: &quot;do you miss interjectable WebSocket level co=
ntrol frame?&quot;<br>

<br>
&gt;Since RFC 6455&#39;s control frames can be interjected in the middle of=
 a data message, to keep this as-is, we need to prevent FRAME boundary from=
 being modified by multiplexing mechanism. Possible solutions are:<br>

&gt;- encapsulate WebSocket frame<br>
&gt;- prepare some mechanism (HTTP/2.0 frame type) in HTTP/2.0 layer simila=
r to HTTP/2.0 PING frame that provides a box to which we can map WebSocket =
level control (PING/PONG/CLOSE/and other reserved control opcodes)<br>

<br>
&gt;Please respond to this thread with your answer and rationale.<br>
<br>
</div>I don&#39;t think this should be removed from RFC6455.<br>
<br>
The extension mechanisms of RFC6455 allow for flexible design of WebSocket =
extensions. Being able to interject existing and new control messages, and =
to interleave multiple data messages is part of that flexibility.<br>
<br>
Also: there seems to be an outburst of &quot;lets remove X&quot; from RFC64=
55 initiatives lately.<br>
<br>
Not sure, but it&#39;s my impression that these are all related to the desi=
re for some kind of WebSocket-over-HTTP2 protocol. And more so: to mold RFC=
6455 so that it &quot;fits into HTTP2&quot;.<br>
<br>
I don&#39;t agree with that approach. I think HTTP2 should take WebSocket &=
quot;as is&quot;, if it wants to act as a transport instead of attempting t=
o redefine it. Or define it&#39;s own messaging (not bound by WS).<br>

<span class=3D"HOEnZb"><font color=3D"#888888"><br>
/Tobias<br>
</font></span></blockquote></div><br></div>

--047d7b33c87e81154104f2c77f6b--


From nobody Wed Feb 19 20:44:41 2014
Return-Path: <ricea@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 8FC4A1A0320 for <hybi@ietfa.amsl.com>; Wed, 19 Feb 2014 20:44:40 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.926
X-Spam-Level: 
X-Spam-Status: No, score=-1.926 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.548, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 4iI48WQWGsfw for <hybi@ietfa.amsl.com>; Wed, 19 Feb 2014 20:44:38 -0800 (PST)
Received: from mail-oa0-x230.google.com (mail-oa0-x230.google.com [IPv6:2607:f8b0:4003:c02::230]) by ietfa.amsl.com (Postfix) with ESMTP id 063311A02CC for <hybi@ietf.org>; Wed, 19 Feb 2014 20:44:37 -0800 (PST)
Received: by mail-oa0-f48.google.com with SMTP id l6so1581988oag.7 for <hybi@ietf.org>; Wed, 19 Feb 2014 20:44:34 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=5FmV/6dvs+jtC+o3ncN43g6uJrBXOZzZ9T8sv1+lHNw=; b=GKIJK3rmZJFBuOGLqR9KQbFcDUpSTWLZB/HhhBl64vtR4ZD170BAP7gLbV8CgdI8b0 WScAR4IseOITcbc5FKgRlPvf4goInDyBcX0oo7x3V9rCLBhgm1Ip30jg10Gjm2JQ4Tqf e5gocc0/0i5xK/X2XTcfl13zJkQ3zKyNHD3a6M2qzjBB0D/YKSS4pBSGpTK/BtjiC56R pHAbvgNA4kpGLtnBnnTYgcsE6CEDDHtbxEkrlpmzi+RLceVDdH4S9RQQcWEWLm2glQb+ oF2Kr5uorX2i+7lVsa/yOt46HZL+d9Rka+lPMRPkgjxaLFNxBI02SfrZ3Z/tmJJzH0UP Sjrw==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=5FmV/6dvs+jtC+o3ncN43g6uJrBXOZzZ9T8sv1+lHNw=; b=YdcajEQryX+KVNy9B72t1olk32f5ztqHnxlFqS/WcRFIKR39xon8Yu9AtVKzPVuJiS VwlSMEzZjzahZHTrOIpeL9hFwiu5hrDsvGuo/aLnWK2K2szLlP42B7H7l3p0+c1vVPWi uQlUJ2SztX5oaJl5K6YWT9HGkUG7SvzgY/KCEj9GFKfPzrXRvfSX38lAstmcZbYcJy1j 1M1ftV4N2keRWaK1jovZQGZjKJ22OCFXhXlLFOb2shXp/rwBl0joRVMK1tJuq2ZFubhe PvO3hYj+vA9jp/RXSe2LmfXIc6qTCQb1c8UWue796urmdaNkM3bwdmMvuGh1sH1iMzMO GDxA==
X-Gm-Message-State: ALoCoQmkfo41JuKThunkAmV7NPnX0QrdluFvhyAVnWCGeEef+9P7eiEqeW37O8+K4SoKQ+NqiwOCS5zdZSqRXyilUQdw7weEYdDMSJBcmti4p0v5CQioIFsQPZQN9Y9A2ZlsJ+V1sN7pJ0O1s+qE/KTrxH87JKkyE0D18/oWTgF/7QZOKuZrGno3ZsEh5XwZvUFfamC6eOKM
MIME-Version: 1.0
X-Received: by 10.182.88.69 with SMTP id be5mr5205822obb.55.1392871474047; Wed, 19 Feb 2014 20:44:34 -0800 (PST)
Received: by 10.182.118.41 with HTTP; Wed, 19 Feb 2014 20:44:33 -0800 (PST)
In-Reply-To: <CAGzyod7JQZJpk_bWP0i9Pj9hOM6+om5RvXVE4j32c5hm2yVRGg@mail.gmail.com>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <634914A010D0B943A035D226786325D4446C537686@EXVMBX020-12.exch020.serverdata.net> <CAGzyod7JQZJpk_bWP0i9Pj9hOM6+om5RvXVE4j32c5hm2yVRGg@mail.gmail.com>
Date: Thu, 20 Feb 2014 13:44:33 +0900
Message-ID: <CAHixhFqoSP6WBLy-i28_7+vttXbWdMc1qXpCdZr70TZ_JFcvYA@mail.gmail.com>
From: Adam Rice <ricea@google.com>
To: Roberto Peon <fenix@google.com>
Content-Type: multipart/alternative; boundary=14dae9399b136dbf9604f2cf2932
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/3mT9pVOLd4SkgiyoVoRMu_66Bz4
Cc: Yutaka Hirano <yhirano@google.com>, "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Thu, 20 Feb 2014 04:44:40 -0000

--14dae9399b136dbf9604f2cf2932
Content-Type: text/plain; charset=UTF-8

Since interjectable control frames are implemented and in active use, I
don't think they should be removed from the successor to RFC6455.

My answer was purely in respect to the question of "would I miss them if
they were gone", to which my answer is no.


On 20 February 2014 04:35, Roberto Peon <fenix@google.com> wrote:

> If they're not used, and implementors don't have plans to use them
> however, they should be removed.
> As we learned from HTTP/1.1, parts of the spec which are not exercised are
> likely to cause problems later.
> -=R
>
>
> On Wed, Feb 19, 2014 at 3:45 AM, Tobias Oberstein <
> tobias.oberstein@tavendo.de> wrote:
>
>> >The second point I want to hear your opinion is: "do you miss
>> interjectable WebSocket level control frame?"
>>
>> >Since RFC 6455's control frames can be interjected in the middle of a
>> data message, to keep this as-is, we need to prevent FRAME boundary from
>> being modified by multiplexing mechanism. Possible solutions are:
>> >- encapsulate WebSocket frame
>> >- prepare some mechanism (HTTP/2.0 frame type) in HTTP/2.0 layer similar
>> to HTTP/2.0 PING frame that provides a box to which we can map WebSocket
>> level control (PING/PONG/CLOSE/and other reserved control opcodes)
>>
>> >Please respond to this thread with your answer and rationale.
>>
>> I don't think this should be removed from RFC6455.
>>
>> The extension mechanisms of RFC6455 allow for flexible design of
>> WebSocket extensions. Being able to interject existing and new control
>> messages, and to interleave multiple data messages is part of that
>> flexibility.
>>
>> Also: there seems to be an outburst of "lets remove X" from RFC6455
>> initiatives lately.
>>
>> Not sure, but it's my impression that these are all related to the desire
>> for some kind of WebSocket-over-HTTP2 protocol. And more so: to mold
>> RFC6455 so that it "fits into HTTP2".
>>
>> I don't agree with that approach. I think HTTP2 should take WebSocket "as
>> is", if it wants to act as a transport instead of attempting to redefine
>> it. Or define it's own messaging (not bound by WS).
>>
>> /Tobias
>>
>
>
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi
>
>

--14dae9399b136dbf9604f2cf2932
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Since interjectable control frames are implemented and in =
active use, I don&#39;t think they should be removed from the successor to =
RFC6455.<div><br></div><div>My answer was purely in respect to the question=
 of &quot;would I miss them if they were gone&quot;, to which my answer is =
no.</div>
</div><div class=3D"gmail_extra"><br><br><div class=3D"gmail_quote">On 20 F=
ebruary 2014 04:35, Roberto Peon <span dir=3D"ltr">&lt;<a href=3D"mailto:fe=
nix@google.com" target=3D"_blank">fenix@google.com</a>&gt;</span> wrote:<br=
><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1=
px #ccc solid;padding-left:1ex">
<div dir=3D"ltr"><div>If they&#39;re not used, and implementors don&#39;t h=
ave plans to use them however, they should be removed.</div><div>As we lear=
ned from HTTP/1.1, parts of the spec which are not exercised are likely to =
cause problems later.</div>
<span class=3D"HOEnZb"><font color=3D"#888888">
<div>-=3DR<br></div></font></span></div><div class=3D"HOEnZb"><div class=3D=
"h5"><div class=3D"gmail_extra"><br><br><div class=3D"gmail_quote">On Wed, =
Feb 19, 2014 at 3:45 AM, Tobias Oberstein <span dir=3D"ltr">&lt;<a href=3D"=
mailto:tobias.oberstein@tavendo.de" target=3D"_blank">tobias.oberstein@tave=
ndo.de</a>&gt;</span> wrote:<br>

<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div>&gt;The second point I want to hear you=
r opinion is: &quot;do you miss interjectable WebSocket level control frame=
?&quot;<br>


<br>
&gt;Since RFC 6455&#39;s control frames can be interjected in the middle of=
 a data message, to keep this as-is, we need to prevent FRAME boundary from=
 being modified by multiplexing mechanism. Possible solutions are:<br>


&gt;- encapsulate WebSocket frame<br>
&gt;- prepare some mechanism (HTTP/2.0 frame type) in HTTP/2.0 layer simila=
r to HTTP/2.0 PING frame that provides a box to which we can map WebSocket =
level control (PING/PONG/CLOSE/and other reserved control opcodes)<br>


<br>
&gt;Please respond to this thread with your answer and rationale.<br>
<br>
</div>I don&#39;t think this should be removed from RFC6455.<br>
<br>
The extension mechanisms of RFC6455 allow for flexible design of WebSocket =
extensions. Being able to interject existing and new control messages, and =
to interleave multiple data messages is part of that flexibility.<br>
<br>
Also: there seems to be an outburst of &quot;lets remove X&quot; from RFC64=
55 initiatives lately.<br>
<br>
Not sure, but it&#39;s my impression that these are all related to the desi=
re for some kind of WebSocket-over-HTTP2 protocol. And more so: to mold RFC=
6455 so that it &quot;fits into HTTP2&quot;.<br>
<br>
I don&#39;t agree with that approach. I think HTTP2 should take WebSocket &=
quot;as is&quot;, if it wants to act as a transport instead of attempting t=
o redefine it. Or define it&#39;s own messaging (not bound by WS).<br>


<span><font color=3D"#888888"><br>
/Tobias<br>
</font></span></blockquote></div><br></div>
</div></div><br>_______________________________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
<br></blockquote></div><br></div>

--14dae9399b136dbf9604f2cf2932--


From nobody Mon Feb 24 03:03:37 2014
Return-Path: <duerst@it.aoyama.ac.jp>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 731841A07E7 for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 03:03:33 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 0.762
X-Spam-Level: 
X-Spam-Status: No, score=0.762 tagged_above=-999 required=5 tests=[BAYES_50=0.8, HELO_EQ_JP=1.244, HOST_EQ_JP=1.265, MIME_8BIT_HEADER=0.3, RCVD_IN_DNSWL_MED=-2.3, RP_MATCHES_RCVD=-0.547] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id t90DJqqtsiJj for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 03:03:31 -0800 (PST)
Received: from scintmta02.scbb.aoyama.ac.jp (scintmta02.scbb.aoyama.ac.jp [133.2.253.34]) by ietfa.amsl.com (Postfix) with ESMTP id 9F4891A07E5 for <hybi@ietf.org>; Mon, 24 Feb 2014 03:03:31 -0800 (PST)
Received: from scmse02.scbb.aoyama.ac.jp ([133.2.253.231]) by scintmta02.scbb.aoyama.ac.jp (secret/secret) with SMTP id s1OB3NZN030533; Mon, 24 Feb 2014 20:03:23 +0900
Received: from (unknown [133.2.206.134]) by scmse02.scbb.aoyama.ac.jp with smtp id 2b50_271e_4772c222_9d43_11e3_a3cb_001e6722eec2; Mon, 24 Feb 2014 20:03:22 +0900
Received: from [IPv6:::1] (unknown [133.2.210.1]) by itmail2.it.aoyama.ac.jp (Postfix) with ESMTP id 06B36BF4CD; Mon, 24 Feb 2014 20:03:23 +0900 (JST)
Message-ID: <530B26EF.4020106@it.aoyama.ac.jp>
Date: Mon, 24 Feb 2014 20:03:11 +0900
From: =?UTF-8?B?Ik1hcnRpbiBKLiBEw7xyc3Qi?= <duerst@it.aoyama.ac.jp>
Organization: Aoyama Gakuin University
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 6.1; en-US; rv:1.9.1.9) Gecko/20100722 Eudora/3.0.4
MIME-Version: 1.0
To: Takeshi Yoshino <tyoshino@google.com>
References: <CAH9hSJa0XvzmS1MmmPmqKjQs0P8EUP8n=x+b0im1S2CO5wL1NA@mail.gmail.com> <CAH9hSJbQ9kWjtYCd6UM9hAGA=gr0oE7M5BVPtwJq1K8-95hAkQ@mail.gmail.com>
In-Reply-To: <CAH9hSJbQ9kWjtYCd6UM9hAGA=gr0oE7M5BVPtwJq1K8-95hAkQ@mail.gmail.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/1u6pr62PCQU06AV5xu5QsmsL_SA
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss per-frame extension?
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 24 Feb 2014 11:03:33 -0000

On 2014/02/19 13:11, Takeshi Yoshino wrote:
> My answer is "no, I don't". I think we should drop per-frame extension
> capability. Even RFC 6455 should be updated to drop it when we get chance
> to do so.

This would be my answer, too.

Regards,   Martin.


From nobody Mon Feb 24 08:54:39 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 98E741A0114 for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 08:54:37 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 0.775
X-Spam-Level: 
X-Spam-Status: No, score=0.775 tagged_above=-999 required=5 tests=[BAYES_50=0.8, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.547, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id z1whdZstOU8h for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 08:54:36 -0800 (PST)
Received: from mail-we0-x22d.google.com (mail-we0-x22d.google.com [IPv6:2a00:1450:400c:c03::22d]) by ietfa.amsl.com (Postfix) with ESMTP id ED3B91A01AB for <hybi@ietf.org>; Mon, 24 Feb 2014 08:54:17 -0800 (PST)
Received: by mail-we0-f173.google.com with SMTP id x48so4935227wes.32 for <hybi@ietf.org>; Mon, 24 Feb 2014 08:54:16 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc:content-type; bh=G0kbRX5uHyJD5fcETmvs54/43hqveSy94GKbYa2Q9Zs=; b=lFFdUXo8qAuUMo6wRuF06gsWJH+tojP+eRBrHj5Tzq6ygaFwvzk4HPZifyCX99s3Is upNqt7do39x/6/0DgZ3qbPysRKzxamtdQqx1F4aJvUJK78NRPMG8VBJl2g2wAov2DxfY pSqYfGFczc75al7EzYkUdGbZu/dfuWPlNDfWETWpfIrxmr2hyb4LDWlMo2FHhigNsZLt xsJ60A51+Pz9LCrt3/imytIiAIUBumAEkDN8fY1edw/v2gbC6LlquJ7AEOy3r6XZ7tre lrMDT28zLwmbPwhKGEH2kNpTsnARytg2uln8l8jt5LVaUFY/Rs4dRm7Pd38PT9hIZgWp PKVg==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc:content-type; bh=G0kbRX5uHyJD5fcETmvs54/43hqveSy94GKbYa2Q9Zs=; b=Tuust1IrAfBOsgldZjzGcH0AjbPrHDcrSB8KrUb7cbfPohrLO/qxJEHRgdIu8DoKDZ Sgqku2l1XBsmwrevZ8sIkW3T38YupJoVJTFvwc/s4JOECdMla7YtN2JivL74aU1vd9KJ D68ue6sfp/1w7jcv5QJmQNc0Mfmc2JGY9IqQbucAB2wxyiUcJrhShTTH658wKuO2z1hv 3I3yrBkiv+cXYwsvegk++Sn1H870lknulWcMPxoG1ii8AQAN9RzJGuCzccFHHdo0TwNG k358Oijn2Djn1hkbqTv/ZTTMDz/HKBmXU2Z6dOHVJK0kyznoBTvhZ39g4+YoJAclMmgb ay/w==
X-Gm-Message-State: ALoCoQnr9Af5bvgE5WAHLKy/1ALJQUqiBpXS1GvDJC8IwpJYyYZaeVU0r+31HjULDj2vi9agY7yDmxS+y4kvUcNlxYuVxL3e0OBIYziTbH0v24X+IPI70Ow6e4YpDpnfghhDYWSFsbcxdp0BE7vk4KOBpNdmPawq9SbRMgptPBjD1sRBtUJqpj/MHDUP7Jcn0olSaHTbseFG
X-Received: by 10.194.143.40 with SMTP id sb8mr20083392wjb.15.1393260856893; Mon, 24 Feb 2014 08:54:16 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Mon, 24 Feb 2014 08:53:56 -0800 (PST)
In-Reply-To: <CAGzyod7JQZJpk_bWP0i9Pj9hOM6+om5RvXVE4j32c5hm2yVRGg@mail.gmail.com>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <634914A010D0B943A035D226786325D4446C537686@EXVMBX020-12.exch020.serverdata.net> <CAGzyod7JQZJpk_bWP0i9Pj9hOM6+om5RvXVE4j32c5hm2yVRGg@mail.gmail.com>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Tue, 25 Feb 2014 01:53:56 +0900
Message-ID: <CAH9hSJa+1TWaJrd7fjDwy+qiB+8HdoSy7ji6xeKNQCXvT1i7aw@mail.gmail.com>
To: Roberto Peon <fenix@google.com>, Francis Brosnan Blazquez <francis@aspl.es>, Adam Rice <ricea@google.com>
Content-Type: multipart/alternative; boundary=089e0115e78c74785f04f329d20c
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/c7ocKWOKNtMzUHrEJdKIa-P6nYU
Cc: Yutaka Hirano <yhirano@google.com>, "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 24 Feb 2014 16:54:37 -0000

--089e0115e78c74785f04f329d20c
Content-Type: text/plain; charset=UTF-8

Thanks all.

Sorry for not being clear. As you see by looking at the title, this is
branch of the mux discussion thread. I asked this question in the context
of WebSocket/HTTP/2.0 discussion. Not to think of removal of the feature on
the next version of the raw WebSocket Protocol. Please note that. Thanks.

--089e0115e78c74785f04f329d20c
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Thanks all.<div><br></div><div>Sorry for not being clear. =
As you see by looking at the title, this is branch of the mux discussion th=
read. I asked this question in the context of WebSocket/HTTP/2.0 discussion=
. Not to think of removal of the feature on the next version of the raw Web=
Socket Protocol. Please note that. Thanks.</div>

</div>

--089e0115e78c74785f04f329d20c--


From nobody Mon Feb 24 09:03:44 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id D0C2B1A01AB for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 09:03:40 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 0.775
X-Spam-Level: 
X-Spam-Status: No, score=0.775 tagged_above=-999 required=5 tests=[BAYES_50=0.8, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.547, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 5BqjgQPZtjKr for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 09:03:39 -0800 (PST)
Received: from mail-wi0-x22e.google.com (mail-wi0-x22e.google.com [IPv6:2a00:1450:400c:c05::22e]) by ietfa.amsl.com (Postfix) with ESMTP id 07C361A00AF for <hybi@ietf.org>; Mon, 24 Feb 2014 09:03:38 -0800 (PST)
Received: by mail-wi0-f174.google.com with SMTP id f8so3347568wiw.7 for <hybi@ietf.org>; Mon, 24 Feb 2014 09:03:38 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :content-type; bh=8+l3EsfZFrsdF/FowrmHl+rGp1B0as/mO9OmnHynmfM=; b=ZOXV5lYVMfoQAjvM6wQjALUH2VUztnv9/+1MUrQQ/CeCgOeRUFvk455rL/fRMj/hZ3 xPsUP5bFVblPrbBfQcI1umIhJBeB007nBRfVNkhVrZWsphnCCoXnZdUC+fMCLNAeEwnD Ds/PPpFCG+AiBHBTOfNW0XLaFgrdKGj9P/2UorP4GejzO+lvaxlZse4GoRprGuKSyQHp nceJO4nxUFfrlD4m0eLiKA5dXu58agMPMURpg+eeNibPbHt0se7sLlmY8dKQ0aCCHia9 Y3bb3qdDWl/zNIzFp+Z8rhGGRUPPewl80lVPRNFsx28fggC9bK7eYQOQplZRywzh5f1x o2xA==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:content-type; bh=8+l3EsfZFrsdF/FowrmHl+rGp1B0as/mO9OmnHynmfM=; b=jWcpeLyUMRNk8EMz/qSqKbZWf4/hPoy3vyTL+gZvdMW0UsJN2jxBA+MawnO508RyA5 mXbnjN2BILFZe+7dTnmrWjAdhDc/1pfGJCm6O4T+g/aSuXgVBUdxIr6O80Dl7g1TOoNG 4AHqTmUbAgQa4EEUjmXi1X3tIsVqCYR1dOyq/pCN58eScx3uE78735WDbFjr+pG63Wy+ FFlwePGB4eOgFsXfRI26cbNp/0kLKh4R6u5720uexJmXIEo0p/eXOLKTj1gNXBGDb3go 1wZnK2lBfCzjl6UCyZCUUxEcDiV7E0yzEHLPco6LsQZxduFS8nq3et9xogJxgXIJCqvX 4CmQ==
X-Gm-Message-State: ALoCoQkSSqYLHC/OtSF11w+RC1I6MTHbMbuNuzFZwk/TwOQcvdoFlKz34SyhmjneJ8EhZ/BWAWGDCzZrfGT2vuLXQEBUh669hmiLwi44US8Nw286qiko/NJaWl2pct/LkL9Y54rbBO3+BhGUGLNyRFrZhetqHCb8a5j62DObtMVP1HnMMdbUap3ufAso7hRfosqb/tFeZjVP
X-Received: by 10.180.73.173 with SMTP id m13mr15380313wiv.52.1393261417954; Mon, 24 Feb 2014 09:03:37 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Mon, 24 Feb 2014 09:03:17 -0800 (PST)
In-Reply-To: <CAH9hSJaYgxHsxM12BC7a-=TG+Ga+x2EUVCcXvsfcYcgrKsvkxQ@mail.gmail.com>
References: <CAH9hSJaYgxHsxM12BC7a-=TG+Ga+x2EUVCcXvsfcYcgrKsvkxQ@mail.gmail.com>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Tue, 25 Feb 2014 02:03:17 +0900
Message-ID: <CAH9hSJZK9K6kxdJ30dLV-gyiyqt1YrYXzsZfBskKj6jE87-voQ@mail.gmail.com>
To: "hybi@ietf.org" <hybi@ietf.org>
Content-Type: multipart/alternative; boundary=f46d043c07e0e59a5004f329f39f
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/JV0Ljycb00caLCEs060pvHNjrOg
Subject: Re: [hybi] Replacing encapsulation method of the mux extension with END_SEGMENT (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 24 Feb 2014 17:03:41 -0000

--f46d043c07e0e59a5004f329f39f
Content-Type: text/plain; charset=UTF-8

draft-ietf-hybi-websocket-multiplexing-11 (mux extension) encapsulates NOT
frames but segments of the original WebSocket message into a series of
encapsulating WebSocket messages. So,
- The mux extension changes fragmentation of a message as it wants, so
frame boundary in the original traffic is NOT guaranteed to be preserved.
- The mux extension CAN transport interjected control frames in the
original traffic.

That's the choice we made and had in -11.

--f46d043c07e0e59a5004f329f39f
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_extra">draft-ietf-hybi-websocket-multi=
plexing-11 (mux extension) encapsulates NOT frames but segments of the orig=
inal WebSocket message into a series of encapsulating WebSocket messages. S=
o,</div>

<div class=3D"gmail_extra">- The mux extension changes fragmentation of a m=
essage as it wants, so frame boundary in the original traffic is NOT guaran=
teed to be preserved.</div><div class=3D"gmail_extra">- The mux extension C=
AN transport interjected control frames in the original traffic.</div>

<div class=3D"gmail_extra"><br></div><div class=3D"gmail_extra">That&#39;s =
the choice we made and had in -11.</div></div>

--f46d043c07e0e59a5004f329f39f--


From nobody Mon Feb 24 10:28:56 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 8DC811A0212 for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 10:28:55 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 0.775
X-Spam-Level: 
X-Spam-Status: No, score=0.775 tagged_above=-999 required=5 tests=[BAYES_50=0.8, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.547, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 2dB7DqSIVMFM for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 10:28:54 -0800 (PST)
Received: from mail-wg0-x22c.google.com (mail-wg0-x22c.google.com [IPv6:2a00:1450:400c:c00::22c]) by ietfa.amsl.com (Postfix) with ESMTP id 134EB1A011C for <hybi@ietf.org>; Mon, 24 Feb 2014 10:28:53 -0800 (PST)
Received: by mail-wg0-f44.google.com with SMTP id k14so4924230wgh.11 for <hybi@ietf.org>; Mon, 24 Feb 2014 10:28:53 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc:content-type; bh=5uRMBOKG7aInQ2H998M96dGaLVJ8aRhC/+x9Xl5kK4o=; b=W4O8nQRK45UbQYn3zcKXkc3B5bRDpnEamPmIWnfsswOwL9apuSWPGk+rd0Le3p9Hn+ VEV+MCCSbP3vpwi6R/PI3v4XWI5I0uSyC0Zew6UYN3ep6u9sDwYBq8DBMbS0fcE1GiyM CURqtHfhrr/LlFmtPArPe9dhEH++hDZpUeys48Ipr1m3TnygCbohhofxmvUAQB9Bd+Vc gH6j0/yXLSXSw7VmNOKvg/VwhWNaO2NC1xCCr8z/45+wBqySUxQtFxddQsAFKalCZRtW 67nrUYaPSStFrnwIRfSNh7Ovkahdx8AMeHpH0+oqP6r0nysMCF0aDWd3IAfRd2seeR1G yixQ==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc:content-type; bh=5uRMBOKG7aInQ2H998M96dGaLVJ8aRhC/+x9Xl5kK4o=; b=GoZ6FJnk57XtR2pPNK8TQCOTKzgl3mz+oigTr8lIPT1+D6EJyS1NHjTQ5EJ0ec4fvO M8OvSksd3MEnArcOKYn1GgSqU4zNCT5paehQCAxkNmcGEAK+2RX1SbNfeYCPIc61qGpS h2Vot3HFdhCtPZiHleu04sbMZIwtwmnWK8/ql17Y2Ohz3HWAJv+ak21MzxlG7hhLB2Lq IbGbm4ypiQGJoQu3t9OW6YQbWYYnXeyipqE7dLMzpnrOgKCl42i6r6TNejmlpHe1hSXb fznW5pVu8+Al2jECd3lBnlr4nSPJVtmU+NqtVs4SoKJo0deFvwfor/OOsRSe0i9Zkqpq QHpA==
X-Gm-Message-State: ALoCoQlkbTZi+zB9+0ABVjOg//y7TTDVtwf+E0ZBZPSRGIzBX+pq9pioaGlzg6qgA/z+WdXeOerIilW4SBIuErWd7cQzsXm1R8hEq/K0dcKHEt45FEMn1YRLGPgWOddl7g4vo681bCp7IjpAhE9gAqoyEaqCwdvZGkxi4vnZeJWUwKyUYyCMlzpuZL9zoxZltOqVdAZeSu1N
X-Received: by 10.181.13.11 with SMTP id eu11mr15440072wid.30.1393266532932; Mon, 24 Feb 2014 10:28:52 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Mon, 24 Feb 2014 10:28:32 -0800 (PST)
In-Reply-To: <CAH9hSJaeCrLjkhHUzaaGDw-apiSv-eaPZeYGEHBuwRoF3yPCOA@mail.gmail.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <CACuKZqEcA1Pv8RpWfmThMjTzi2BbVMMKXqujs6BxVfxRPZJ9NQ@mail.gmail.com> <CAG4zZZCr4aTfVpw2coX2g0qw++4kdgNCFVze6tHKZ+fJNqb0aQ@mail.gmail.com> <CAG4zZZA4WR50Ea2e36XEgTgpJURwiMFn+DXeYuxHFfZatYzDOQ@mail.gmail.com> <CAH9hSJYv5VXGjS7AfG12-ArEvE6Uj_eE6pPxAiQcchcgV8vHcg@mail.gmail.com> <CAH9hSJbzra7uz7yfKQwfZaP_jhnxwdZyx8JnwCmBGhiMk6rbtg@mail.gmail.com> <CAG4zZZC1cmaH36znAvVLZE_MyJ+ThNk1Ky7tQ-QnKf_qwiQNDw@mail.gmail.com> <CAG4zZZB3h6TTFLUp-ucYQP7tMf-8=h5fh1UqufMh3w-0JN+hSw@mail.gmail.com> <CAH9hSJaeCrLjkhHUzaaGDw-apiSv-eaPZeYGEHBuwRoF3yPCOA@mail.gmail.com>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Tue, 25 Feb 2014 03:28:32 +0900
Message-ID: <CAH9hSJY4NResx4DskJM8agd5ZXo9yHELYaXOpWG-xXK8P4+9zw@mail.gmail.com>
To: Joakim Erdfelt <joakim@intalio.com>, Tobias Oberstein <tobias.oberstein@tavendo.de>,  Peter Thorson <webmaster@zaphoyd.com>, micheil@brandedcode.com
Content-Type: multipart/alternative; boundary=f46d043c7de6c5f25104f32b241b
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/fj0g3oHxrCZPrpCEkiKMLO5Pjmg
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 24 Feb 2014 18:28:55 -0000

--f46d043c7de6c5f25104f32b241b
Content-Type: text/plain; charset=UTF-8

Thanks all for your comments.

First, please note that I wanted to discuss the question in the context of
WS/HTTP/2.0 layering. Not about redesign of RFC 6455 itself.

I asked this question since the length header is the biggest field when
encoded into HTTP/2.0 HEADERS format. Whether we can eliminate it or not in
WS/HTTP/2.0 is one of the most important points to evaluate the plans
employing HTTP/2.0 HEADERS.

Though the ranges the length headers represent differ (HTTP/2.0: 14 bit, WS
length format: 63 bit), it seems everyone can live without a header of
"original WebSocket frame size".

Please reply to this post if you have any objection, but please don't write
objection to use of HTTP/2.0 HEADERS itself in this thread.

Thanks

--f46d043c7de6c5f25104f32b241b
Content-Type: text/html; charset=UTF-8

<div dir="ltr"><div>Thanks all for your comments.</div><div><br></div><div>First, please note that I wanted to discuss the question in the context of WS/HTTP/2.0 layering. Not about redesign of RFC 6455 itself.</div><div><br>

</div>I asked this question since the length header is the biggest field when encoded into HTTP/2.0 HEADERS format. Whether we can eliminate it or not in WS/HTTP/2.0 is one of the most important points to evaluate the plans employing HTTP/2.0 HEADERS.<div>

<br></div><div>Though the ranges the length headers represent differ (HTTP/2.0: 14 bit, WS length format: 63 bit), it seems everyone can live without a header of &quot;original WebSocket frame size&quot;.</div><div><br></div>

<div>Please reply to this post if you have any objection, but please don&#39;t write objection to use of HTTP/2.0 HEADERS itself in this thread.</div><div><br></div><div>Thanks</div></div>

--f46d043c7de6c5f25104f32b241b--


From nobody Mon Feb 24 10:43:19 2014
Return-Path: <joakim@intalio.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 4195B1A0235 for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 10:43:18 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 0.723
X-Spam-Level: 
X-Spam-Status: No, score=0.723 tagged_above=-999 required=5 tests=[BAYES_50=0.8, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_LOW=-0.7] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id cK-Vc1m07Jip for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 10:43:14 -0800 (PST)
Received: from mail-ea0-f171.google.com (mail-ea0-f171.google.com [209.85.215.171]) by ietfa.amsl.com (Postfix) with ESMTP id 3184B1A024E for <hybi@ietf.org>; Mon, 24 Feb 2014 10:43:12 -0800 (PST)
Received: by mail-ea0-f171.google.com with SMTP id f15so3306603eak.2 for <hybi@ietf.org>; Mon, 24 Feb 2014 10:43:11 -0800 (PST)
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=LdhS7OCGymlQkO5fzCGcZtyigv3hEXgqXe7Kxswj7oc=; b=WWZG4mGDPW+dj1ZjjQPMYuSRKb19FwuKvLrie7yADrDu3aHmXdy7y9CSpluaHafZam vRKKgA+vgFImvNocKv2g3ZhNrZ2PBnBiJK6CuA0DscOWj9sW6joj1sQX1Wh1EwAUgVXq CQf+B3d+7YB24letk95zew8cRPxioS+o0PscGAwXIz8bfd3KCWOaYy702URSQDCf8Zrn 57O7CTmqg1+5VK2/QPeQZxfQp7v+AHLnjmOGhV8GDWqT3OVT+71Arvy+sRLqdIaQoHlX PTlSYFLMvxtLlN+Hsz5KeEQgnpcvOMdNzvOy8/a2H/93OGiaX5iS5AZSFI48+CDG8WSv gAeA==
X-Gm-Message-State: ALoCoQknKBCRzrT7L1mUCw3FVetqNtkx6t5IHAEWbAWdqTol9/9oXAxhoH6TBfIw2fFDvQJ08TKw
MIME-Version: 1.0
X-Received: by 10.14.103.67 with SMTP id e43mr25828930eeg.94.1393267391126; Mon, 24 Feb 2014 10:43:11 -0800 (PST)
Received: by 10.15.23.134 with HTTP; Mon, 24 Feb 2014 10:43:11 -0800 (PST)
In-Reply-To: <CAH9hSJY4NResx4DskJM8agd5ZXo9yHELYaXOpWG-xXK8P4+9zw@mail.gmail.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <CACuKZqEcA1Pv8RpWfmThMjTzi2BbVMMKXqujs6BxVfxRPZJ9NQ@mail.gmail.com> <CAG4zZZCr4aTfVpw2coX2g0qw++4kdgNCFVze6tHKZ+fJNqb0aQ@mail.gmail.com> <CAG4zZZA4WR50Ea2e36XEgTgpJURwiMFn+DXeYuxHFfZatYzDOQ@mail.gmail.com> <CAH9hSJYv5VXGjS7AfG12-ArEvE6Uj_eE6pPxAiQcchcgV8vHcg@mail.gmail.com> <CAH9hSJbzra7uz7yfKQwfZaP_jhnxwdZyx8JnwCmBGhiMk6rbtg@mail.gmail.com> <CAG4zZZC1cmaH36znAvVLZE_MyJ+ThNk1Ky7tQ-QnKf_qwiQNDw@mail.gmail.com> <CAG4zZZB3h6TTFLUp-ucYQP7tMf-8=h5fh1UqufMh3w-0JN+hSw@mail.gmail.com> <CAH9hSJaeCrLjkhHUzaaGDw-apiSv-eaPZeYGEHBuwRoF3yPCOA@mail.gmail.com> <CAH9hSJY4NResx4DskJM8agd5ZXo9yHELYaXOpWG-xXK8P4+9zw@mail.gmail.com>
Date: Mon, 24 Feb 2014 11:43:11 -0700
Message-ID: <CAG4zZZC_S-+sALay3h+cD3nUK33_ayef5QC6v--TiAAR3Oc33w@mail.gmail.com>
From: Joakim Erdfelt <joakim@intalio.com>
To: Takeshi Yoshino <tyoshino@google.com>
Content-Type: multipart/alternative; boundary=001a11c289acecee2704f32b579c
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/YgE6bwO8shbC-pZaPDcT10AoJqo
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 24 Feb 2014 18:43:18 -0000

--001a11c289acecee2704f32b579c
Content-Type: text/plain; charset=ISO-8859-1

Let me see if I grok this entirely...

You are saying that HTTP/2.0 HEADERS has a 14 bit value already.
You would like to utilize this existing 14 bit value to represent websocket
payload data.
You would like to avoid having to adjust or supplement HTTP/2.0 HEADERS
just for WebSocket behaviors.
You are indicating that this 14 bit value is not a 1 to 1 representation of
"original WebSocket frame size" as determined by various APIs or extensions
before hitting the HTTP/2.0 layer.
You hint at the behavior of large (over 14 bit length) "original WebSocket
frames" being likely sent in multiple HTTP/2.0 DATA/END_SEGMENT portions.

Wouldn't this also mean that any "original WebSocket frame" length
information not being present on the receiving side. right?
The receiving side would just see HTTP/2.0 DATA/END_SEGMENT with some
indicator that the DATA is for WebSocket, right?
The receiving side would never see the "original WebSocket frame" as this
information is now lost.  Is that correct?
Instead you have HTTP/2.0 DATA length header for WebSocket payload data,
along with an END_SEGMENT indicating the separation between WebSocket
messages.


--
Joakim Erdfelt <joakim@intalio.com>
webtide.com <http://www.webtide.com/> - intalio.com/jetty
Expert advice, services and support from from the Jetty & CometD experts
eclipse.org/jetty - cometd.org


On Mon, Feb 24, 2014 at 11:28 AM, Takeshi Yoshino <tyoshino@google.com>wrote:

> Thanks all for your comments.
>
> First, please note that I wanted to discuss the question in the context of
> WS/HTTP/2.0 layering. Not about redesign of RFC 6455 itself.
>
> I asked this question since the length header is the biggest field when
> encoded into HTTP/2.0 HEADERS format. Whether we can eliminate it or not in
> WS/HTTP/2.0 is one of the most important points to evaluate the plans
> employing HTTP/2.0 HEADERS.
>
> Though the ranges the length headers represent differ (HTTP/2.0: 14 bit,
> WS length format: 63 bit), it seems everyone can live without a header of
> "original WebSocket frame size".
>
> Please reply to this post if you have any objection, but please don't
> write objection to use of HTTP/2.0 HEADERS itself in this thread.
>
> Thanks
>

--001a11c289acecee2704f32b579c
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div>Let me see if I grok this entirely...</div><div><br><=
/div>You are saying that HTTP/2.0 HEADERS has a 14 bit value already. =A0<d=
iv>You would like to utilize this existing 14 bit value to represent websoc=
ket payload data.</div>
<div>You would like to avoid having to adjust or supplement HTTP/2.0 HEADER=
S just for WebSocket behaviors.<br></div><div>You are indicating that this =
14 bit value is not a 1 to 1 representation of &quot;original WebSocket fra=
me size&quot; as determined by various APIs or extensions before hitting th=
e HTTP/2.0 layer.</div>
<div>You hint at the behavior of large (over 14 bit length) &quot;original =
WebSocket frames&quot; being likely sent in multiple HTTP/2.0 DATA/END_SEGM=
ENT portions.</div><div><br></div><div>Wouldn&#39;t this also mean that any=
 &quot;original WebSocket frame&quot; length information not being present =
on the receiving side. right?</div>
<div>The receiving side would just see HTTP/2.0 DATA/END_SEGMENT with some =
indicator that the DATA is for WebSocket, right?</div><div>The receiving si=
de would never see the &quot;original WebSocket frame&quot; as this informa=
tion is now lost. =A0Is that correct?</div>
<div>Instead you have HTTP/2.0 DATA length header for WebSocket payload dat=
a, along with an END_SEGMENT indicating the separation between WebSocket me=
ssages.</div><div><div><br></div></div></div><div class=3D"gmail_extra"><br=
 clear=3D"all">
<div><div dir=3D"ltr"><div>--</div><div>Joakim Erdfelt &lt;<a href=3D"mailt=
o:joakim@intalio.com" target=3D"_blank">joakim@intalio.com</a>&gt;</div><di=
v><a href=3D"http://www.webtide.com/" target=3D"_blank">webtide.com</a>=A0-=
 <a href=3D"http://intalio.com/jetty" target=3D"_blank">intalio.com/jetty</=
a></div>
<div><span>Expert advice, services and support from=A0</span><span>from the=
 Jetty &amp; CometD experts</span></div><div><a href=3D"http://eclipse.org/=
jetty/" target=3D"_blank">eclipse.org/jetty</a>=A0-=A0<a href=3D"http://com=
etd.org/" target=3D"_blank">cometd.org</a></div>
</div></div>
<br><br><div class=3D"gmail_quote">On Mon, Feb 24, 2014 at 11:28 AM, Takesh=
i Yoshino <span dir=3D"ltr">&lt;<a href=3D"mailto:tyoshino@google.com" targ=
et=3D"_blank">tyoshino@google.com</a>&gt;</span> wrote:<br><blockquote clas=
s=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;pad=
ding-left:1ex">
<div dir=3D"ltr"><div>Thanks all for your comments.</div><div><br></div><di=
v>First, please note that I wanted to discuss the question in the context o=
f WS/HTTP/2.0 layering. Not about redesign of RFC 6455 itself.</div><div>
<br>

</div>I asked this question since the length header is the biggest field wh=
en encoded into HTTP/2.0 HEADERS format. Whether we can eliminate it or not=
 in WS/HTTP/2.0 is one of the most important points to evaluate the plans e=
mploying HTTP/2.0 HEADERS.<div>


<br></div><div>Though the ranges the length headers represent differ (HTTP/=
2.0: 14 bit, WS length format: 63 bit), it seems everyone can live without =
a header of &quot;original WebSocket frame size&quot;.</div><div><br></div>


<div>Please reply to this post if you have any objection, but please don&#3=
9;t write objection to use of HTTP/2.0 HEADERS itself in this thread.</div>=
<div><br></div><div>Thanks</div></div>
</blockquote></div><br></div>

--001a11c289acecee2704f32b579c--


From nobody Mon Feb 24 10:59:31 2014
Return-Path: <fenix@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id DC9661A0186 for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 10:59:29 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -0.525
X-Spam-Level: 
X-Spam-Status: No, score=-0.525 tagged_above=-999 required=5 tests=[BAYES_05=-0.5, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.547, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 8Mqy_VRzt82j for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 10:59:26 -0800 (PST)
Received: from mail-qc0-x22f.google.com (mail-qc0-x22f.google.com [IPv6:2607:f8b0:400d:c01::22f]) by ietfa.amsl.com (Postfix) with ESMTP id A9BAA1A0233 for <hybi@ietf.org>; Mon, 24 Feb 2014 10:59:25 -0800 (PST)
Received: by mail-qc0-f175.google.com with SMTP id e16so2502139qcx.20 for <hybi@ietf.org>; Mon, 24 Feb 2014 10:59:24 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=dSCWtjPCZIxJPoCZYwNaXDYcpzisYVugrYJwjysNq+8=; b=TTSnrfX8kEajU1eyDHIDW39jlJ1c1ljqovNHWKjW5gzPu/1dT7QAWITFP5Z0WLRZvT Xd7UJujm0u1ExI7wN/JcjMJ8IK9l9HKml/3jqivAsfq5fH5g076R0MFITYpsdlWsQ8WQ RjejHz4s3ymxfgjPLubFqujmD+cFE1kW2hovEbdmAeKoPQgjKUjkczE15DvAgVnWIaQ7 ucraZ5Xs96Wej8J0iNdY6V14Ao+irzaSAwiEO5ikDD4q7lk6Rg8JRVqSH1E4ig7rMc73 uahJbM1etba2Zql+xSLnO7WJ7d3BaHgUpqIsbwBSPWTwHWq9BDjYFeK4OPm8i3bY+j4V INhw==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=dSCWtjPCZIxJPoCZYwNaXDYcpzisYVugrYJwjysNq+8=; b=cLp982IWL9rv9aCbVKB6jZePD8+B+9NF2eyPp20wBwi2QCOm4/0lSlujHy7YEz1W9O SMQhculJIi6UBRKNIkoHKotobLX2iDSO9xQbtMghjLSgKGQIw/dORFv9+VUAWRJXL8F1 4VgfHEZ9amS3WkXxRbEd7cZgPSpadVKeSqL9cBP0ha/G++GTZsY2Vv7J5KtTKbfCe7eS gZ0/MupLV80Ql6hJQaSXrXLtuEMdYfOJXX7D7v7/xptszpXpVMlVHd3qJEzlVKuJCJ0i WpwrS6J/VwZNFtxd4XbVmuGvYjSL5i3r/HsxjiKuCAnFYJi8aBVVh60f/0yHZVBe7PNt aIMQ==
X-Gm-Message-State: ALoCoQkdMzfm34iwcBs1zHKogTXfe9zI5RMp/gsxsIHzXSmOc27Mt4rH5wubTby60L5YqQB9qyoit8vRsfaAGlmuTAx8P5FTFnAMDzO/wZHYkxx9sOSeANguu4EHFvKDS+UjYzpT8HjS0nHeDUU2AJYdpRFLDwddKfjtzTfAxAsC0mk7d52K3uCsrFwwN7Us0f22rMO2L4x0
MIME-Version: 1.0
X-Received: by 10.140.106.225 with SMTP id e88mr30329374qgf.66.1393268364696;  Mon, 24 Feb 2014 10:59:24 -0800 (PST)
Received: by 10.229.13.7 with HTTP; Mon, 24 Feb 2014 10:59:24 -0800 (PST)
In-Reply-To: <CAG4zZZC_S-+sALay3h+cD3nUK33_ayef5QC6v--TiAAR3Oc33w@mail.gmail.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <CACuKZqEcA1Pv8RpWfmThMjTzi2BbVMMKXqujs6BxVfxRPZJ9NQ@mail.gmail.com> <CAG4zZZCr4aTfVpw2coX2g0qw++4kdgNCFVze6tHKZ+fJNqb0aQ@mail.gmail.com> <CAG4zZZA4WR50Ea2e36XEgTgpJURwiMFn+DXeYuxHFfZatYzDOQ@mail.gmail.com> <CAH9hSJYv5VXGjS7AfG12-ArEvE6Uj_eE6pPxAiQcchcgV8vHcg@mail.gmail.com> <CAH9hSJbzra7uz7yfKQwfZaP_jhnxwdZyx8JnwCmBGhiMk6rbtg@mail.gmail.com> <CAG4zZZC1cmaH36znAvVLZE_MyJ+ThNk1Ky7tQ-QnKf_qwiQNDw@mail.gmail.com> <CAG4zZZB3h6TTFLUp-ucYQP7tMf-8=h5fh1UqufMh3w-0JN+hSw@mail.gmail.com> <CAH9hSJaeCrLjkhHUzaaGDw-apiSv-eaPZeYGEHBuwRoF3yPCOA@mail.gmail.com> <CAH9hSJY4NResx4DskJM8agd5ZXo9yHELYaXOpWG-xXK8P4+9zw@mail.gmail.com> <CAG4zZZC_S-+sALay3h+cD3nUK33_ayef5QC6v--TiAAR3Oc33w@mail.gmail.com>
Date: Mon, 24 Feb 2014 10:59:24 -0800
Message-ID: <CAGzyod6isscjV6TD=2DUo_dTSDRZxz6Mzmr7MbYNATwFPE7HXA@mail.gmail.com>
From: Roberto Peon <fenix@google.com>
To: Joakim Erdfelt <joakim@intalio.com>
Content-Type: multipart/alternative; boundary=001a113a8626f4a12804f32b913d
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/_uVaCz3aLY1Y5Bzw1d3oLLi52W8
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 24 Feb 2014 18:59:30 -0000

--001a113a8626f4a12804f32b913d
Content-Type: text/plain; charset=UTF-8

As I understand it:

If the previous DATA frame on the stream included END_SEGMENT, and the
current DATA frame on the stream includes END_SEGMENT, then the HTTP2 frame
length == the websocket frame length.
This is extremely likely to be the common case.

If the websocket frame needs to be broken up (generally only because it has
a large payload), then some other signaling of the frame length is
necessary, e.g. by using an HTTP/2 HEADERS frame with content-length key.
-=R


On Mon, Feb 24, 2014 at 10:43 AM, Joakim Erdfelt <joakim@intalio.com> wrote:

> Let me see if I grok this entirely...
>
> You are saying that HTTP/2.0 HEADERS has a 14 bit value already.
> You would like to utilize this existing 14 bit value to represent
> websocket payload data.
> You would like to avoid having to adjust or supplement HTTP/2.0 HEADERS
> just for WebSocket behaviors.
> You are indicating that this 14 bit value is not a 1 to 1 representation
> of "original WebSocket frame size" as determined by various APIs or
> extensions before hitting the HTTP/2.0 layer.
> You hint at the behavior of large (over 14 bit length) "original WebSocket
> frames" being likely sent in multiple HTTP/2.0 DATA/END_SEGMENT portions.
>
> Wouldn't this also mean that any "original WebSocket frame" length
> information not being present on the receiving side. right?
> The receiving side would just see HTTP/2.0 DATA/END_SEGMENT with some
> indicator that the DATA is for WebSocket, right?
> The receiving side would never see the "original WebSocket frame" as this
> information is now lost.  Is that correct?
> Instead you have HTTP/2.0 DATA length header for WebSocket payload data,
> along with an END_SEGMENT indicating the separation between WebSocket
> messages.
>
>
> --
> Joakim Erdfelt <joakim@intalio.com>
> webtide.com <http://www.webtide.com/> - intalio.com/jetty
> Expert advice, services and support from from the Jetty & CometD experts
> eclipse.org/jetty - cometd.org
>
>
> On Mon, Feb 24, 2014 at 11:28 AM, Takeshi Yoshino <tyoshino@google.com>wrote:
>
>> Thanks all for your comments.
>>
>> First, please note that I wanted to discuss the question in the context
>> of WS/HTTP/2.0 layering. Not about redesign of RFC 6455 itself.
>>
>> I asked this question since the length header is the biggest field when
>> encoded into HTTP/2.0 HEADERS format. Whether we can eliminate it or not in
>> WS/HTTP/2.0 is one of the most important points to evaluate the plans
>> employing HTTP/2.0 HEADERS.
>>
>> Though the ranges the length headers represent differ (HTTP/2.0: 14 bit,
>> WS length format: 63 bit), it seems everyone can live without a header of
>> "original WebSocket frame size".
>>
>> Please reply to this post if you have any objection, but please don't
>> write objection to use of HTTP/2.0 HEADERS itself in this thread.
>>
>> Thanks
>>
>
>
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi
>
>

--001a113a8626f4a12804f32b913d
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">As I understand it:<div><br><div>If the previous DATA fram=
e on the stream included END_SEGMENT, and the current DATA frame on the str=
eam includes END_SEGMENT, then the HTTP2 frame length =3D=3D the websocket =
frame length.</div>
<div>This is extremely likely to be the common case.</div><div><br></div><d=
iv>If the websocket frame needs to be broken up (generally only because it =
has a large payload), then some other signaling of the frame length is nece=
ssary, e.g. by using an HTTP/2 HEADERS frame with content-length key.</div>
<div>-=3DR</div></div></div><div class=3D"gmail_extra"><br><br><div class=
=3D"gmail_quote">On Mon, Feb 24, 2014 at 10:43 AM, Joakim Erdfelt <span dir=
=3D"ltr">&lt;<a href=3D"mailto:joakim@intalio.com" target=3D"_blank">joakim=
@intalio.com</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div dir=3D"ltr"><div>Let me see if I grok t=
his entirely...</div><div><br></div>You are saying that HTTP/2.0 HEADERS ha=
s a 14 bit value already. =C2=A0<div>
You would like to utilize this existing 14 bit value to represent websocket=
 payload data.</div>
<div>You would like to avoid having to adjust or supplement HTTP/2.0 HEADER=
S just for WebSocket behaviors.<br></div><div>You are indicating that this =
14 bit value is not a 1 to 1 representation of &quot;original WebSocket fra=
me size&quot; as determined by various APIs or extensions before hitting th=
e HTTP/2.0 layer.</div>

<div>You hint at the behavior of large (over 14 bit length) &quot;original =
WebSocket frames&quot; being likely sent in multiple HTTP/2.0 DATA/END_SEGM=
ENT portions.</div><div><br></div><div>Wouldn&#39;t this also mean that any=
 &quot;original WebSocket frame&quot; length information not being present =
on the receiving side. right?</div>

<div>The receiving side would just see HTTP/2.0 DATA/END_SEGMENT with some =
indicator that the DATA is for WebSocket, right?</div><div>The receiving si=
de would never see the &quot;original WebSocket frame&quot; as this informa=
tion is now lost. =C2=A0Is that correct?</div>

<div>Instead you have HTTP/2.0 DATA length header for WebSocket payload dat=
a, along with an END_SEGMENT indicating the separation between WebSocket me=
ssages.</div><div><div><br></div></div></div><div class=3D"gmail_extra">
<div class=3D""><br clear=3D"all">
<div><div dir=3D"ltr"><div>--</div><div>Joakim Erdfelt &lt;<a href=3D"mailt=
o:joakim@intalio.com" target=3D"_blank">joakim@intalio.com</a>&gt;</div><di=
v><a href=3D"http://www.webtide.com/" target=3D"_blank">webtide.com</a>=C2=
=A0- <a href=3D"http://intalio.com/jetty" target=3D"_blank">intalio.com/jet=
ty</a></div>

<div><span>Expert advice, services and support from=C2=A0</span><span>from =
the Jetty &amp; CometD experts</span></div><div><a href=3D"http://eclipse.o=
rg/jetty/" target=3D"_blank">eclipse.org/jetty</a>=C2=A0-=C2=A0<a href=3D"h=
ttp://cometd.org/" target=3D"_blank">cometd.org</a></div>

</div></div>
<br><br></div><div><div class=3D"h5"><div class=3D"gmail_quote">On Mon, Feb=
 24, 2014 at 11:28 AM, Takeshi Yoshino <span dir=3D"ltr">&lt;<a href=3D"mai=
lto:tyoshino@google.com" target=3D"_blank">tyoshino@google.com</a>&gt;</spa=
n> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
<div dir=3D"ltr"><div>Thanks all for your comments.</div><div><br></div><di=
v>First, please note that I wanted to discuss the question in the context o=
f WS/HTTP/2.0 layering. Not about redesign of RFC 6455 itself.</div><div>

<br>

</div>I asked this question since the length header is the biggest field wh=
en encoded into HTTP/2.0 HEADERS format. Whether we can eliminate it or not=
 in WS/HTTP/2.0 is one of the most important points to evaluate the plans e=
mploying HTTP/2.0 HEADERS.<div>



<br></div><div>Though the ranges the length headers represent differ (HTTP/=
2.0: 14 bit, WS length format: 63 bit), it seems everyone can live without =
a header of &quot;original WebSocket frame size&quot;.</div><div><br></div>



<div>Please reply to this post if you have any objection, but please don&#3=
9;t write objection to use of HTTP/2.0 HEADERS itself in this thread.</div>=
<div><br></div><div>Thanks</div></div>
</blockquote></div><br></div></div></div>
<br>_______________________________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
<br></blockquote></div><br></div>

--001a113a8626f4a12804f32b913d--


From nobody Mon Feb 24 12:29:08 2014
Return-Path: <zhong.j.yu@gmail.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 8A2E01A02BB for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 12:29:06 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2
X-Spam-Level: 
X-Spam-Status: No, score=-2 tagged_above=-999 required=5 tests=[BAYES_00=-1.9,  DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FREEMAIL_FROM=0.001, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 20mzBtBMMqP4 for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 12:29:05 -0800 (PST)
Received: from mail-ob0-x22f.google.com (mail-ob0-x22f.google.com [IPv6:2607:f8b0:4003:c01::22f]) by ietfa.amsl.com (Postfix) with ESMTP id 642701A02A0 for <hybi@ietf.org>; Mon, 24 Feb 2014 12:29:05 -0800 (PST)
Received: by mail-ob0-f175.google.com with SMTP id va2so4594626obc.6 for <hybi@ietf.org>; Mon, 24 Feb 2014 12:29:04 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113;  h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=sA6C11eckZhD1bz4dCw1rAJMotrp5hh/GzcRq0jlsrs=; b=f2NibJcdnPI5vCdOlUIMYZhfsgprqen14WvS0X+ISPpgLMBN/C0CXJEP5/G3/Ymcgf WaJH2WsoeVTP4ImuX797zC0W08Ez4GnmSo74JFBPYkh7rse6tuH97OiRIEeVqfJSDP3A oG0xs+2cEwVKsrC7cAk5H+wcZfRsCLpY5obrlWVc468EsViHusaNW5mdYvo12aDawEw9 czE8vQLkPDhL+co/KUuRUshH3kA6D+HOMVcgoiE5EC33GTr+Jp4JsUxz0af64at12rO4 nSZj7kKhjzMny0I4LUgDjYyytcGYe3+RWQmdoRqUYRX1p/jYo0b/1TyKd8uQMx8gxhkF vzOg==
MIME-Version: 1.0
X-Received: by 10.60.70.142 with SMTP id m14mr9067617oeu.65.1393273744726; Mon, 24 Feb 2014 12:29:04 -0800 (PST)
Received: by 10.76.106.162 with HTTP; Mon, 24 Feb 2014 12:29:04 -0800 (PST)
In-Reply-To: <CAH9hSJY4NResx4DskJM8agd5ZXo9yHELYaXOpWG-xXK8P4+9zw@mail.gmail.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <CACuKZqEcA1Pv8RpWfmThMjTzi2BbVMMKXqujs6BxVfxRPZJ9NQ@mail.gmail.com> <CAG4zZZCr4aTfVpw2coX2g0qw++4kdgNCFVze6tHKZ+fJNqb0aQ@mail.gmail.com> <CAG4zZZA4WR50Ea2e36XEgTgpJURwiMFn+DXeYuxHFfZatYzDOQ@mail.gmail.com> <CAH9hSJYv5VXGjS7AfG12-ArEvE6Uj_eE6pPxAiQcchcgV8vHcg@mail.gmail.com> <CAH9hSJbzra7uz7yfKQwfZaP_jhnxwdZyx8JnwCmBGhiMk6rbtg@mail.gmail.com> <CAG4zZZC1cmaH36znAvVLZE_MyJ+ThNk1Ky7tQ-QnKf_qwiQNDw@mail.gmail.com> <CAG4zZZB3h6TTFLUp-ucYQP7tMf-8=h5fh1UqufMh3w-0JN+hSw@mail.gmail.com> <CAH9hSJaeCrLjkhHUzaaGDw-apiSv-eaPZeYGEHBuwRoF3yPCOA@mail.gmail.com> <CAH9hSJY4NResx4DskJM8agd5ZXo9yHELYaXOpWG-xXK8P4+9zw@mail.gmail.com>
Date: Mon, 24 Feb 2014 14:29:04 -0600
Message-ID: <CACuKZqHTr0vufQr5mEHM2__NconW_d6jrzTpD0JKEzsiP09uwg@mail.gmail.com>
From: Zhong Yu <zhong.j.yu@gmail.com>
To: Takeshi Yoshino <tyoshino@google.com>
Content-Type: text/plain; charset=ISO-8859-1
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/r6lAyXELEIsH7jE-tCcTyT-nBXw
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 24 Feb 2014 20:29:06 -0000

It there any problem if we mandate that 1 http frame corresponds to 1
ws frame, and 1 ws frame must reside in 1 http frame? What's the point
of supporting ws frames that need multiple http frames?

On Mon, Feb 24, 2014 at 12:28 PM, Takeshi Yoshino <tyoshino@google.com> wrote:
> Thanks all for your comments.
>
> First, please note that I wanted to discuss the question in the context of
> WS/HTTP/2.0 layering. Not about redesign of RFC 6455 itself.
>
> I asked this question since the length header is the biggest field when
> encoded into HTTP/2.0 HEADERS format. Whether we can eliminate it or not in
> WS/HTTP/2.0 is one of the most important points to evaluate the plans
> employing HTTP/2.0 HEADERS.
>
> Though the ranges the length headers represent differ (HTTP/2.0: 14 bit, WS
> length format: 63 bit), it seems everyone can live without a header of
> "original WebSocket frame size".
>
> Please reply to this post if you have any objection, but please don't write
> objection to use of HTTP/2.0 HEADERS itself in this thread.
>
> Thanks


From nobody Mon Feb 24 12:32:57 2014
Return-Path: <zhong.j.yu@gmail.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id E63841A0202 for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 12:32:55 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2
X-Spam-Level: 
X-Spam-Status: No, score=-2 tagged_above=-999 required=5 tests=[BAYES_00=-1.9,  DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FREEMAIL_FROM=0.001, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id AxeQymEBqfFQ for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 12:32:54 -0800 (PST)
Received: from mail-ob0-x235.google.com (mail-ob0-x235.google.com [IPv6:2607:f8b0:4003:c01::235]) by ietfa.amsl.com (Postfix) with ESMTP id 26C661A02C6 for <hybi@ietf.org>; Mon, 24 Feb 2014 12:32:54 -0800 (PST)
Received: by mail-ob0-f181.google.com with SMTP id wp4so5801892obc.12 for <hybi@ietf.org>; Mon, 24 Feb 2014 12:32:53 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113;  h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=0NGvtvNgY6I+294izdeVx0UyP5HLqnhJEim3sugBTu4=; b=NHZAXI4O3n1He0LbJv4igETLX6DLfU3IgClMX0T5wEWp0GJjl5NHQtthO4j2X8b5ci 8/OEs7SgoAZLyMctTItxULWuP39fx2BWgkUQC7mGQ2ibT/pavn5iV+2qrKyytMNghBP0 ihRQRC9eKYB76rUREy8/11hzT+BjhG4Lfb9K/u2INS5skFGPXmQyKDC1iKPIXFnGN6iG e87TsEZu/kHkKquGKXefu+u7h2fDG3VzfoNLa7m4GePkd2jLRFfiyC/BmG4YnPKCw3Da eHbE1xlhN4zSkhLBTzicH25F/FS/5Lh3y8PtJM1w8tpHrAolSTXk2iTegJTAA/AxOoZ4 vuyg==
MIME-Version: 1.0
X-Received: by 10.182.87.69 with SMTP id v5mr2553544obz.77.1393273973537; Mon, 24 Feb 2014 12:32:53 -0800 (PST)
Received: by 10.76.106.162 with HTTP; Mon, 24 Feb 2014 12:32:53 -0800 (PST)
In-Reply-To: <CAG4zZZC_S-+sALay3h+cD3nUK33_ayef5QC6v--TiAAR3Oc33w@mail.gmail.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <CACuKZqEcA1Pv8RpWfmThMjTzi2BbVMMKXqujs6BxVfxRPZJ9NQ@mail.gmail.com> <CAG4zZZCr4aTfVpw2coX2g0qw++4kdgNCFVze6tHKZ+fJNqb0aQ@mail.gmail.com> <CAG4zZZA4WR50Ea2e36XEgTgpJURwiMFn+DXeYuxHFfZatYzDOQ@mail.gmail.com> <CAH9hSJYv5VXGjS7AfG12-ArEvE6Uj_eE6pPxAiQcchcgV8vHcg@mail.gmail.com> <CAH9hSJbzra7uz7yfKQwfZaP_jhnxwdZyx8JnwCmBGhiMk6rbtg@mail.gmail.com> <CAG4zZZC1cmaH36znAvVLZE_MyJ+ThNk1Ky7tQ-QnKf_qwiQNDw@mail.gmail.com> <CAG4zZZB3h6TTFLUp-ucYQP7tMf-8=h5fh1UqufMh3w-0JN+hSw@mail.gmail.com> <CAH9hSJaeCrLjkhHUzaaGDw-apiSv-eaPZeYGEHBuwRoF3yPCOA@mail.gmail.com> <CAH9hSJY4NResx4DskJM8agd5ZXo9yHELYaXOpWG-xXK8P4+9zw@mail.gmail.com> <CAG4zZZC_S-+sALay3h+cD3nUK33_ayef5QC6v--TiAAR3Oc33w@mail.gmail.com>
Date: Mon, 24 Feb 2014 14:32:53 -0600
Message-ID: <CACuKZqFvUbRh+HeDnOK_VOUP+Ycjk_O-zzjSXFNHMe1-=Vy-MQ@mail.gmail.com>
From: Zhong Yu <zhong.j.yu@gmail.com>
To: Joakim Erdfelt <joakim@intalio.com>
Content-Type: text/plain; charset=ISO-8859-1
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/PABPuD9i32dkzs31UGFcXKdjFz8
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 24 Feb 2014 20:32:56 -0000

On Mon, Feb 24, 2014 at 12:43 PM, Joakim Erdfelt <joakim@intalio.com> wrote:
> Let me see if I grok this entirely...
>
> You are saying that HTTP/2.0 HEADERS has a 14 bit value already.
> You would like to utilize this existing 14 bit value to represent websocket
> payload data.
> You would like to avoid having to adjust or supplement HTTP/2.0 HEADERS just
> for WebSocket behaviors.
> You are indicating that this 14 bit value is not a 1 to 1 representation of
> "original WebSocket frame size" as determined by various APIs or extensions

I never understand why APIs expose the concept of frames. It's like
TCP APIs exposing the concept of TCP segments.

> before hitting the HTTP/2.0 layer.
> You hint at the behavior of large (over 14 bit length) "original WebSocket
> frames" being likely sent in multiple HTTP/2.0 DATA/END_SEGMENT portions.
>
> Wouldn't this also mean that any "original WebSocket frame" length
> information not being present on the receiving side. right?
> The receiving side would just see HTTP/2.0 DATA/END_SEGMENT with some
> indicator that the DATA is for WebSocket, right?
> The receiving side would never see the "original WebSocket frame" as this
> information is now lost.  Is that correct?
> Instead you have HTTP/2.0 DATA length header for WebSocket payload data,
> along with an END_SEGMENT indicating the separation between WebSocket
> messages.
>
>
> --
> Joakim Erdfelt <joakim@intalio.com>
> webtide.com - intalio.com/jetty
> Expert advice, services and support from from the Jetty & CometD experts
> eclipse.org/jetty - cometd.org
>
>
> On Mon, Feb 24, 2014 at 11:28 AM, Takeshi Yoshino <tyoshino@google.com>
> wrote:
>>
>> Thanks all for your comments.
>>
>> First, please note that I wanted to discuss the question in the context of
>> WS/HTTP/2.0 layering. Not about redesign of RFC 6455 itself.
>>
>> I asked this question since the length header is the biggest field when
>> encoded into HTTP/2.0 HEADERS format. Whether we can eliminate it or not in
>> WS/HTTP/2.0 is one of the most important points to evaluate the plans
>> employing HTTP/2.0 HEADERS.
>>
>> Though the ranges the length headers represent differ (HTTP/2.0: 14 bit,
>> WS length format: 63 bit), it seems everyone can live without a header of
>> "original WebSocket frame size".
>>
>> Please reply to this post if you have any objection, but please don't
>> write objection to use of HTTP/2.0 HEADERS itself in this thread.
>>
>> Thanks
>
>


From nobody Mon Feb 24 12:39:16 2014
Return-Path: <fenix@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 393031A02F4 for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 12:39:15 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -0.525
X-Spam-Level: 
X-Spam-Status: No, score=-0.525 tagged_above=-999 required=5 tests=[BAYES_05=-0.5, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.547, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id je88al78Qxop for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 12:39:13 -0800 (PST)
Received: from mail-qg0-x233.google.com (mail-qg0-x233.google.com [IPv6:2607:f8b0:400d:c04::233]) by ietfa.amsl.com (Postfix) with ESMTP id 4EE461A0177 for <hybi@ietf.org>; Mon, 24 Feb 2014 12:39:13 -0800 (PST)
Received: by mail-qg0-f51.google.com with SMTP id q108so16324145qgd.10 for <hybi@ietf.org>; Mon, 24 Feb 2014 12:39:12 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=KMAOWJVhV3cd/VGcChdzQbcS8M70mRG9o6DpnjGUUx4=; b=YW/WiK44QjP/VTyyIJqpH+VHVEeT19XrJy0DOnemofdqvA76DX56nU3PPRLzG4tLmN TxN48Y9Ve3MfK9IoA1qIu27Mz3r7JPzF61/YfkF6pTF8E0RghWx41uB2OdH11P95g15D Z56nJQyeVB6Ba6xeftLYs7UN9JuVDTnHFZd+BjtdgyqQKDivrRQog/jjnH3noEUgy19K bNTq1/sS2lwbX0bbHTh+9kQN1weLcG/Rdt2HYXiwp41HH4Y7wn7o/EZtqFzPcToDtPHV tJE5KfXISPq9o+lF6NOuXV3H3KjXv5u/VsE+QZhMuMLjxl2VZJk0YAGqgIESSStOiyJE 0nlg==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=KMAOWJVhV3cd/VGcChdzQbcS8M70mRG9o6DpnjGUUx4=; b=kqUhSbWe1pdDzEZ4roy9ds2j7dxi9pT0Vug8zFm2RUyAwr9vC4ifDWT05wTTnYvzMT wDELATIWVUErffJAkM+yVpQ0629l24G9iN/vrRwCb/P1JD5M05pxm9uSZzmNDdrk8jSj aN2pojW9TVCDkwp3E9rFChlM0gtopZtozx7N5ZVaeA6w++Px0kzK6g1NxVE2RdnlSbKQ zaRml7ftgukNSipywVLb1oAHSJI5VfiG0p/ZCPIwHk3sMX2HKjdXAxFqAPUh5eeez3o4 DPhtQv1IKzEncEqKQ+tE5KMTtmsKZAuXPBy9Mgvj2tQmoEO02e1L2OmlaWBFy/SHqlML oLMQ==
X-Gm-Message-State: ALoCoQl3qhRRTuXG/cd1Ln0T4k140rhA4wBrutVbyqmsr7MwpQM71hUEkUY5RU5BfD5MYPxZqq77ANK8AHE6dPCfpCvXHZc9pKZhX76LxKSnoDHb7ZMMbTEPcSBCJX22YSjgYCm3SEoS4U1XlKkb+RaB3dYY6k+spKkPwRXhZLp9JuJOhPDzxAaIAZ9/jpXhh2CJX6GoAnYo
MIME-Version: 1.0
X-Received: by 10.224.122.20 with SMTP id j20mr6257158qar.79.1393274352450; Mon, 24 Feb 2014 12:39:12 -0800 (PST)
Received: by 10.229.13.7 with HTTP; Mon, 24 Feb 2014 12:39:12 -0800 (PST)
In-Reply-To: <CACuKZqHTr0vufQr5mEHM2__NconW_d6jrzTpD0JKEzsiP09uwg@mail.gmail.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <CACuKZqEcA1Pv8RpWfmThMjTzi2BbVMMKXqujs6BxVfxRPZJ9NQ@mail.gmail.com> <CAG4zZZCr4aTfVpw2coX2g0qw++4kdgNCFVze6tHKZ+fJNqb0aQ@mail.gmail.com> <CAG4zZZA4WR50Ea2e36XEgTgpJURwiMFn+DXeYuxHFfZatYzDOQ@mail.gmail.com> <CAH9hSJYv5VXGjS7AfG12-ArEvE6Uj_eE6pPxAiQcchcgV8vHcg@mail.gmail.com> <CAH9hSJbzra7uz7yfKQwfZaP_jhnxwdZyx8JnwCmBGhiMk6rbtg@mail.gmail.com> <CAG4zZZC1cmaH36znAvVLZE_MyJ+ThNk1Ky7tQ-QnKf_qwiQNDw@mail.gmail.com> <CAG4zZZB3h6TTFLUp-ucYQP7tMf-8=h5fh1UqufMh3w-0JN+hSw@mail.gmail.com> <CAH9hSJaeCrLjkhHUzaaGDw-apiSv-eaPZeYGEHBuwRoF3yPCOA@mail.gmail.com> <CAH9hSJY4NResx4DskJM8agd5ZXo9yHELYaXOpWG-xXK8P4+9zw@mail.gmail.com> <CACuKZqHTr0vufQr5mEHM2__NconW_d6jrzTpD0JKEzsiP09uwg@mail.gmail.com>
Date: Mon, 24 Feb 2014 12:39:12 -0800
Message-ID: <CAGzyod767Aj5AeNF8yp4+Whm2kVrG_Dg0D7nechpbpa6G36Bqw@mail.gmail.com>
From: Roberto Peon <fenix@google.com>
To: Zhong Yu <zhong.j.yu@gmail.com>
Content-Type: multipart/alternative; boundary=089e0149c186da6e8804f32cf6b9
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/hHqwX-IQbxbUIHLjx5-nfRopPfw
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 24 Feb 2014 20:39:15 -0000

--089e0149c186da6e8804f32cf6b9
Content-Type: text/plain; charset=UTF-8

The browser API doesn't have a limit on the size of a message, so the
layers below that API should probably support it.
It would be incomplete not to, for sure.

As for why someone would do it? Who knows. Perhaps someone is exploiting
the fact that WS stuff never gets put into a cache for security reasons? Or
perhaps because using the API is easier and better expresses the
application semantics, even if they're not doing web browsing? Any of these
seem reasonable. There is no requirement for WS to include only small
messages...

-=R


On Mon, Feb 24, 2014 at 12:29 PM, Zhong Yu <zhong.j.yu@gmail.com> wrote:

> It there any problem if we mandate that 1 http frame corresponds to 1
> ws frame, and 1 ws frame must reside in 1 http frame? What's the point
> of supporting ws frames that need multiple http frames?
>
> On Mon, Feb 24, 2014 at 12:28 PM, Takeshi Yoshino <tyoshino@google.com>
> wrote:
> > Thanks all for your comments.
> >
> > First, please note that I wanted to discuss the question in the context
> of
> > WS/HTTP/2.0 layering. Not about redesign of RFC 6455 itself.
> >
> > I asked this question since the length header is the biggest field when
> > encoded into HTTP/2.0 HEADERS format. Whether we can eliminate it or not
> in
> > WS/HTTP/2.0 is one of the most important points to evaluate the plans
> > employing HTTP/2.0 HEADERS.
> >
> > Though the ranges the length headers represent differ (HTTP/2.0: 14 bit,
> WS
> > length format: 63 bit), it seems everyone can live without a header of
> > "original WebSocket frame size".
> >
> > Please reply to this post if you have any objection, but please don't
> write
> > objection to use of HTTP/2.0 HEADERS itself in this thread.
> >
> > Thanks
>
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi
>

--089e0149c186da6e8804f32cf6b9
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">The browser API doesn&#39;t have a limit on the size of a =
message, so the layers below that API should probably support it.<div>It wo=
uld be incomplete not to, for sure.</div><div><br></div><div>As for why som=
eone would do it? Who knows. Perhaps someone is exploiting the fact that WS=
 stuff never gets put into a cache for security reasons? Or perhaps because=
 using the API is easier and better expresses the application semantics, ev=
en if they&#39;re not doing web browsing? Any of these seem reasonable. The=
re is no requirement for WS to include only small messages...</div>
<div><br></div><div>-=3DR</div></div><div class=3D"gmail_extra"><br><br><di=
v class=3D"gmail_quote">On Mon, Feb 24, 2014 at 12:29 PM, Zhong Yu <span di=
r=3D"ltr">&lt;<a href=3D"mailto:zhong.j.yu@gmail.com" target=3D"_blank">zho=
ng.j.yu@gmail.com</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">It there any problem if we mandate that 1 ht=
tp frame corresponds to 1<br>
ws frame, and 1 ws frame must reside in 1 http frame? What&#39;s the point<=
br>
of supporting ws frames that need multiple http frames?<br>
<div class=3D"HOEnZb"><div class=3D"h5"><br>
On Mon, Feb 24, 2014 at 12:28 PM, Takeshi Yoshino &lt;<a href=3D"mailto:tyo=
shino@google.com">tyoshino@google.com</a>&gt; wrote:<br>
&gt; Thanks all for your comments.<br>
&gt;<br>
&gt; First, please note that I wanted to discuss the question in the contex=
t of<br>
&gt; WS/HTTP/2.0 layering. Not about redesign of RFC 6455 itself.<br>
&gt;<br>
&gt; I asked this question since the length header is the biggest field whe=
n<br>
&gt; encoded into HTTP/2.0 HEADERS format. Whether we can eliminate it or n=
ot in<br>
&gt; WS/HTTP/2.0 is one of the most important points to evaluate the plans<=
br>
&gt; employing HTTP/2.0 HEADERS.<br>
&gt;<br>
&gt; Though the ranges the length headers represent differ (HTTP/2.0: 14 bi=
t, WS<br>
&gt; length format: 63 bit), it seems everyone can live without a header of=
<br>
&gt; &quot;original WebSocket frame size&quot;.<br>
&gt;<br>
&gt; Please reply to this post if you have any objection, but please don&#3=
9;t write<br>
&gt; objection to use of HTTP/2.0 HEADERS itself in this thread.<br>
&gt;<br>
&gt; Thanks<br>
<br>
</div></div><div class=3D"HOEnZb"><div class=3D"h5">_______________________=
________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
</div></div></blockquote></div><br></div>

--089e0149c186da6e8804f32cf6b9--


From nobody Mon Feb 24 12:58:41 2014
Return-Path: <zhong.j.yu@gmail.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 9A7AD1A0280 for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 12:58:38 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2
X-Spam-Level: 
X-Spam-Status: No, score=-2 tagged_above=-999 required=5 tests=[BAYES_00=-1.9,  DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FREEMAIL_FROM=0.001, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id OHlqf5Y46ifJ for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 12:58:36 -0800 (PST)
Received: from mail-ob0-x231.google.com (mail-ob0-x231.google.com [IPv6:2607:f8b0:4003:c01::231]) by ietfa.amsl.com (Postfix) with ESMTP id 8FDF41A0257 for <hybi@ietf.org>; Mon, 24 Feb 2014 12:58:36 -0800 (PST)
Received: by mail-ob0-f177.google.com with SMTP id wp18so7716948obc.36 for <hybi@ietf.org>; Mon, 24 Feb 2014 12:58:35 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113;  h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=sBcM4pv7oIM5DH7MxHm8XYDEHogffi9KDH+8jEezUM4=; b=tC4t2+GZ/a5lxUgUApiuCZFSZlkHoPrDh6BXk2sIl0mzdEIf3wBnq/kj1UDvGMk1Uw EVHh1/jGVucHH8Lj01oCtkFioD8pEkDoSi6miskPwqBfcVOQPkXV+g13xx2K2tYYfmYE O5F4FikFacWZwhlIQxJKKv8E7+/Iynbgc8AV37vGWO1UlUzTdhheeTX1W8acDimx0PjI Ovrmpkm84y5zJJqVsuTW5RUeKrA79CIJzSoso+nQCEsm0LTVF1NWcA5bhIfoUfIGqeWg OeJ8bEdXbYcuiMzLahmCk1DUcYxwESm9tgLByTmwF5cS7jnsDvbbVIM+5kQa6bh2R5N2 S0Wg==
MIME-Version: 1.0
X-Received: by 10.60.174.170 with SMTP id bt10mr23484871oec.47.1393275515870;  Mon, 24 Feb 2014 12:58:35 -0800 (PST)
Received: by 10.76.106.162 with HTTP; Mon, 24 Feb 2014 12:58:35 -0800 (PST)
In-Reply-To: <CAGzyod767Aj5AeNF8yp4+Whm2kVrG_Dg0D7nechpbpa6G36Bqw@mail.gmail.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <CACuKZqEcA1Pv8RpWfmThMjTzi2BbVMMKXqujs6BxVfxRPZJ9NQ@mail.gmail.com> <CAG4zZZCr4aTfVpw2coX2g0qw++4kdgNCFVze6tHKZ+fJNqb0aQ@mail.gmail.com> <CAG4zZZA4WR50Ea2e36XEgTgpJURwiMFn+DXeYuxHFfZatYzDOQ@mail.gmail.com> <CAH9hSJYv5VXGjS7AfG12-ArEvE6Uj_eE6pPxAiQcchcgV8vHcg@mail.gmail.com> <CAH9hSJbzra7uz7yfKQwfZaP_jhnxwdZyx8JnwCmBGhiMk6rbtg@mail.gmail.com> <CAG4zZZC1cmaH36znAvVLZE_MyJ+ThNk1Ky7tQ-QnKf_qwiQNDw@mail.gmail.com> <CAG4zZZB3h6TTFLUp-ucYQP7tMf-8=h5fh1UqufMh3w-0JN+hSw@mail.gmail.com> <CAH9hSJaeCrLjkhHUzaaGDw-apiSv-eaPZeYGEHBuwRoF3yPCOA@mail.gmail.com> <CAH9hSJY4NResx4DskJM8agd5ZXo9yHELYaXOpWG-xXK8P4+9zw@mail.gmail.com> <CACuKZqHTr0vufQr5mEHM2__NconW_d6jrzTpD0JKEzsiP09uwg@mail.gmail.com> <CAGzyod767Aj5AeNF8yp4+Whm2kVrG_Dg0D7nechpbpa6G36Bqw@mail.gmail.com>
Date: Mon, 24 Feb 2014 14:58:35 -0600
Message-ID: <CACuKZqGs+F_Jh7qatwCF7YoZ3qE6rBU5Vd2xzwLbn8eAc6+OtA@mail.gmail.com>
From: Zhong Yu <zhong.j.yu@gmail.com>
To: Roberto Peon <fenix@google.com>
Content-Type: text/plain; charset=ISO-8859-1
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/j_PgGf_u9ddY9klOLUehsfygBJQ
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 24 Feb 2014 20:58:38 -0000

Uh? I was talking about ws frames, not ws messages. We can always
break up big frames into small frames (unless frame boundary has any
semantics to anyone).

On Mon, Feb 24, 2014 at 2:39 PM, Roberto Peon <fenix@google.com> wrote:
> The browser API doesn't have a limit on the size of a message, so the layers
> below that API should probably support it.
> It would be incomplete not to, for sure.
>
> As for why someone would do it? Who knows. Perhaps someone is exploiting the
> fact that WS stuff never gets put into a cache for security reasons? Or
> perhaps because using the API is easier and better expresses the application
> semantics, even if they're not doing web browsing? Any of these seem
> reasonable. There is no requirement for WS to include only small messages...
>
> -=R
>
>
> On Mon, Feb 24, 2014 at 12:29 PM, Zhong Yu <zhong.j.yu@gmail.com> wrote:
>>
>> It there any problem if we mandate that 1 http frame corresponds to 1
>> ws frame, and 1 ws frame must reside in 1 http frame? What's the point
>> of supporting ws frames that need multiple http frames?
>>
>> On Mon, Feb 24, 2014 at 12:28 PM, Takeshi Yoshino <tyoshino@google.com>
>> wrote:
>> > Thanks all for your comments.
>> >
>> > First, please note that I wanted to discuss the question in the context
>> > of
>> > WS/HTTP/2.0 layering. Not about redesign of RFC 6455 itself.
>> >
>> > I asked this question since the length header is the biggest field when
>> > encoded into HTTP/2.0 HEADERS format. Whether we can eliminate it or not
>> > in
>> > WS/HTTP/2.0 is one of the most important points to evaluate the plans
>> > employing HTTP/2.0 HEADERS.
>> >
>> > Though the ranges the length headers represent differ (HTTP/2.0: 14 bit,
>> > WS
>> > length format: 63 bit), it seems everyone can live without a header of
>> > "original WebSocket frame size".
>> >
>> > Please reply to this post if you have any objection, but please don't
>> > write
>> > objection to use of HTTP/2.0 HEADERS itself in this thread.
>> >
>> > Thanks
>>
>> _______________________________________________
>> hybi mailing list
>> hybi@ietf.org
>> https://www.ietf.org/mailman/listinfo/hybi
>
>


From nobody Mon Feb 24 13:24:03 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 660111A02CA for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 13:23:59 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.925
X-Spam-Level: 
X-Spam-Status: No, score=-1.925 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.547, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Pm9Zii0DW_9e for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 13:23:57 -0800 (PST)
Received: from mail-we0-x231.google.com (mail-we0-x231.google.com [IPv6:2a00:1450:400c:c03::231]) by ietfa.amsl.com (Postfix) with ESMTP id 518531A026D for <hybi@ietf.org>; Mon, 24 Feb 2014 13:23:57 -0800 (PST)
Received: by mail-we0-f177.google.com with SMTP id t61so5083669wes.22 for <hybi@ietf.org>; Mon, 24 Feb 2014 13:23:56 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc:content-type; bh=6hLaGy7Lv2NdvMaVDJPs7rHEyZVFFDZbLQ3/eXYbuTA=; b=XBc+YE7iCt+qOFTwV0zi6RdZfC/woYv7hovdbJAWV565RVoHz8X7heLttH+3fS3Q8s an7aA5J2uq6KMGIgNdkJWVH3lmF/iVopENQvWxzpYb0o1MUcwe2ZZmdzIl87kG/MHHg7 tWZazCIqgsJzQlvt9mAGQ3HpRsZjNpHho/mKxkhg2jLpMUWX/kcWw+MYdd1kYhBfTL9t sHQrpFoMa+pWPUVcqV0zXc1SF3yn+39gMnenhWBA4Z0rOdNfF8oLBchI3NacQpXix93Z IfWVBctQhQ+Fin5gvYDfFvre8wlmcHYqpb8hteBfdFl4h5vhVcvYuIddHVf9vym5TmML UNIQ==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc:content-type; bh=6hLaGy7Lv2NdvMaVDJPs7rHEyZVFFDZbLQ3/eXYbuTA=; b=lK9XueKm4mSbmhE1GhwyItboIJJ/vey1LmqcVsQhwUuo2uq0hcJOFm9JYgBRKqyq3v j/RN7OL54hxIXKib2+u3M+TTq4WJy42F336mWgNpp7YxSakY4+p7CDV4Wu/vly5soC3C Dc6cdyHqW7ZjpdGZOmYOvRSE6D7TeuB6bqpsW1zZfgi00/b/2Jt/vUDn1bA4TclPW2wI 8kjj9NhNKxAux+OB+ZffHvVxdx6CACpGbAvSsC65sCDVR+n2TR0bFCjFma+ntnKSjBRj a3hxuKQF6cNtm7NbhWwKE5WFcXd32B6hz5H/HRIHFWh5ab2ohYPztf7NfZ6lKuDNRKVk RTmw==
X-Gm-Message-State: ALoCoQkyAQn4j7vvTu6VDqoWztvs+s96MSfN9h0LqGD6wo99DWioUYaeZns8toDgkTflXM2k4g0Aklgjjy3Tpaet2SCT+movpmMioPyg9CIQ5uGTTqCh4ppwsGf9DsmjBOxq6JPM5Wdnm6Qm68mQ4agmpwUpvkEXALr3jxF4wsrN8SNYjG3FeOHx8wMo9sh/QIawrzmxwPJF
X-Received: by 10.180.107.136 with SMTP id hc8mr16392479wib.11.1393277036240;  Mon, 24 Feb 2014 13:23:56 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Mon, 24 Feb 2014 13:23:36 -0800 (PST)
In-Reply-To: <CACuKZqHTr0vufQr5mEHM2__NconW_d6jrzTpD0JKEzsiP09uwg@mail.gmail.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <CACuKZqEcA1Pv8RpWfmThMjTzi2BbVMMKXqujs6BxVfxRPZJ9NQ@mail.gmail.com> <CAG4zZZCr4aTfVpw2coX2g0qw++4kdgNCFVze6tHKZ+fJNqb0aQ@mail.gmail.com> <CAG4zZZA4WR50Ea2e36XEgTgpJURwiMFn+DXeYuxHFfZatYzDOQ@mail.gmail.com> <CAH9hSJYv5VXGjS7AfG12-ArEvE6Uj_eE6pPxAiQcchcgV8vHcg@mail.gmail.com> <CAH9hSJbzra7uz7yfKQwfZaP_jhnxwdZyx8JnwCmBGhiMk6rbtg@mail.gmail.com> <CAG4zZZC1cmaH36znAvVLZE_MyJ+ThNk1Ky7tQ-QnKf_qwiQNDw@mail.gmail.com> <CAG4zZZB3h6TTFLUp-ucYQP7tMf-8=h5fh1UqufMh3w-0JN+hSw@mail.gmail.com> <CAH9hSJaeCrLjkhHUzaaGDw-apiSv-eaPZeYGEHBuwRoF3yPCOA@mail.gmail.com> <CAH9hSJY4NResx4DskJM8agd5ZXo9yHELYaXOpWG-xXK8P4+9zw@mail.gmail.com> <CACuKZqHTr0vufQr5mEHM2__NconW_d6jrzTpD0JKEzsiP09uwg@mail.gmail.com>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Tue, 25 Feb 2014 06:23:36 +0900
Message-ID: <CAH9hSJaw5YpQ2bsDVZ+AXZSa7PLEzjcXb55Lm7hefZ855=kr9Q@mail.gmail.com>
To: Zhong Yu <zhong.j.yu@gmail.com>
Content-Type: multipart/alternative; boundary=e89a8f3bad45d1bd0704f32d9630
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/Kv_54oWWPGc3B9s0xUGHI1gF2Wc
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 24 Feb 2014 21:23:59 -0000

--e89a8f3bad45d1bd0704f32d9630
Content-Type: text/plain; charset=UTF-8

On Tue, Feb 25, 2014 at 5:29 AM, Zhong Yu <zhong.j.yu@gmail.com> wrote:

> It there any problem if we mandate that 1 http frame corresponds to 1
> ws frame, and 1 ws frame must reside in 1 http frame? What's the point
> of supporting ws frames that need multiple http frames?
>

That's not what I'd like to discuss in this thread. I wanted to get
agreement on that HTTP/2.0 level length + END_SEGMENT is enough regardless
of whether we keep WebSocket frame boundary or not.

So, let's discuss that in another thread.


>
> On Mon, Feb 24, 2014 at 12:28 PM, Takeshi Yoshino <tyoshino@google.com>
> wrote:
> > Thanks all for your comments.
> >
> > First, please note that I wanted to discuss the question in the context
> of
> > WS/HTTP/2.0 layering. Not about redesign of RFC 6455 itself.
> >
> > I asked this question since the length header is the biggest field when
> > encoded into HTTP/2.0 HEADERS format. Whether we can eliminate it or not
> in
> > WS/HTTP/2.0 is one of the most important points to evaluate the plans
> > employing HTTP/2.0 HEADERS.
> >
> > Though the ranges the length headers represent differ (HTTP/2.0: 14 bit,
> WS
> > length format: 63 bit), it seems everyone can live without a header of
> > "original WebSocket frame size".
> >
> > Please reply to this post if you have any objection, but please don't
> write
> > objection to use of HTTP/2.0 HEADERS itself in this thread.
> >
> > Thanks
>

--e89a8f3bad45d1bd0704f32d9630
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote">On T=
ue, Feb 25, 2014 at 5:29 AM, Zhong Yu <span dir=3D"ltr">&lt;<a href=3D"mail=
to:zhong.j.yu@gmail.com" target=3D"_blank">zhong.j.yu@gmail.com</a>&gt;</sp=
an> wrote:<br>

<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">It there any problem if we mandate that 1 ht=
tp frame corresponds to 1<br>
ws frame, and 1 ws frame must reside in 1 http frame? What&#39;s the point<=
br>
of supporting ws frames that need multiple http frames?<br></blockquote><di=
v><br></div><div>That&#39;s not what I&#39;d like to discuss in this thread=
. I wanted to get agreement on that HTTP/2.0 level length + END_SEGMENT is =
enough regardless of whether we keep WebSocket frame boundary or not.</div>

<div><br></div><div>So, let&#39;s discuss that in another thread.</div><div=
>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;b=
order-left:1px #ccc solid;padding-left:1ex">
<div class=3D"HOEnZb"><div class=3D"h5"><br>
On Mon, Feb 24, 2014 at 12:28 PM, Takeshi Yoshino &lt;<a href=3D"mailto:tyo=
shino@google.com">tyoshino@google.com</a>&gt; wrote:<br>
&gt; Thanks all for your comments.<br>
&gt;<br>
&gt; First, please note that I wanted to discuss the question in the contex=
t of<br>
&gt; WS/HTTP/2.0 layering. Not about redesign of RFC 6455 itself.<br>
&gt;<br>
&gt; I asked this question since the length header is the biggest field whe=
n<br>
&gt; encoded into HTTP/2.0 HEADERS format. Whether we can eliminate it or n=
ot in<br>
&gt; WS/HTTP/2.0 is one of the most important points to evaluate the plans<=
br>
&gt; employing HTTP/2.0 HEADERS.<br>
&gt;<br>
&gt; Though the ranges the length headers represent differ (HTTP/2.0: 14 bi=
t, WS<br>
&gt; length format: 63 bit), it seems everyone can live without a header of=
<br>
&gt; &quot;original WebSocket frame size&quot;.<br>
&gt;<br>
&gt; Please reply to this post if you have any objection, but please don&#3=
9;t write<br>
&gt; objection to use of HTTP/2.0 HEADERS itself in this thread.<br>
&gt;<br>
&gt; Thanks<br>
</div></div></blockquote></div><br></div></div>

--e89a8f3bad45d1bd0704f32d9630--


From nobody Mon Feb 24 13:32:10 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 1A2511A026A for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 13:32:10 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.325
X-Spam-Level: 
X-Spam-Status: No, score=-1.325 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, J_CHICKENPOX_47=0.6, RP_MATCHES_RCVD=-0.547, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 7c2mnm_qpbWp for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 13:32:08 -0800 (PST)
Received: from mail-wg0-x231.google.com (mail-wg0-x231.google.com [IPv6:2a00:1450:400c:c00::231]) by ietfa.amsl.com (Postfix) with ESMTP id 2C77E1A026D for <hybi@ietf.org>; Mon, 24 Feb 2014 13:32:07 -0800 (PST)
Received: by mail-wg0-f49.google.com with SMTP id y10so5268048wgg.16 for <hybi@ietf.org>; Mon, 24 Feb 2014 13:32:07 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc:content-type; bh=cBOhQl278ty6L0GYU23sJ73nQsqQUVwyUozCY/cz200=; b=f7VOc96XCpYfliYZSlhQtwlJ48wos5g5OarWzOrLh+L7x2c4eL9bSStk5r4FbdmlFz LGywB9e+n+9HBNo358xXygaMImxvVwuFnsCfQeZSwpnouMD60jS76/agGuKlV/QBKEmM WJc2UYkQXfuyaIWf4G1tnF/ZzHrPI0ix2sCc2rTcmqmqPt/HnV6NpVfDNc4+IHlBNWol 0MnsFYrj+W6Y0iJpICvRgQOjXWz7AtBY6TpxAGS9ZBgKdXQOXzBRboDgeCBKoJ5YYaBh GSoGzCp19ISVsxRkXIME6qoE1ruOavzpdXJS7AMg9pdyrNj73MIcPjqtlXtLV00kqyKE mL1Q==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc:content-type; bh=cBOhQl278ty6L0GYU23sJ73nQsqQUVwyUozCY/cz200=; b=Ic7+oDqdtPeHpN6QvrjqSkNwDUjKwkut+6ttarkERQ/Blgz+pZIAVnmjMPdXnm+XH+ bNZbY/Yk6nV85LrPI6EEuVFHJWFJ9tILFyBgI8gQeLWcmlVjjyDeXrNpd5wg513Noz6R K87Pd1rxvTpKM0G4hFHYoAcd8YilCQ9lZ2UqISvED6bKc/vAkFJcHigbgB4HF40dNZHB hPEyhIsNP0DQZ6vIKNaMsYPcQ+YTIG4WuPeBJ5anl8Hq6Z3RFJ8M7O5+t/PU7tLtYFDU OlfKWpaELLIGRX8+OqcMlOMnFZapS1gDetntPdEqs7CNjWXW2lRCM90ezk1WTkPblcBZ o5Wg==
X-Gm-Message-State: ALoCoQkGUmQ5QWgFCigeN74VYpW/RuXDzOc9d+/nR0jmE69P5BNH8exl1lzaaJnCKEolyTmQjz0Ipn7d6f7zJdXUmIhFVHedzSivTShEIIleqG2c6CJFFo6hWBs2wWiWD4C2+5WtccyD7bXoftQ7IVSXiyXFjhyiOrsUP0IbLEUpXD/JnugDshIQi/430cOmz5qqHpf4HNTa
X-Received: by 10.181.13.11 with SMTP id eu11mr16049709wid.30.1393277526927; Mon, 24 Feb 2014 13:32:06 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Mon, 24 Feb 2014 13:31:46 -0800 (PST)
In-Reply-To: <CAGzyod6isscjV6TD=2DUo_dTSDRZxz6Mzmr7MbYNATwFPE7HXA@mail.gmail.com>
References: <CAH9hSJbf_ABT7ECL9eS=_ADrncX8qBtxZv=uLcdu9_6GUv23Uw@mail.gmail.com> <CACuKZqEcA1Pv8RpWfmThMjTzi2BbVMMKXqujs6BxVfxRPZJ9NQ@mail.gmail.com> <CAG4zZZCr4aTfVpw2coX2g0qw++4kdgNCFVze6tHKZ+fJNqb0aQ@mail.gmail.com> <CAG4zZZA4WR50Ea2e36XEgTgpJURwiMFn+DXeYuxHFfZatYzDOQ@mail.gmail.com> <CAH9hSJYv5VXGjS7AfG12-ArEvE6Uj_eE6pPxAiQcchcgV8vHcg@mail.gmail.com> <CAH9hSJbzra7uz7yfKQwfZaP_jhnxwdZyx8JnwCmBGhiMk6rbtg@mail.gmail.com> <CAG4zZZC1cmaH36znAvVLZE_MyJ+ThNk1Ky7tQ-QnKf_qwiQNDw@mail.gmail.com> <CAG4zZZB3h6TTFLUp-ucYQP7tMf-8=h5fh1UqufMh3w-0JN+hSw@mail.gmail.com> <CAH9hSJaeCrLjkhHUzaaGDw-apiSv-eaPZeYGEHBuwRoF3yPCOA@mail.gmail.com> <CAH9hSJY4NResx4DskJM8agd5ZXo9yHELYaXOpWG-xXK8P4+9zw@mail.gmail.com> <CAG4zZZC_S-+sALay3h+cD3nUK33_ayef5QC6v--TiAAR3Oc33w@mail.gmail.com> <CAGzyod6isscjV6TD=2DUo_dTSDRZxz6Mzmr7MbYNATwFPE7HXA@mail.gmail.com>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Tue, 25 Feb 2014 06:31:46 +0900
Message-ID: <CAH9hSJaPCd83QF7DsgbcbHMwLQg9XwUbFX--Jeq8t322-Crwew@mail.gmail.com>
To: Roberto Peon <fenix@google.com>
Content-Type: multipart/alternative; boundary=f46d043c7de6110fcf04f32db446
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/6LSI61DvH7TaKmjJ2SwldrQaumI
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Is it important to know frame length at the start of frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 24 Feb 2014 21:32:10 -0000

--f46d043c7de6110fcf04f32db446
Content-Type: text/plain; charset=UTF-8

On Tue, Feb 25, 2014 at 3:59 AM, Roberto Peon <fenix@google.com> wrote:

> As I understand it:
>
> If the previous DATA frame on the stream included END_SEGMENT, and the
> current DATA frame on the stream includes END_SEGMENT, then the HTTP2 frame
> length == the websocket frame length.
> This is extremely likely to be the common case.
>
> If the websocket frame needs to be broken up (generally only because it
> has a large payload), then some other signaling of the frame length is
> necessary, e.g. by using an HTTP/2 HEADERS frame with content-length key.
>  -=R
>
>
> On Mon, Feb 24, 2014 at 10:43 AM, Joakim Erdfelt <joakim@intalio.com>wrote:
>
>> Let me see if I grok this entirely...
>>
>> You are saying that HTTP/2.0 HEADERS has a 14 bit value already.
>> You would like to utilize this existing 14 bit value to represent
>> websocket payload data.
>>
>
Yes. Using the 14 bit length combined with the END_SEGMENT signal, we can
represent a block by sequence of HTTP/2.0 DATA frames.

As Roberto said, the simplest case is (for simplicity, padding in DATA
frames is assumed to be 0):

DATA(length=n, END_SEGMENT(2nd bit in flags)=true, Data=payload data of n
bytes)

Where Data field contains one byte to represent FIN, RSV1-3 and opcode. Or,
instead we can send HEADERS before the DATA which contains FIN, RSV1-3 and
opcode header since they would compress well compared to length header.

If payload is large, a sequence

DATA(length=s_0, END_SEGMENT=false, Data=payload data of s_0 bytes),
DATA(length=s_1, END_SEGMENT=false, Data=payload data of s_1 bytes),
...
DATA(length=s_n, END_SEGMENT=true, Data=payload data of s_n bytes)

represents the original frame. Where the first one byte in the
concatenation of Data fields has FIN, RSV1-3 and opcode, or instead we can
send HEADERS before the sequence of DATAs which contains FIN, RSV1-3 and
opcode.


>  You would like to avoid having to adjust or supplement HTTP/2.0 HEADERS
>> just for WebSocket behaviors.
>>
>
Yes. I'm trying to use what we have now as-is.


> You are indicating that this 14 bit value is not a 1 to 1 representation
>> of "original WebSocket frame size" as determined by various APIs or
>> extensions before hitting the HTTP/2.0 layer.
>>
>
Right


>  You hint at the behavior of large (over 14 bit length) "original
>> WebSocket frames" being likely sent in multiple HTTP/2.0 DATA/END_SEGMENT
>> portions.
>>
>
Right


>  Wouldn't this also mean that any "original WebSocket frame" length
>> information not being present on the receiving side. right?
>>
>
 Not present on receiving the first DATA, but will be revealed on receiving
the length header of the last DATA frame (with END_SEGMENT on).


>  The receiving side would just see HTTP/2.0 DATA/END_SEGMENT with some
>> indicator that the DATA is for WebSocket, right?
>>
>
I don't quite understand this question. It's already known that DATAs
contain WebSocket traffic by ":scheme" in the opening handshake.


> The receiving side would never see the "original WebSocket frame" as this
>> information is now lost.  Is that correct?
>>
>
END_SEGMENT tells where the boundary of the frames was.


>  Instead you have HTTP/2.0 DATA length header for WebSocket payload data,
>> along with an END_SEGMENT indicating the separation between WebSocket
>> messages.
>>
>
That depends on another topic. If we separate messages with END_SEGMENT,
then yes. If we separate frames with END_SEGMENT, then no.

I'd like to get consensus on that length header in HTTP/2.0 DATA frame is
sufficient. It's a topic independent from whether we preserve frame
boundary or not.


>
>>
>> --
>> Joakim Erdfelt <joakim@intalio.com>
>> webtide.com <http://www.webtide.com/> - intalio.com/jetty
>> Expert advice, services and support from from the Jetty & CometD experts
>> eclipse.org/jetty - cometd.org
>>
>>
>> On Mon, Feb 24, 2014 at 11:28 AM, Takeshi Yoshino <tyoshino@google.com>wrote:
>>
>>> Thanks all for your comments.
>>>
>>> First, please note that I wanted to discuss the question in the context
>>> of WS/HTTP/2.0 layering. Not about redesign of RFC 6455 itself.
>>>
>>> I asked this question since the length header is the biggest field when
>>> encoded into HTTP/2.0 HEADERS format. Whether we can eliminate it or not in
>>> WS/HTTP/2.0 is one of the most important points to evaluate the plans
>>> employing HTTP/2.0 HEADERS.
>>>
>>> Though the ranges the length headers represent differ (HTTP/2.0: 14 bit,
>>> WS length format: 63 bit), it seems everyone can live without a header of
>>> "original WebSocket frame size".
>>>
>>> Please reply to this post if you have any objection, but please don't
>>> write objection to use of HTTP/2.0 HEADERS itself in this thread.
>>>
>>> Thanks
>>>
>>
>>
>> _______________________________________________
>> hybi mailing list
>> hybi@ietf.org
>> https://www.ietf.org/mailman/listinfo/hybi
>>
>>
>
[1] http://tools.ietf.org/html/draft-hirano-httpbis-websocket-over-http2-00

--f46d043c7de6110fcf04f32db446
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_extra">On Tue, Feb 25, 2014 at 3:59 AM=
, Roberto Peon <span dir=3D"ltr">&lt;<a href=3D"mailto:fenix@google.com" ta=
rget=3D"_blank">fenix@google.com</a>&gt;</span> wrote:<br></div><div class=
=3D"gmail_extra">


<div class=3D"gmail_quote"><blockquote class=3D"gmail_quote" style=3D"margi=
n:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204=
);border-left-style:solid;padding-left:1ex"><div dir=3D"ltr">As I understan=
d it:<div>


<br><div>If the previous DATA frame on the stream included END_SEGMENT, and=
 the current DATA frame on the stream includes END_SEGMENT, then the HTTP2 =
frame length =3D=3D the websocket frame length.</div>
<div>This is extremely likely to be the common case.</div><div><br></div><d=
iv>If the websocket frame needs to be broken up (generally only because it =
has a large payload), then some other signaling of the frame length is nece=
ssary, e.g. by using an HTTP/2 HEADERS frame with content-length key.</div>


<span><font color=3D"#888888">
<div>-=3DR</div></font></span></div></div><div class=3D"gmail_extra"><br><b=
r><div class=3D"gmail_quote"><div><div>On Mon, Feb 24, 2014 at 10:43 AM, Jo=
akim Erdfelt <span dir=3D"ltr">&lt;<a href=3D"mailto:joakim@intalio.com" ta=
rget=3D"_blank">joakim@intalio.com</a>&gt;</span> wrote:<br>



</div></div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0=
.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left-s=
tyle:solid;padding-left:1ex"><div><div><div dir=3D"ltr"><div>Let me see if =
I grok this entirely...</div>


<div><br></div>You are saying that HTTP/2.0 HEADERS has a 14 bit value alre=
ady. =C2=A0<div>
You would like to utilize this existing 14 bit value to represent websocket=
 payload data.</div></div></div></div></blockquote></div></div></blockquote=
><div><br></div><div>Yes. Using the 14 bit length combined with the END_SEG=
MENT signal, we can represent a block by sequence of HTTP/2.0 DATA frames.<=
/div>


<div><br></div><div>As Roberto said, the simplest case is (for simplicity, =
padding in DATA frames is assumed to be 0):</div><div><br></div><div>DATA(l=
ength=3Dn, END_SEGMENT(2nd bit in flags)=3Dtrue, Data=3Dpayload data of n b=
ytes)</div>


<div><br></div><div>Where Data field contains one byte to represent FIN, RS=
V1-3 and opcode. Or, instead we can send HEADERS before the DATA which cont=
ains FIN, RSV1-3 and opcode header since they would compress well compared =
to length header.</div>


<div><br></div><div>If payload is large, a sequence</div><div><br></div><di=
v>DATA(length=3Ds_0, END_SEGMENT=3Dfalse, Data=3Dpayload data of s_0 bytes)=
,<br></div><div>DATA(length=3Ds_1, END_SEGMENT=3Dfalse, Data=3Dpayload data=
 of s_1 bytes),</div>


<div>...</div><div>DATA(length=3Ds_n, END_SEGMENT=3Dtrue, Data=3Dpayload da=
ta of s_n bytes)</div><div><br></div><div>represents the original frame. Wh=
ere the first one byte in the concatenation of Data fields has FIN, RSV1-3 =
and opcode, or instead we can send HEADERS before the sequence of DATAs whi=
ch contains FIN, RSV1-3 and opcode.</div>

<div>=C2=A0<br></div><blockquote class=3D"gmail_quote" style=3D"margin:0px =
0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);bord=
er-left-style:solid;padding-left:1ex">
<div class=3D"gmail_extra"><div class=3D"gmail_quote"><blockquote class=3D"=
gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border=
-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex"><div=
><div>


<div dir=3D"ltr">
<div>You would like to avoid having to adjust or supplement HTTP/2.0 HEADER=
S just for WebSocket behaviors.<br></div></div></div></div></blockquote></d=
iv></div></blockquote><div><br></div><div>Yes. I&#39;m trying to use what w=
e have now as-is.</div>


<div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px =
0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-l=
eft-style:solid;padding-left:1ex"><div class=3D"gmail_extra"><div class=3D"=
gmail_quote">


<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div><div><div dir=3D"ltr"><div></div><div>You are indicat=
ing that this 14 bit value is not a 1 to 1 representation of &quot;original=
 WebSocket frame size&quot; as determined by various APIs or extensions bef=
ore hitting the HTTP/2.0 layer.</div>


</div></div></div></blockquote></div></div></blockquote><div><br></div><div=
>Right</div><div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"mar=
gin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,2=
04);border-left-style:solid;padding-left:1ex">


<div class=3D"gmail_extra"><div class=3D"gmail_quote"><blockquote class=3D"=
gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border=
-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex"><div=
><div>


<div dir=3D"ltr">

<div>You hint at the behavior of large (over 14 bit length) &quot;original =
WebSocket frames&quot; being likely sent in multiple HTTP/2.0 DATA/END_SEGM=
ENT portions.</div></div></div></div></blockquote></div></div></blockquote>


<div><br></div><div>Right</div><div>=C2=A0</div><blockquote class=3D"gmail_=
quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-=
color:rgb(204,204,204);border-left-style:solid;padding-left:1ex"><div class=
=3D"gmail_extra">


<div class=3D"gmail_quote"><blockquote class=3D"gmail_quote" style=3D"margi=
n:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204=
);border-left-style:solid;padding-left:1ex"><div><div><div dir=3D"ltr">
<div>Wouldn&#39;t this also mean that any &quot;original WebSocket frame&qu=
ot; length information not being present on the receiving side. right?<br><=
/div></div></div></div></blockquote></div></div></blockquote><div><br>

</div>
<div>Not present on receiving the first DATA, but will be revealed on recei=
ving the length header of the last DATA frame (with END_SEGMENT on).</div><=
div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0=
px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-le=
ft-style:solid;padding-left:1ex">


<div class=3D"gmail_extra"><div class=3D"gmail_quote"><blockquote class=3D"=
gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border=
-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex"><div=
><div>


<div dir=3D"ltr"><div></div>

<div>The receiving side would just see HTTP/2.0 DATA/END_SEGMENT with some =
indicator that the DATA is for WebSocket, right?</div></div></div></div></b=
lockquote></div></div></blockquote><div><br></div><div>I don&#39;t quite un=
derstand this question. It&#39;s already known that DATAs contain WebSocket=
 traffic by &quot;:scheme&quot; in the opening handshake.</div>


<div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px =
0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-l=
eft-style:solid;padding-left:1ex"><div class=3D"gmail_extra"><div class=3D"=
gmail_quote">


<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div><div><div dir=3D"ltr"><div>The receiving side would n=
ever see the &quot;original WebSocket frame&quot; as this information is no=
w lost. =C2=A0Is that correct?</div>


</div></div></div></blockquote></div></div></blockquote><div><br></div><div=
>END_SEGMENT tells where the boundary of the frames was.</div><div>=C2=A0</=
div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;bor=
der-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:sol=
id;padding-left:1ex">


<div class=3D"gmail_extra"><div class=3D"gmail_quote"><blockquote class=3D"=
gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border=
-left-color:rgb(204,204,204);border-left-style:solid;padding-left:1ex"><div=
><div>


<div dir=3D"ltr">

<div>Instead you have HTTP/2.0 DATA length header for WebSocket payload dat=
a, along with an END_SEGMENT indicating the separation between WebSocket me=
ssages.</div></div></div></div></blockquote></div></div></blockquote><div>

<br></div><div>That depends on another topic. If we separate messages with =
END_SEGMENT, then yes. If we separate frames with END_SEGMENT, then no.</di=
v><div><br></div><div>I&#39;d like to get consensus on that length header i=
n HTTP/2.0 DATA frame is sufficient. It&#39;s a topic independent from whet=
her we preserve frame boundary or not.</div>

<div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px =
0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-l=
eft-style:solid;padding-left:1ex"><div class=3D"gmail_extra"><div class=3D"=
gmail_quote">

<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div><div><div dir=3D"ltr"><div><div><br></div></div></div=
>
<div class=3D"gmail_extra">


<div><br clear=3D"all">
<div><div dir=3D"ltr"><div>--</div><div>Joakim Erdfelt &lt;<a href=3D"mailt=
o:joakim@intalio.com" target=3D"_blank">joakim@intalio.com</a>&gt;</div><di=
v><a href=3D"http://www.webtide.com/" target=3D"_blank">webtide.com</a>=C2=
=A0- <a href=3D"http://intalio.com/jetty" target=3D"_blank">intalio.com/jet=
ty</a></div>




<div><span>Expert advice, services and support from=C2=A0</span><span>from =
the Jetty &amp; CometD experts</span></div><div><a href=3D"http://eclipse.o=
rg/jetty/" target=3D"_blank">eclipse.org/jetty</a>=C2=A0-=C2=A0<a href=3D"h=
ttp://cometd.org/" target=3D"_blank">cometd.org</a></div>




</div></div>
<br><br></div><div><div><div class=3D"gmail_quote">On Mon, Feb 24, 2014 at =
11:28 AM, Takeshi Yoshino <span dir=3D"ltr">&lt;<a href=3D"mailto:tyoshino@=
google.com" target=3D"_blank">tyoshino@google.com</a>&gt;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex">
<div dir=3D"ltr"><div>Thanks all for your comments.</div><div><br></div><di=
v>First, please note that I wanted to discuss the question in the context o=
f WS/HTTP/2.0 layering. Not about redesign of RFC 6455 itself.</div><div>




<br>

</div>I asked this question since the length header is the biggest field wh=
en encoded into HTTP/2.0 HEADERS format. Whether we can eliminate it or not=
 in WS/HTTP/2.0 is one of the most important points to evaluate the plans e=
mploying HTTP/2.0 HEADERS.<div>






<br></div><div>Though the ranges the length headers represent differ (HTTP/=
2.0: 14 bit, WS length format: 63 bit), it seems everyone can live without =
a header of &quot;original WebSocket frame size&quot;.</div><div><br></div>






<div>Please reply to this post if you have any objection, but please don&#3=
9;t write objection to use of HTTP/2.0 HEADERS itself in this thread.</div>=
<div><br></div><div>Thanks</div></div>
</blockquote></div><br></div></div></div>
<br></div></div><div>_______________________________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
<br></div></blockquote></div><br></div>
</blockquote></div><br></div><div class=3D"gmail_extra">[1]=C2=A0<a href=3D=
"http://tools.ietf.org/html/draft-hirano-httpbis-websocket-over-http2-00" t=
arget=3D"_blank">http://tools.ietf.org/html/draft-hirano-httpbis-websocket-=
over-http2-00</a></div>


<div class=3D"gmail_extra"><br></div></div>

--f46d043c7de6110fcf04f32db446--


From nobody Mon Feb 24 14:36:10 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 18C561A02CC for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 14:36:08 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -0.026
X-Spam-Level: 
X-Spam-Status: No, score=-0.026 tagged_above=-999 required=5 tests=[BAYES_40=-0.001, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.547, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id DR4X08pvQg0Q for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 14:36:06 -0800 (PST)
Received: from mail-wi0-x22f.google.com (mail-wi0-x22f.google.com [IPv6:2a00:1450:400c:c05::22f]) by ietfa.amsl.com (Postfix) with ESMTP id BC9B41A02C7 for <hybi@ietf.org>; Mon, 24 Feb 2014 14:36:05 -0800 (PST)
Received: by mail-wi0-f175.google.com with SMTP id hm4so3704464wib.14 for <hybi@ietf.org>; Mon, 24 Feb 2014 14:36:04 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc:content-type; bh=00tbSCkEXcxcqbZL08mB51QKaXnne7JyMVEBjfMfGFw=; b=ePann47Nsid8VQE7e9KjnKjevLLO6Y+q9VECnWtfly57gToVnLgUqmhcX/UbvDlRLd N76oFmcbaa5k0PVyZnHawnJHzNFhoxwPNLV1u6QVPWrKyIHouupaznNlgTFsRNXBmYib 8lJPMIPbh1W+s4HW8MrFgg4k9G+FFJ2hR6dhTO7NxtxgHHSJygJSibsJRFEHRW63QzME 3Of5TJf7A/Qv5FkRwcoX/TSppd4lyujRLg+nIxQicMkE3+53sn6OakX4fy1mdpfGZ5S7 MQ8ZXWWOkUERoO03Pp418s37SvqNa08mysHhoZAI5N5QI1rsc3wb7LwOcuslcMVzn4wv w6mA==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc:content-type; bh=00tbSCkEXcxcqbZL08mB51QKaXnne7JyMVEBjfMfGFw=; b=NS7blqzZmArLt0MXoJlu0kdljRueLhwaY9Wm5uhkMF/g7cUUfHWBpdbwMBN0qZozOC Fi3KGE2lT1QEQ93v3Eq/FixsUuI4VspLVrk7l0WmBce0B05LuqWd2IXtMubVQWgs4m2w J3dT2nDHGWtcjkWfB35zJNFf8C+aCEh0bQK1VwySxu9vngvK/ANit1xIlm/7iHfURrpG x3uBJZRI2zuRagnIn0BB+PPMaaH88j0nmN0EJOKriliWOfoqn0F+ZYzD3CkDzCCDDdVS IytJnnmYhj/r5P/Kj2huJVUb7AeUbdyVkqiAlvZ+OJFqBYQQUcgRtj1bn4EKsqCiNF+Z CG9w==
X-Gm-Message-State: ALoCoQlfaGjohIU+fQtfhIzM/UlUZA4o2fQY8mh8MyM+CK8TwUIbLoFuCVYe/4Hq7Xtr/JaBost8mUqrUf0IfkYgbd9gTxqhf5jloovsZVu5zikI9Fk4gxKj9AQqFFba3burhbW0ml1ZK1nquAwUN/8gui3eJKmCiu5tPkxh7cYVc0+zLVuC7OoODLLatTkIHVuwlpIZ7yxV
X-Received: by 10.194.71.47 with SMTP id r15mr21431077wju.19.1393281364365; Mon, 24 Feb 2014 14:36:04 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Mon, 24 Feb 2014 14:35:44 -0800 (PST)
In-Reply-To: <634914A010D0B943A035D226786325D4446C537686@EXVMBX020-12.exch020.serverdata.net>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <634914A010D0B943A035D226786325D4446C537686@EXVMBX020-12.exch020.serverdata.net>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Tue, 25 Feb 2014 07:35:44 +0900
Message-ID: <CAH9hSJaALTNiC9rUC-qVYKK74_YJA5psvGhbsFegTwaLsg9Dug@mail.gmail.com>
To: Tobias Oberstein <tobias.oberstein@tavendo.de>
Content-Type: multipart/alternative; boundary=047d7bfd0172cbb89404f32e98f9
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/sv5IlvZokma3xZugLpbB3-LCEc4
Cc: "hybi@ietf.org" <hybi@ietf.org>, Yutaka Hirano <yhirano@google.com>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Mon, 24 Feb 2014 22:36:08 -0000

--047d7bfd0172cbb89404f32e98f9
Content-Type: text/plain; charset=UTF-8

On Wed, Feb 19, 2014 at 8:45 PM, Tobias Oberstein <
tobias.oberstein@tavendo.de> wrote:

> I don't agree with that approach. I think HTTP2 should take WebSocket "as
> is", if it wants to act as a transport instead of attempting to redefine
> it. Or define it's own messaging (not bound by WS).
>

I want to understand what you mean by this, Tobias, and maybe Peter.

As one of Chromium developers, I just want to build and use a scalable
transport for the WebSocket API. For this goal, it doesn't matter whether
HTTP/2.0 works as a transparent transport or not. There's no such
implication in the term "WS/HTTP/2.0" at least when used by me.

What Roberto and Martin are suggesting sounds like one to be categorized as
"define it's own messaging". Some of Yutaka's plans, too. What's your
actual position for them. Not the best option for you but not against?

--047d7bfd0172cbb89404f32e98f9
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote">On W=
ed, Feb 19, 2014 at 8:45 PM, Tobias Oberstein <span dir=3D"ltr">&lt;<a href=
=3D"mailto:tobias.oberstein@tavendo.de" target=3D"_blank">tobias.oberstein@=
tavendo.de</a>&gt;</span> wrote:<br>

<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div class=3D"">I don&#39;t agree with that approach. I th=
ink HTTP2 should take WebSocket &quot;as is&quot;, if it wants to act as a =
transport instead of attempting to redefine it. Or define it&#39;s own mess=
aging (not bound by WS).<br>

</div></blockquote><div><br></div><div>I want to understand what you mean b=
y this, Tobias, and maybe Peter.</div><div><br></div><div>As one of Chromiu=
m developers, I just want to build and use a scalable transport for the Web=
Socket API. For this goal, it doesn&#39;t matter whether HTTP/2.0 works as =
a transparent transport or not. There&#39;s no such implication in the term=
 &quot;WS/HTTP/2.0&quot; at least when used by me.</div>

<div><br></div><div>What Roberto and Martin are suggesting sounds like one =
to be categorized as &quot;define it&#39;s own messaging&quot;. Some of Yut=
aka&#39;s plans, too. What&#39;s your actual position for them. Not the bes=
t option for you but not against?</div>

<div>=C2=A0</div></div></div></div>

--047d7bfd0172cbb89404f32e98f9--


From nobody Mon Feb 24 20:59:01 2014
Return-Path: <yhirano@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 706F01A02D3 for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 20:58:56 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 1.375
X-Spam-Level: *
X-Spam-Status: No, score=1.375 tagged_above=-999 required=5 tests=[BAYES_50=0.8, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, J_CHICKENPOX_26=0.6, RP_MATCHES_RCVD=-0.547, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id O2EDLRZPAM8T for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 20:58:55 -0800 (PST)
Received: from mail-yk0-x230.google.com (mail-yk0-x230.google.com [IPv6:2607:f8b0:4002:c07::230]) by ietfa.amsl.com (Postfix) with ESMTP id 094BB1A028F for <hybi@ietf.org>; Mon, 24 Feb 2014 20:58:54 -0800 (PST)
Received: by mail-yk0-f176.google.com with SMTP id 19so17351039ykq.7 for <hybi@ietf.org>; Mon, 24 Feb 2014 20:58:54 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=19v1WnZnliBJBw/BJW/8ol8szh4hJZ/Nrkc8gk5wt0Q=; b=P///O4twHBWnxpI2mbVX2RjLU6S7PXqflDIR/xhSFqnIFDY4dMx9WVpY3l0CSVbEw+ fn/iuggr1Fd9qvlyPrExN0MZ+XhCQ9T93Ib7dj6pjaDYRf+AixZtjIx/Q4wMUUaZFCGH bXZqJXO/EUd8hZuon2o7QPVDEAAiXr5VJqK4YKKxZx8Hzwy6rinDYnfzp5hqHlbiBFcd DgRr0fVn0r+SMPd0Q7qb5/U7cizhzqDwnptHjKJBnRYgnsgG1OEepJ23UYxwqsbHJxPf epj9NX1UYoB6YJJRU6SxWZ5zg6Ss4Tulbcj4rCB2jLaZmb3TQEvhA7upPiXg2Oomni/i 7x+g==
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=chromium.org; s=google; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=19v1WnZnliBJBw/BJW/8ol8szh4hJZ/Nrkc8gk5wt0Q=; b=Kpo1LINhDjsWTlY1TAb3pd3ol7w6QgHzQzPaMGlJWUc2lGGMNWgNPsvipcsfThRRio EYG64+beovxdhtFG8b31P2xuNZbW1CBKoLYU7tY14eD28nmfnJI4YmR+w/BqgFVaZywF EvcM4PxtG/TeltCP30Z0vZ0M5lCtVDWLv8Jak=
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:sender:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=19v1WnZnliBJBw/BJW/8ol8szh4hJZ/Nrkc8gk5wt0Q=; b=caPhlswknwboBXzEkyslpLxWJf9jWtItXwqhidtGFazmzeTMDhSCF5Z9dX1uNyvlS7 gHXZqDlkGmQFLhyy6MDUc9x0j+Zpnrhgt9Z6t6ApNtyJxDwAsHyCwzGHUy3EIJoyTfNr pLD2DIAZImNgbKRxbpDNQ2CaPPZSqxKxAoAea4eiB14w6ogAdPvFWKyCDyHr470WqrGL njL/hWRcuS42wElupOfE8YbHNSnMEt2+TFeXLPgl0AUHMgFhXILnGiUDBNss8/owKUNL ifx5XM0GuWE8Sm+AT6lj+tKM+/1OdNgUJm1OdojBzXU8GofFRlB1SiOqz18XL2topYcK hGJA==
X-Gm-Message-State: ALoCoQkjyg7siRwJLoHzg/yEQeLxYabmG47aCazQPFj2j26Bon8fgKQhbWaY0M4dnG9fGe/C68ubyMPebkRXmrWN3I6EVXbs9eaX4ERH7hs+6pov5jP0fHHr6FLDJkBi+k29YfFGSKeF6G1GPrqG1JP5j/2OaZt+p+cD/E7EM+ygaAZ/MoO6lEWtr6cmWwnSpp7E/okJTVI8
MIME-Version: 1.0
X-Received: by 10.236.221.167 with SMTP id r37mr3298552yhp.85.1393304334050; Mon, 24 Feb 2014 20:58:54 -0800 (PST)
Sender: yhirano@google.com
Received: by 10.170.164.87 with HTTP; Mon, 24 Feb 2014 20:58:53 -0800 (PST)
In-Reply-To: <CAHixhFq=wfmYH8-ij_WtsQLN=NUTJwRQ=k8jCPepQDM8V8ZZYA@mail.gmail.com>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <CAH9hSJbBmvNPBSSAk-khdWXgWw0GTt0FG3KsdzYeJcfiAPDk0A@mail.gmail.com> <CAHixhFq=wfmYH8-ij_WtsQLN=NUTJwRQ=k8jCPepQDM8V8ZZYA@mail.gmail.com>
Date: Tue, 25 Feb 2014 13:58:53 +0900
X-Google-Sender-Auth: icurPzXmWGkNfuZh7nMoyab2sPY
Message-ID: <CABihn6EN7V6XEwf6NWn78orxvr3XjGHxROJC4JjQ6RYYKEeCug@mail.gmail.com>
From: Yutaka Hirano <yhirano@chromium.org>
To: Adam Rice <ricea@google.com>
Content-Type: multipart/alternative; boundary=001a11c2ddaee557fc04f333f12c
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/R0htFKeRqHW3labElc0U8H_E-y8
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 25 Feb 2014 04:58:56 -0000

--001a11c2ddaee557fc04f333f12c
Content-Type: text/plain; charset=UTF-8

>
> * Close: Can be used to perform a "clean" close in the middle of sending a
> message. However, I cannot think of any applications that would consider a
> situation in which a in-transit message was aborted to close the connection
> to be a "clean close" scenario.

An endpoint should be able to cancel the ongoing transmission, i.e. The
close call should abort transferring very_very_long_message in the
following code.


ws = new WebSocket(...);
ws.onopen = function() {
  ws.send(very_very_long_message);
  ws.close();
};

It is possible to _Close the WebSocket connection_ without the closing
handshake in such cases, but since the WebSocket API doesn't expose the
transmission status, doing so makes the closing handshake completely
unreliable.



On Wed, Feb 19, 2014 at 4:09 PM, Adam Rice <ricea@google.com> wrote:

> With the current set of control frames, I would not miss the ability to
> interject them in the middle of a data message.
>
> Breaking them down one by one:
>
> * Pong: useless in the middle of a data message (the peer already knows
> that it is receiving data, and useless for RTT measurement because it has
> an unknown amount of OS queueing and buffer bloat in front of it).
> * Ping: almost always useless in the middle of a data message. It could be
> useful if, for example, you had a middle box which closed "idle"
> connections, and considered any connection "idle" if there was no TCP/IP
> payload traffic originating from the client, and the server was in the
> middle of sending a huge message. However, such a middle box would also
> break HTTP downloads so seems unlikely to exist in practice. Even if it
> existed, how would the server detect it to know that it needed to send a
> Ping?
> * Close: Can be used to perform a "clean" close in the middle of sending a
> message. However, I cannot think of any applications that would consider a
> situation in which a in-transit message was aborted to close the connection
> to be a "clean close" scenario.
>
>
> On 19 February 2014 13:24, Takeshi Yoshino <tyoshino@google.com> wrote:
>
>> I'm inclined to say "Yes" for this question so said yes for #2 of
>> Yutaka's original question [1], but I could also cope with if we decide to
>> drop this.
>>
>> [1] http://www.ietf.org/mail-archive/web/hybi/current/msg10403.html
>>
>> _______________________________________________
>> hybi mailing list
>> hybi@ietf.org
>> https://www.ietf.org/mailman/listinfo/hybi
>>
>>
>

--001a11c2ddaee557fc04f333f12c
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px =
0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-l=
eft-style:solid;padding-left:1ex"><span style=3D"font-family:arial,sans-ser=
if;font-size:13px">* Close: Can be used to perform a &quot;clean&quot; clos=
e in the middle of sending a message. However, I cannot think of any applic=
ations that would consider a situation in which a in-transit message was ab=
orted to close the connection to be a &quot;clean close&quot; scenario.</sp=
an></blockquote>
<div>An endpoint should be able to cancel the ongoing transmission, i.e. Th=
e close call should abort transferring very_very_long_message in the follow=
ing code.</div><div><blockquote style=3D"margin:0 0 0 40px;border:none;padd=
ing:0px">
<div><br></div><div>ws =3D new WebSocket(...);</div><div>ws.onopen =3D func=
tion() {</div><div>=C2=A0 ws.send(very_very_long_message);</div><div>=C2=A0=
 ws.close();</div><div>};</div><div><br></div></blockquote>It is possible t=
o _Close the WebSocket connection_ without the closing handshake in such ca=
ses, but since the WebSocket API doesn&#39;t expose the transmission status=
, doing so makes the closing handshake completely unreliable.</div>
<div><br></div></div><div class=3D"gmail_extra"><br><br><div class=3D"gmail=
_quote">On Wed, Feb 19, 2014 at 4:09 PM, Adam Rice <span dir=3D"ltr">&lt;<a=
 href=3D"mailto:ricea@google.com" target=3D"_blank">ricea@google.com</a>&gt=
;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div dir=3D"ltr">With the current set of con=
trol frames, I would not miss the ability to interject them in the middle o=
f a data message.<div>
<br></div><div>Breaking them down one by one:</div><div><br></div><div>* Po=
ng: useless in the middle of a data message (the peer already knows that it=
 is receiving data, and useless for RTT measurement because it has an unkno=
wn amount of OS queueing and buffer bloat in front of it).</div>

<div>* Ping: almost always useless in the middle of a data message. It coul=
d be useful if, for example, you had a middle box which closed &quot;idle&q=
uot; connections, and considered any connection &quot;idle&quot; if there w=
as no TCP/IP payload traffic originating from the client, and the server wa=
s in the middle of sending a huge message. However, such a middle box would=
 also break HTTP downloads so seems unlikely to exist in practice. Even if =
it existed, how would the server detect it to know that it needed to send a=
 Ping?</div>

<div>* Close: Can be used to perform a &quot;clean&quot; close in the middl=
e of sending a message. However, I cannot think of any applications that wo=
uld consider a situation in which a in-transit message was aborted to close=
 the connection to be a &quot;clean close&quot; scenario.</div>

</div><div class=3D"gmail_extra"><br><br><div class=3D"gmail_quote"><div cl=
ass=3D"">On 19 February 2014 13:24, Takeshi Yoshino <span dir=3D"ltr">&lt;<=
a href=3D"mailto:tyoshino@google.com" target=3D"_blank">tyoshino@google.com=
</a>&gt;</span> wrote:<br>

</div><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-l=
eft:1px #ccc solid;padding-left:1ex"><div class=3D""><div dir=3D"ltr"><div =
class=3D"gmail_extra">I&#39;m inclined to say &quot;Yes&quot; for this ques=
tion so said yes for #2 of Yutaka&#39;s original question [1], but I could =
also cope with if we decide to drop this.</div>

<div class=3D"gmail_extra">

<br></div><div class=3D"gmail_extra">[1]=C2=A0<a href=3D"http://www.ietf.or=
g/mail-archive/web/hybi/current/msg10403.html" target=3D"_blank">http://www=
.ietf.org/mail-archive/web/hybi/current/msg10403.html</a></div></div>
<br></div>_______________________________________________<br>
hybi mailing list<br>
<a href=3D"mailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><br>
<a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blank">ht=
tps://www.ietf.org/mailman/listinfo/hybi</a><br>
<br></blockquote></div><br></div>
</blockquote></div><br></div>

--001a11c2ddaee557fc04f333f12c--


From nobody Mon Feb 24 21:41:00 2014
Return-Path: <zhong.j.yu@gmail.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id B6A791A044B for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 21:40:58 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 1.3
X-Spam-Level: *
X-Spam-Status: No, score=1.3 tagged_above=-999 required=5 tests=[BAYES_50=0.8,  DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FREEMAIL_FROM=0.001, J_CHICKENPOX_26=0.6, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id zMqJ43dhmavj for <hybi@ietfa.amsl.com>; Mon, 24 Feb 2014 21:40:57 -0800 (PST)
Received: from mail-ob0-x22b.google.com (mail-ob0-x22b.google.com [IPv6:2607:f8b0:4003:c01::22b]) by ietfa.amsl.com (Postfix) with ESMTP id 1A0801A0274 for <hybi@ietf.org>; Mon, 24 Feb 2014 21:40:57 -0800 (PST)
Received: by mail-ob0-f171.google.com with SMTP id vb8so8027635obc.16 for <hybi@ietf.org>; Mon, 24 Feb 2014 21:40:56 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113;  h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=EhtwAVphpxSwu9ayany5RH4SqM4xJ9o4J39RaEoFkt0=; b=T/RmElgJHbczrHy/Ardxch4FOl28qZD/QIqucrtL9urWEdkk7UdOlr9IqfY/Uw2xk2 gd+CF5Suwd2uyxMvIvxN07mmbQwDa6W1XZU3LBBfQ/gTH8PvYPt4pc0nmxUCoZRCR7vB IXnn3357zhV6ybMmohmZ9t9bgKsVanZRzcZUY4lRh4I0Ut8iHiyVS8wHqUsVBmbLXSfC +HsS2vB6DQFjx08UNhFjQojFpT730uylr9KxS3oI/pBlNUQWLshT9MQCVX5Pf/JApPn7 HfGB55PkTzXKSc6lFsXT5iu4RLWfyHL716MWqqpyQDPbfYvnkD6Sf5lgSLW2Wo1MkvZ2 cLlQ==
MIME-Version: 1.0
X-Received: by 10.183.3.102 with SMTP id bv6mr21091102obd.18.1393306856304; Mon, 24 Feb 2014 21:40:56 -0800 (PST)
Received: by 10.76.106.162 with HTTP; Mon, 24 Feb 2014 21:40:56 -0800 (PST)
In-Reply-To: <CABihn6EN7V6XEwf6NWn78orxvr3XjGHxROJC4JjQ6RYYKEeCug@mail.gmail.com>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <CAH9hSJbBmvNPBSSAk-khdWXgWw0GTt0FG3KsdzYeJcfiAPDk0A@mail.gmail.com> <CAHixhFq=wfmYH8-ij_WtsQLN=NUTJwRQ=k8jCPepQDM8V8ZZYA@mail.gmail.com> <CABihn6EN7V6XEwf6NWn78orxvr3XjGHxROJC4JjQ6RYYKEeCug@mail.gmail.com>
Date: Mon, 24 Feb 2014 23:40:56 -0600
Message-ID: <CACuKZqHNoR5GQmWyzbXAszZCOT2P4pjSmT3SF6ZG3X7hTY=1xw@mail.gmail.com>
From: Zhong Yu <zhong.j.yu@gmail.com>
To: Yutaka Hirano <yhirano@chromium.org>
Content-Type: text/plain; charset=ISO-8859-1
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/Qc4h0nv89TCoH4ADylqjQEATgIo
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 25 Feb 2014 05:40:59 -0000

On Mon, Feb 24, 2014 at 10:58 PM, Yutaka Hirano <yhirano@chromium.org> wrote:
>> * Close: Can be used to perform a "clean" close in the middle of sending a
>> message. However, I cannot think of any applications that would consider a
>> situation in which a in-transit message was aborted to close the connection
>> to be a "clean close" scenario.
>
> An endpoint should be able to cancel the ongoing transmission, i.e. The
> close call should abort transferring very_very_long_message in the following
> code.
>
>
> ws = new WebSocket(...);
> ws.onopen = function() {
>   ws.send(very_very_long_message);
>   ws.close();
> };

Shouldn't ws.close() allow queued messages to be flushed before the
actual closing?

>
> It is possible to _Close the WebSocket connection_ without the closing
> handshake in such cases, but since the WebSocket API doesn't expose the
> transmission status, doing so makes the closing handshake completely
> unreliable.
>
>
>
> On Wed, Feb 19, 2014 at 4:09 PM, Adam Rice <ricea@google.com> wrote:
>>
>> With the current set of control frames, I would not miss the ability to
>> interject them in the middle of a data message.
>>
>> Breaking them down one by one:
>>
>> * Pong: useless in the middle of a data message (the peer already knows
>> that it is receiving data, and useless for RTT measurement because it has an
>> unknown amount of OS queueing and buffer bloat in front of it).
>> * Ping: almost always useless in the middle of a data message. It could be
>> useful if, for example, you had a middle box which closed "idle"
>> connections, and considered any connection "idle" if there was no TCP/IP
>> payload traffic originating from the client, and the server was in the
>> middle of sending a huge message. However, such a middle box would also
>> break HTTP downloads so seems unlikely to exist in practice. Even if it
>> existed, how would the server detect it to know that it needed to send a
>> Ping?
>> * Close: Can be used to perform a "clean" close in the middle of sending a
>> message. However, I cannot think of any applications that would consider a
>> situation in which a in-transit message was aborted to close the connection
>> to be a "clean close" scenario.
>>
>>
>> On 19 February 2014 13:24, Takeshi Yoshino <tyoshino@google.com> wrote:
>>>
>>> I'm inclined to say "Yes" for this question so said yes for #2 of
>>> Yutaka's original question [1], but I could also cope with if we decide to
>>> drop this.
>>>
>>> [1] http://www.ietf.org/mail-archive/web/hybi/current/msg10403.html
>>>
>>> _______________________________________________
>>> hybi mailing list
>>> hybi@ietf.org
>>> https://www.ietf.org/mailman/listinfo/hybi
>>>
>>
>
>
> _______________________________________________
> hybi mailing list
> hybi@ietf.org
> https://www.ietf.org/mailman/listinfo/hybi
>


From nobody Tue Feb 25 00:27:20 2014
Return-Path: <yhirano@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 6D1011A016B for <hybi@ietfa.amsl.com>; Tue, 25 Feb 2014 00:27:19 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 1.375
X-Spam-Level: *
X-Spam-Status: No, score=1.375 tagged_above=-999 required=5 tests=[BAYES_50=0.8, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, J_CHICKENPOX_26=0.6, RP_MATCHES_RCVD=-0.547, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 585KCWjRNSwH for <hybi@ietfa.amsl.com>; Tue, 25 Feb 2014 00:27:17 -0800 (PST)
Received: from mail-yk0-x229.google.com (mail-yk0-x229.google.com [IPv6:2607:f8b0:4002:c07::229]) by ietfa.amsl.com (Postfix) with ESMTP id 0DFA01A035C for <hybi@ietf.org>; Tue, 25 Feb 2014 00:27:16 -0800 (PST)
Received: by mail-yk0-f169.google.com with SMTP id 142so17792628ykq.0 for <hybi@ietf.org>; Tue, 25 Feb 2014 00:27:16 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=QmMFoZQ4ezmuTS/aHVnODnrvK+e1RpeMSmT/voWwU2o=; b=LAzDn+D1PmJ+0Be5+h0SVqXKGmboFKXRi3hLqMbusZUmvvly2SVXOqhiFKcelLGvmU sdhitvYqFBquGfcp+5lo/wLm+blfEeqKT+rsmSEk/UNjgiZL/dyS4Nw6AjFxRwp+mYR+ OsPoQkc8rVxPP8sETQMQ3q/WCYDLlDWS00HMPSVku8RzHzLfX/KiUZaqPZqEGzx2BZZJ neGkUrQ+m9h4WCzGZg9G8blDvglkHR3i6p4d4FkL2vRMEQ0wYYsTymZNzwtUdIVYch6D xDFiPynlzbyvSuJ5EKLcP6vikPpB7Vkkohb511FMg9Q+VIXJ3KYIZPLvfxgS19nknP87 FbKg==
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=chromium.org; s=google; h=mime-version:sender:in-reply-to:references:date:message-id:subject :from:to:cc:content-type; bh=QmMFoZQ4ezmuTS/aHVnODnrvK+e1RpeMSmT/voWwU2o=; b=OH7NJTgqhR6+DUBEq6IvKwGoAtwEeiUMxNJZHUonGMtKtXLKnHxXgMpQ3rNVCYw/Am Y2jsBMXMP50uqWY20uAlr+XPRZDLI2o8SxhpKOvwu7RrblV6s9Iacp4rTBPk1vejTnLj Q/kWW9uNOD4vdb2JNK4f1V8Pm+T+cyWSBinis=
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:sender:in-reply-to:references:date :message-id:subject:from:to:cc:content-type; bh=QmMFoZQ4ezmuTS/aHVnODnrvK+e1RpeMSmT/voWwU2o=; b=fD0WxOdalYP69u55cIJ6Cp2mMjA+UiVX7zcUSsazgQWR8MfMgOWBkKNGQV7GCCeVcb qlSXlIX771vogqX5pM8f0l6Q9cz/WOQyK8zX2/u9PX2Mar4hXmjf66FAk+SoUw8/xMrB DhPSe60nrWF9LckcK6xDo4jVhoIjws6mwGPxaGXTaxre9Z4bs6oMVZ2SV651fBIRd/Me Ww4fHgGtFr88U7PUeQp8cS2HkYnWANrgI6GS9vHXqx040zJX+ARIDSArtI9T5QHxwu3j W+L6pgun2z4SjKr6jLoY8P9cH0X3HgzfYl9PHbPoKngiH8cl5hitLfi0Y6kehjujm80/ F7iw==
X-Gm-Message-State: ALoCoQmpEFj8IVeDBiqUetXxHrwfmADclgHcoJ3HoGX5Xw2WItsFHYZJwIVdP7WGTYAgdBm3JQyD58zphaYu+Dj++OOguc2NE2BUMNx3hEO5ny8hoJZ5bug7SgYjp0h2FE0AjkJUlJHzXtbaNaBKXQlV8DskwiEN5o+xMwcLd5pQ2lt4WXAQ3KMoSTqiZ217YDT33U6k16IF
MIME-Version: 1.0
X-Received: by 10.236.231.234 with SMTP id l100mr83337yhq.135.1393316835803; Tue, 25 Feb 2014 00:27:15 -0800 (PST)
Sender: yhirano@google.com
Received: by 10.170.164.87 with HTTP; Tue, 25 Feb 2014 00:27:15 -0800 (PST)
In-Reply-To: <CACuKZqHNoR5GQmWyzbXAszZCOT2P4pjSmT3SF6ZG3X7hTY=1xw@mail.gmail.com>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <CAH9hSJbBmvNPBSSAk-khdWXgWw0GTt0FG3KsdzYeJcfiAPDk0A@mail.gmail.com> <CAHixhFq=wfmYH8-ij_WtsQLN=NUTJwRQ=k8jCPepQDM8V8ZZYA@mail.gmail.com> <CABihn6EN7V6XEwf6NWn78orxvr3XjGHxROJC4JjQ6RYYKEeCug@mail.gmail.com> <CACuKZqHNoR5GQmWyzbXAszZCOT2P4pjSmT3SF6ZG3X7hTY=1xw@mail.gmail.com>
Date: Tue, 25 Feb 2014 17:27:15 +0900
X-Google-Sender-Auth: IQEyQNBFBOnEUH4bTF5webYmsT8
Message-ID: <CABihn6GC4VM2AHza-F7ML=FfHLZu7FNqx+BhbuVsfJLWk0P92w@mail.gmail.com>
From: Yutaka Hirano <yhirano@chromium.org>
To: Zhong Yu <zhong.j.yu@gmail.com>
Content-Type: multipart/alternative; boundary=089e0160a4c80efd8404f336dbba
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/VyWTsWbJbdQU2YK8rBWAXLfp4VY
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 25 Feb 2014 08:27:19 -0000

--089e0160a4c80efd8404f336dbba
Content-Type: text/plain; charset=UTF-8

>
> Shouldn't ws.close() allow queued messages to be flushed before the
> actual closing?

It is allowed to wait until the endpoint finishes sending the message.
Is it the desired behavior? I thought it is undesired, but I'm not
confident now.

On Tue, Feb 25, 2014 at 2:40 PM, Zhong Yu <zhong.j.yu@gmail.com> wrote:

> On Mon, Feb 24, 2014 at 10:58 PM, Yutaka Hirano <yhirano@chromium.org>
> wrote:
> >> * Close: Can be used to perform a "clean" close in the middle of
> sending a
> >> message. However, I cannot think of any applications that would
> consider a
> >> situation in which a in-transit message was aborted to close the
> connection
> >> to be a "clean close" scenario.
> >
> > An endpoint should be able to cancel the ongoing transmission, i.e. The
> > close call should abort transferring very_very_long_message in the
> following
> > code.
> >
> >
> > ws = new WebSocket(...);
> > ws.onopen = function() {
> >   ws.send(very_very_long_message);
> >   ws.close();
> > };
>
> Shouldn't ws.close() allow queued messages to be flushed before the
> actual closing?
>
> >
> > It is possible to _Close the WebSocket connection_ without the closing
> > handshake in such cases, but since the WebSocket API doesn't expose the
> > transmission status, doing so makes the closing handshake completely
> > unreliable.
> >
> >
> >
> > On Wed, Feb 19, 2014 at 4:09 PM, Adam Rice <ricea@google.com> wrote:
> >>
> >> With the current set of control frames, I would not miss the ability to
> >> interject them in the middle of a data message.
> >>
> >> Breaking them down one by one:
> >>
> >> * Pong: useless in the middle of a data message (the peer already knows
> >> that it is receiving data, and useless for RTT measurement because it
> has an
> >> unknown amount of OS queueing and buffer bloat in front of it).
> >> * Ping: almost always useless in the middle of a data message. It could
> be
> >> useful if, for example, you had a middle box which closed "idle"
> >> connections, and considered any connection "idle" if there was no TCP/IP
> >> payload traffic originating from the client, and the server was in the
> >> middle of sending a huge message. However, such a middle box would also
> >> break HTTP downloads so seems unlikely to exist in practice. Even if it
> >> existed, how would the server detect it to know that it needed to send a
> >> Ping?
> >> * Close: Can be used to perform a "clean" close in the middle of
> sending a
> >> message. However, I cannot think of any applications that would
> consider a
> >> situation in which a in-transit message was aborted to close the
> connection
> >> to be a "clean close" scenario.
> >>
> >>
> >> On 19 February 2014 13:24, Takeshi Yoshino <tyoshino@google.com> wrote:
> >>>
> >>> I'm inclined to say "Yes" for this question so said yes for #2 of
> >>> Yutaka's original question [1], but I could also cope with if we
> decide to
> >>> drop this.
> >>>
> >>> [1] http://www.ietf.org/mail-archive/web/hybi/current/msg10403.html
> >>>
> >>> _______________________________________________
> >>> hybi mailing list
> >>> hybi@ietf.org
> >>> https://www.ietf.org/mailman/listinfo/hybi
> >>>
> >>
> >
> >
> > _______________________________________________
> > hybi mailing list
> > hybi@ietf.org
> > https://www.ietf.org/mailman/listinfo/hybi
> >
>

--089e0160a4c80efd8404f336dbba
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px =
0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-l=
eft-style:solid;padding-left:1ex"><span style=3D"font-family:arial,sans-ser=
if;font-size:13px">Shouldn&#39;t ws.close() allow queued messages to be flu=
shed before the<br>

</span><span style=3D"font-family:arial,sans-serif;font-size:13px">actual c=
losing?</span></blockquote><div>It is allowed to wait until the endpoint fi=
nishes sending the message.</div><div>Is it the desired behavior? I thought=
 it is undesired, but I&#39;m not confident now.</div>
<div><br></div><div class=3D"gmail_extra"><div class=3D"gmail_quote">On Tue=
, Feb 25, 2014 at 2:40 PM, Zhong Yu <span dir=3D"ltr">&lt;<a href=3D"mailto=
:zhong.j.yu@gmail.com" target=3D"_blank">zhong.j.yu@gmail.com</a>&gt;</span=
> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
<div>On Mon, Feb 24, 2014 at 10:58 PM, Yutaka Hirano &lt;<a href=3D"mailto:=
yhirano@chromium.org" target=3D"_blank">yhirano@chromium.org</a>&gt; wrote:=
<br>
&gt;&gt; * Close: Can be used to perform a &quot;clean&quot; close in the m=
iddle of sending a<br>
&gt;&gt; message. However, I cannot think of any applications that would co=
nsider a<br>
&gt;&gt; situation in which a in-transit message was aborted to close the c=
onnection<br>
&gt;&gt; to be a &quot;clean close&quot; scenario.<br>
&gt;<br>
&gt; An endpoint should be able to cancel the ongoing transmission, i.e. Th=
e<br>
&gt; close call should abort transferring very_very_long_message in the fol=
lowing<br>
&gt; code.<br>
&gt;<br>
&gt;<br>
&gt; ws =3D new WebSocket(...);<br>
&gt; ws.onopen =3D function() {<br>
&gt; =C2=A0 ws.send(very_very_long_message);<br>
&gt; =C2=A0 ws.close();<br>
&gt; };<br>
<br>
</div>Shouldn&#39;t ws.close() allow queued messages to be flushed before t=
he<br>
actual closing?<br>
<div><div><br>
&gt;<br>
&gt; It is possible to _Close the WebSocket connection_ without the closing=
<br>
&gt; handshake in such cases, but since the WebSocket API doesn&#39;t expos=
e the<br>
&gt; transmission status, doing so makes the closing handshake completely<b=
r>
&gt; unreliable.<br>
&gt;<br>
&gt;<br>
&gt;<br>
&gt; On Wed, Feb 19, 2014 at 4:09 PM, Adam Rice &lt;<a href=3D"mailto:ricea=
@google.com" target=3D"_blank">ricea@google.com</a>&gt; wrote:<br>
&gt;&gt;<br>
&gt;&gt; With the current set of control frames, I would not miss the abili=
ty to<br>
&gt;&gt; interject them in the middle of a data message.<br>
&gt;&gt;<br>
&gt;&gt; Breaking them down one by one:<br>
&gt;&gt;<br>
&gt;&gt; * Pong: useless in the middle of a data message (the peer already =
knows<br>
&gt;&gt; that it is receiving data, and useless for RTT measurement because=
 it has an<br>
&gt;&gt; unknown amount of OS queueing and buffer bloat in front of it).<br=
>
&gt;&gt; * Ping: almost always useless in the middle of a data message. It =
could be<br>
&gt;&gt; useful if, for example, you had a middle box which closed &quot;id=
le&quot;<br>
&gt;&gt; connections, and considered any connection &quot;idle&quot; if the=
re was no TCP/IP<br>
&gt;&gt; payload traffic originating from the client, and the server was in=
 the<br>
&gt;&gt; middle of sending a huge message. However, such a middle box would=
 also<br>
&gt;&gt; break HTTP downloads so seems unlikely to exist in practice. Even =
if it<br>
&gt;&gt; existed, how would the server detect it to know that it needed to =
send a<br>
&gt;&gt; Ping?<br>
&gt;&gt; * Close: Can be used to perform a &quot;clean&quot; close in the m=
iddle of sending a<br>
&gt;&gt; message. However, I cannot think of any applications that would co=
nsider a<br>
&gt;&gt; situation in which a in-transit message was aborted to close the c=
onnection<br>
&gt;&gt; to be a &quot;clean close&quot; scenario.<br>
&gt;&gt;<br>
&gt;&gt;<br>
&gt;&gt; On 19 February 2014 13:24, Takeshi Yoshino &lt;<a href=3D"mailto:t=
yoshino@google.com" target=3D"_blank">tyoshino@google.com</a>&gt; wrote:<br=
>
&gt;&gt;&gt;<br>
&gt;&gt;&gt; I&#39;m inclined to say &quot;Yes&quot; for this question so s=
aid yes for #2 of<br>
&gt;&gt;&gt; Yutaka&#39;s original question [1], but I could also cope with=
 if we decide to<br>
&gt;&gt;&gt; drop this.<br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt; [1] <a href=3D"http://www.ietf.org/mail-archive/web/hybi/curre=
nt/msg10403.html" target=3D"_blank">http://www.ietf.org/mail-archive/web/hy=
bi/current/msg10403.html</a><br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt; _______________________________________________<br>
&gt;&gt;&gt; hybi mailing list<br>
&gt;&gt;&gt; <a href=3D"mailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.o=
rg</a><br>
&gt;&gt;&gt; <a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=
=3D"_blank">https://www.ietf.org/mailman/listinfo/hybi</a><br>
&gt;&gt;&gt;<br>
&gt;&gt;<br>
&gt;<br>
&gt;<br>
&gt; _______________________________________________<br>
&gt; hybi mailing list<br>
&gt; <a href=3D"mailto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><b=
r>
&gt; <a href=3D"https://www.ietf.org/mailman/listinfo/hybi" target=3D"_blan=
k">https://www.ietf.org/mailman/listinfo/hybi</a><br>
&gt;<br>
</div></div></blockquote></div><br></div></div>

--089e0160a4c80efd8404f336dbba--


From nobody Tue Feb 25 04:31:44 2014
Return-Path: <arman@noemax.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 398BC1A06C2 for <hybi@ietfa.amsl.com>; Tue, 25 Feb 2014 04:31:42 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 0.153
X-Spam-Level: 
X-Spam-Status: No, score=0.153 tagged_above=-999 required=5 tests=[BAYES_50=0.8, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.547, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id DGavkMjVotC1 for <hybi@ietfa.amsl.com>; Tue, 25 Feb 2014 04:31:39 -0800 (PST)
Received: from mail.noemax.com (mail.noemax.com [74.208.113.37]) by ietfa.amsl.com (Postfix) with ESMTP id 97B2F1A06CD for <hybi@ietf.org>; Tue, 25 Feb 2014 04:31:39 -0800 (PST)
DKIM-Signature: a=rsa-sha1; t=1393331429; x=1393936229; s=m2048; d=noemax.com;  c=relaxed/relaxed; v=1; bh=HmJMS0quqHvAo8rdUJW3zPcpMLQ=; h=From:Subject:Date:Message-ID:To:Cc:MIME-Version:Content-Type:In-Reply-To:References; b=LUZ9lDXbAgMZFQHrEs/1k2TvVIgH9aBMeg75e2UhxVyXbquzkU/csYEUdOB7kHanCLruc//W/b/u5O7QGOBxHRuVDS2bMAHiJJU1HuTuKTIvcyCr1DL/ldAWNDHsxENtNJ3HzyyIk/+6KUya/MZepuHbrmdzmso/2Dyv8wNwFA5OxkhsPuAlE25fX6sfQlEtWyseGoq2VoeAprqmiIDCVlp5dolgGuatWl8ujU8f+6xSXyjn5Ip/Itq+jrAjPmcLx3g4/KEPbuRp/62B8BDja0Lv6Ib+GQUrOvQbYPmHAt93ub9n8GYJCmGeWa/9Vs0szgGqDMpNeIHsKcETDMtT6Q==
Received: from ArmanPCW8 by mail.noemax.com (Noemax Mail Server) with ASMTP (SSL) id 201402251430289450; Tue, 25 Feb 2014 14:30:28 +0200
From: "Arman Djusupov" <arman@noemax.com>
To: "'Takeshi Yoshino'" <tyoshino@google.com>, "'Tobias Oberstein'" <tobias.oberstein@tavendo.de>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <634914A010D0B943A035D226786325D4446C537686@EXVMBX020-12.exch020.serverdata.net> <CAH9hSJaALTNiC9rUC-qVYKK74_YJA5psvGhbsFegTwaLsg9Dug@mail.gmail.com>
In-Reply-To: <CAH9hSJaALTNiC9rUC-qVYKK74_YJA5psvGhbsFegTwaLsg9Dug@mail.gmail.com>
Date: Tue, 25 Feb 2014 14:31:29 +0200
Message-ID: <00e501cf3225$84f0b520$8ed21f60$@noemax.com>
MIME-Version: 1.0
Content-Type: multipart/alternative; boundary="----=_NextPart_000_00E6_01CF3236.487BCF10"
X-Mailer: Microsoft Outlook 14.0
Thread-Index: AQLZFU8Gf4QDl/nttx8R1PjVcX14zAJ3oJOBAcC/tj+YkFs2gA==
Content-Language: en-us
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/BCQ-O8rMC4Oz0EgY2g1FmXjkXQI
Cc: hybi@ietf.org, 'Yutaka Hirano' <yhirano@google.com>, 'Peter Thorson' <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 25 Feb 2014 12:31:42 -0000

This is a multipart message in MIME format.

------=_NextPart_000_00E6_01CF3236.487BCF10
Content-Type: text/plain;
	charset="utf-8"
Content-Transfer-Encoding: quoted-printable

Takeshi,

=20

I assume that Tobias means that HTTP/2 should tunnel the WebSocket =
transport =E2=80=9Cas is=E2=80=9D without stripping off the WebSocket =
headers or functionality. If so then I am in agreement with Tobias.

=20

I can=E2=80=99t think of a good reason why can=E2=80=99t we just forward =
a complete WS frame with its full header through the HTTP/2. If I missed =
an argument that was put forward in a previous discussions please do =
point me to it. IMO plain tunneling, just like HTTP is tunneled through =
TLS, would do fine.

=20

If we strip off almost the entire frame header and we fragment the =
payload so that it is carried by an HTTP/2 frame, we are practically =
creating a new protocol. Such a protocol doesn=E2=80=99t really need to =
be related to WebSocket, however (if we do create this new protocol) it =
can be transparently exposed through the same browser WS API when HTTP/2 =
is available.

=20

With best regards,

Arman

=20

=20

From: hybi [mailto:hybi-bounces@ietf.org] On Behalf Of Takeshi Yoshino
Sent: Tuesday, February 25, 2014 12:36 AM
To: Tobias Oberstein
Cc: hybi@ietf.org; Yutaka Hirano; Peter Thorson
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket =
level control frame? (was: Re: Discontinuation of mux ...)

=20

On Wed, Feb 19, 2014 at 8:45 PM, Tobias Oberstein =
<tobias.oberstein@tavendo.de> wrote:

I don't agree with that approach. I think HTTP2 should take WebSocket =
"as is", if it wants to act as a transport instead of attempting to =
redefine it. Or define it's own messaging (not bound by WS).

=20

I want to understand what you mean by this, Tobias, and maybe Peter.

=20

As one of Chromium developers, I just want to build and use a scalable =
transport for the WebSocket API. For this goal, it doesn't matter =
whether HTTP/2.0 works as a transparent transport or not. There's no =
such implication in the term "WS/HTTP/2.0" at least when used by me.

=20

What Roberto and Martin are suggesting sounds like one to be categorized =
as "define it's own messaging". Some of Yutaka's plans, too. What's your =
actual position for them. Not the best option for you but not against?

=20


------=_NextPart_000_00E6_01CF3236.487BCF10
Content-Type: text/html;
	charset="utf-8"
Content-Transfer-Encoding: quoted-printable

<html xmlns:v=3D"urn:schemas-microsoft-com:vml" =
xmlns:o=3D"urn:schemas-microsoft-com:office:office" =
xmlns:w=3D"urn:schemas-microsoft-com:office:word" =
xmlns:m=3D"http://schemas.microsoft.com/office/2004/12/omml" =
xmlns=3D"http://www.w3.org/TR/REC-html40"><head><meta =
http-equiv=3DContent-Type content=3D"text/html; charset=3Dutf-8"><meta =
name=3DGenerator content=3D"Microsoft Word 14 (filtered =
medium)"><style><!--
/* Font Definitions */
@font-face
	{font-family:"Cambria Math";
	panose-1:2 4 5 3 5 4 6 3 2 4;}
@font-face
	{font-family:Calibri;
	panose-1:2 15 5 2 2 2 4 3 2 4;}
@font-face
	{font-family:Tahoma;
	panose-1:2 11 6 4 3 5 4 4 2 4;}
/* Style Definitions */
p.MsoNormal, li.MsoNormal, div.MsoNormal
	{margin:0in;
	margin-bottom:.0001pt;
	font-size:12.0pt;
	font-family:"Times New Roman","serif";}
a:link, span.MsoHyperlink
	{mso-style-priority:99;
	color:blue;
	text-decoration:underline;}
a:visited, span.MsoHyperlinkFollowed
	{mso-style-priority:99;
	color:purple;
	text-decoration:underline;}
span.EmailStyle17
	{mso-style-type:personal-reply;
	font-family:"Calibri","sans-serif";
	color:#44546A;}
.MsoChpDefault
	{mso-style-type:export-only;
	font-size:10.0pt;}
@page WordSection1
	{size:8.5in 11.0in;
	margin:1.0in 1.0in 1.0in 1.0in;}
div.WordSection1
	{page:WordSection1;}
--></style><!--[if gte mso 9]><xml>
<o:shapedefaults v:ext=3D"edit" spidmax=3D"1026" />
</xml><![endif]--><!--[if gte mso 9]><xml>
<o:shapelayout v:ext=3D"edit">
<o:idmap v:ext=3D"edit" data=3D"1" />
</o:shapelayout></xml><![endif]--></head><body lang=3DEN-US link=3Dblue =
vlink=3Dpurple><div class=3DWordSection1><p class=3DMsoNormal><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>Takeshi,<o:p></o:p></span></p><p class=3DMsoNormal><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'><o:p>&nbsp;</o:p></span></p><p class=3DMsoNormal><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>I assume that Tobias </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'>mean</span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>s</span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'> that HTTP/2 should tunnel </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>the </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'>WebSocket transport =E2=80=9Cas is=E2=80=9D without stri</span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>p</span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'>ping off </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>the </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'>WebSocket headers or functionality. </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>If so then I am in agreement with Tobias.</span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'><o:p></o:p></span></p><p class=3DMsoNormal><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'><o:p>&nbsp;</o:p></span></p><p class=3DMsoNormal><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>I can=E2=80=99t think of a good reason w</span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'>hy can=E2=80=99t we just forward </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>a </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'>complete WS frame with </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>its </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'>full header through the HTTP/2</span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>. If I missed an argument that was put forward in a previous =
discussions please do point me to it. IMO p</span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'>lain tunneling</span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>, </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'>just like HTTP </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>is </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'>tunneled through TLS</span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>, would do fine.</span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'><o:p></o:p></span></p><p class=3DMsoNormal><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'><o:p>&nbsp;</o:p></span></p><p class=3DMsoNormal><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'>If we strip off almost </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>the </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'>entire frame header and </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>we </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'>fragment </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>the </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'>payload </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>so that it is carried by an</span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'> HTTP/2 frame, we are practically creating a new protocol. Such =
</span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>a </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'>protocol doesn=E2=80=99t really need to be related to =
WebSocket</span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>, however (if we do create this new protocol) it </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'>can be </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>transparently </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'>exposed through the </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#1F497=
D'>same </span><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'>browser WS API when HTTP/2 is available.<o:p></o:p></span></p><p =
class=3DMsoNormal><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'><o:p>&nbsp;</o:p></span></p><p class=3DMsoNormal><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'>With best regards,<o:p></o:p></span></p><p class=3DMsoNormal><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'>Arman<o:p></o:p></span></p><p class=3DMsoNormal><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'><o:p>&nbsp;</o:p></span></p><p class=3DMsoNormal><span =
style=3D'font-size:11.0pt;font-family:"Calibri","sans-serif";color:#44546=
A'><o:p>&nbsp;</o:p></span></p><p class=3DMsoNormal><b><span =
style=3D'font-size:10.0pt;font-family:"Tahoma","sans-serif"'>From:</span>=
</b><span style=3D'font-size:10.0pt;font-family:"Tahoma","sans-serif"'> =
hybi [mailto:hybi-bounces@ietf.org] <b>On Behalf Of </b>Takeshi =
Yoshino<br><b>Sent:</b> Tuesday, February 25, 2014 12:36 =
AM<br><b>To:</b> Tobias Oberstein<br><b>Cc:</b> hybi@ietf.org; Yutaka =
Hirano; Peter Thorson<br><b>Subject:</b> Re: [hybi] Straw poll: Do you =
miss interjectable WebSocket level control frame? (was: Re: =
Discontinuation of mux ...)<o:p></o:p></span></p><p =
class=3DMsoNormal><o:p>&nbsp;</o:p></p><div><div><div><p =
class=3DMsoNormal>On Wed, Feb 19, 2014 at 8:45 PM, Tobias Oberstein =
&lt;<a href=3D"mailto:tobias.oberstein@tavendo.de" =
target=3D"_blank">tobias.oberstein@tavendo.de</a>&gt; =
wrote:<o:p></o:p></p><div><p class=3DMsoNormal>I don't agree with that =
approach. I think HTTP2 should take WebSocket &quot;as is&quot;, if it =
wants to act as a transport instead of attempting to redefine it. Or =
define it's own messaging (not bound by WS).<o:p></o:p></p></div><div><p =
class=3DMsoNormal><o:p>&nbsp;</o:p></p></div><div><p class=3DMsoNormal>I =
want to understand what you mean by this, Tobias, and maybe =
Peter.<o:p></o:p></p></div><div><p =
class=3DMsoNormal><o:p>&nbsp;</o:p></p></div><div><p =
class=3DMsoNormal>As one of Chromium developers, I just want to build =
and use a scalable transport for the WebSocket API. For this goal, it =
doesn't matter whether HTTP/2.0 works as a transparent transport or not. =
There's no such implication in the term &quot;WS/HTTP/2.0&quot; at least =
when used by me.<o:p></o:p></p></div><div><p =
class=3DMsoNormal><o:p>&nbsp;</o:p></p></div><div><p =
class=3DMsoNormal>What Roberto and Martin are suggesting sounds like one =
to be categorized as &quot;define it's own messaging&quot;. Some of =
Yutaka's plans, too. What's your actual position for them. Not the best =
option for you but not against?<o:p></o:p></p></div><div><p =
class=3DMsoNormal>&nbsp;<o:p></o:p></p></div></div></div></div></div></bo=
dy></html>
------=_NextPart_000_00E6_01CF3236.487BCF10--


From nobody Tue Feb 25 11:11:54 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 641191A01D3 for <hybi@ietfa.amsl.com>; Tue, 25 Feb 2014 11:11:48 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -0.026
X-Spam-Level: 
X-Spam-Status: No, score=-0.026 tagged_above=-999 required=5 tests=[BAYES_40=-0.001, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.547, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id u4V9J94reOv1 for <hybi@ietfa.amsl.com>; Tue, 25 Feb 2014 11:11:44 -0800 (PST)
Received: from mail-wi0-x236.google.com (mail-wi0-x236.google.com [IPv6:2a00:1450:400c:c05::236]) by ietfa.amsl.com (Postfix) with ESMTP id 820CC1A0122 for <hybi@ietf.org>; Tue, 25 Feb 2014 11:11:43 -0800 (PST)
Received: by mail-wi0-f182.google.com with SMTP id f8so1200149wiw.3 for <hybi@ietf.org>; Tue, 25 Feb 2014 11:11:42 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc:content-type; bh=vBz2XnHe6kqGkPo34MI3bodAOfCiQJZb0HuXdGVFvuo=; b=N6Q2YSo4mAYjwOaF0F0998ausa6bR+GKTH7aDegFyA7Aob2SvT9WRqgvQxttCLGlTA DsMe6lNluxL6NvHOpsTXQmWdtl5bWKj1T+9mFN4BRX52I+c8pNFA2RZyfrQ+S0odrr7Z ZowdkJ8vPuaDkNCd5zqACVRaaFpxKv9O9r9A+q2mfI+NoXl+3p5KN7c1DSUCCRb2/fT8 vbzYDBcqTJyg4anMbt0+NwNkhUaM85+JCnLbxIgo0ThZa+gDqcNKA0cojmH5q7hGV6lx es+W2wVdd7wlUhBMCSJhxIA2v9UdUpzSbIanmptu4iHf/Z6Fj1wscMxMTTuO6PHpquqt i0gw==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc:content-type; bh=vBz2XnHe6kqGkPo34MI3bodAOfCiQJZb0HuXdGVFvuo=; b=YiOWtLQV3bgIQCwIo8iz7ybDXOJ3ZPCFcXnkYcpWeceoUAOTgEBzBiEOx3nZevNJd5 HE2FZcDPX4ZWh5ad9yKKXC0XDUWVZhVIjn4CV7EXUb3mhvUkE31klM8C4LFKBSe6Fcgt hucIkZYGcmWlVizRH5YjSwdbaKrvBLTykklr4P2yOlH6F5cI6vz9Re0feh3g8a0hheEf sQcHBXJLb5VvNCCOaWcKeFwRFCTBJazzzcfQp61e7XU+21LltPEaFfRqTv0LWhaBmPym 0kSW1iYT7DdcasvO/xHuKsg94rAFP3xTHBiZBnws8M52yLB6jaLkwxZJV2av7BN3vSqu GWhA==
X-Gm-Message-State: ALoCoQmq18wbZChV1mT6WeFoOsjizelFFtM0aJcqP4EEIWqDZTfqSesBkqa2saVUUkOH+r47EJW44GzFX1fRM0iRSRxPXe1iCCZYWVkD57ExzyVOt/Js4ygM5YqdfZJ5al2Cs/go/n9ByAB+ifhe410hMHAMYpywbS740RrKBJ+iXuD0Mqr+vPlF+2jDzhXr8hBNBZVWH6cC
X-Received: by 10.180.73.173 with SMTP id m13mr1482828wiv.52.1393355501949; Tue, 25 Feb 2014 11:11:41 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Tue, 25 Feb 2014 11:11:21 -0800 (PST)
In-Reply-To: <00e501cf3225$84f0b520$8ed21f60$@noemax.com>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <634914A010D0B943A035D226786325D4446C537686@EXVMBX020-12.exch020.serverdata.net> <CAH9hSJaALTNiC9rUC-qVYKK74_YJA5psvGhbsFegTwaLsg9Dug@mail.gmail.com> <00e501cf3225$84f0b520$8ed21f60$@noemax.com>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Wed, 26 Feb 2014 04:11:21 +0900
Message-ID: <CAH9hSJb91zndtZnUq=SQKWNRGh3B1yLWfV9L-qo0=vS1KkVF4A@mail.gmail.com>
To: Arman Djusupov <arman@noemax.com>
Content-Type: multipart/alternative; boundary=f46d043c07e0bd707504f33fdb59
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/JHCC_DI4K8yOqWwv9kuHrNN2Q2c
Cc: Yutaka Hirano <yhirano@google.com>, "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 25 Feb 2014 19:11:48 -0000

--f46d043c07e0bd707504f33fdb59
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

Hi Arman,

On Tue, Feb 25, 2014 at 9:31 PM, Arman Djusupov <arman@noemax.com> wrote:

> Takeshi,
>
>
>
> I assume that Tobias means that HTTP/2 should tunnel the WebSocket
> transport =E2=80=9Cas is=E2=80=9D without stripping off the WebSocket hea=
ders or
> functionality. If so then I am in agreement with Tobias.
>

I understand this argument. For developers who've built and are already
using RFC 6455 based infrastructure, this approach is very simple and
allows for faster and easier migration.


>
>
> I can=E2=80=99t think of a good reason why can=E2=80=99t we just forward =
a complete WS
> frame with its full header through the HTTP/2. If I missed an argument
> that was put forward in a previous discussions please do point me to it.
> IMO plain tunneling, just like HTTP is tunneled through TLS, would do
> fine.
>

Please search for Roberto's posts in the HyBi archive [1] and Roberto and
Martin's posts to topics including "WebSocket" in httpbis archive [2].
Basically, for those who haven't deployed RFC 6455 based software but
(planning to) introducing HTTP/2.0, inheriting RFC 6455 architecture
provides no benefit but some design points of RFC 6455 considered to be
harmful rather motivate them to throw away RFC 6455 and redesign messaging
architecture inside HTTP/2.0 that can be applied to the WS API. Another
Roberto's point is that basic messaging units and meta data should be
easily recognized and investigated by load balancers with least knowledge
in addition to HTTP/2.0. Clearly separated thinner layering provides
flexibility but also requires more code and tests.


>
>
> If we strip off almost the entire frame header and we fragment the payloa=
d
> so that it is carried by an HTTP/2 frame, we are practically creating a
> new protocol. Such a protocol doesn=E2=80=99t really need to be related t=
o
> WebSocket, however (if we do create this new protocol) it can be transpar=
ently
> exposed through the same browser WS API when HTTP/2 is available.
>
>
>

Yes. I just want to deliver some protocol that is scalable under the
browser WS API. Whether it's related to the WebSocket Protocol or not does
not matter so much. Since we don't want to implement two or more protocol
in Chromium to achieve the same goal (multiplexed protocol for WS API), we
need to choose one.

There seems to be disjoint between the two parties. I understand both
argument but need to find some landing point.

[1] http://www.ietf.org/mail-archive/web/hybi/current/maillist.html
[2] http://lists.w3.org/Archives/Public/ietf-http-wg/2014JanMar/

--f46d043c07e0bd707504f33fdb59
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote">Hi A=
rman,</div><div class=3D"gmail_quote"><br></div><div class=3D"gmail_quote">=
On Tue, Feb 25, 2014 at 9:31 PM, Arman Djusupov <span dir=3D"ltr">&lt;<a hr=
ef=3D"mailto:arman@noemax.com" target=3D"_blank">arman@noemax.com</a>&gt;</=
span> wrote:<br>

<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div lang=3D"EN-US" link=3D"blue" vlink=3D"purple"><div><p=
 class=3D"MsoNormal">

<span style=3D"font-size:11pt;font-family:Calibri,sans-serif;color:rgb(31,7=
3,125)">Takeshi,<u></u><u></u></span></p><p class=3D"MsoNormal"><span style=
=3D"font-size:11pt;font-family:Calibri,sans-serif;color:rgb(31,73,125)"><u>=
</u>=C2=A0<u></u></span></p>

<p class=3D"MsoNormal"><span style=3D"font-size:11pt;font-family:Calibri,sa=
ns-serif;color:rgb(31,73,125)">I assume that Tobias </span><span style=3D"f=
ont-size:11pt;font-family:Calibri,sans-serif;color:rgb(68,84,106)">mean</sp=
an><span style=3D"font-size:11pt;font-family:Calibri,sans-serif;color:rgb(3=
1,73,125)">s</span><span style=3D"font-size:11pt;font-family:Calibri,sans-s=
erif;color:rgb(68,84,106)"> that HTTP/2 should tunnel </span><span style=3D=
"font-size:11pt;font-family:Calibri,sans-serif;color:rgb(31,73,125)">the </=
span><span style=3D"font-size:11pt;font-family:Calibri,sans-serif;color:rgb=
(68,84,106)">WebSocket transport =E2=80=9Cas is=E2=80=9D without stri</span=
><span style=3D"font-size:11pt;font-family:Calibri,sans-serif;color:rgb(31,=
73,125)">p</span><span style=3D"font-size:11pt;font-family:Calibri,sans-ser=
if;color:rgb(68,84,106)">ping off </span><span style=3D"font-size:11pt;font=
-family:Calibri,sans-serif;color:rgb(31,73,125)">the </span><span style=3D"=
font-size:11pt;font-family:Calibri,sans-serif;color:rgb(68,84,106)">WebSock=
et headers or functionality. </span><span style=3D"font-size:11pt;font-fami=
ly:Calibri,sans-serif;color:rgb(31,73,125)">If so then I am in agreement wi=
th Tobias.</span></p>

</div></div></blockquote><div><br></div><div>I understand this argument. Fo=
r developers who&#39;ve built and are already using RFC 6455 based infrastr=
ucture, this approach is very simple and allows for faster and easier migra=
tion.</div>

<div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px =
0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-l=
eft-style:solid;padding-left:1ex"><div lang=3D"EN-US" link=3D"blue" vlink=
=3D"purple">
<div>
<p class=3D"MsoNormal"><span style=3D"font-size:11pt;font-family:Calibri,sa=
ns-serif;color:rgb(68,84,106)"><u></u><u></u></span></p><p class=3D"MsoNorm=
al"><span style=3D"font-size:11pt;font-family:Calibri,sans-serif;color:rgb(=
68,84,106)"><u></u>=C2=A0<u></u></span></p>

<p class=3D"MsoNormal"><span style=3D"font-size:11pt;font-family:Calibri,sa=
ns-serif;color:rgb(31,73,125)">I can=E2=80=99t think of a good reason w</sp=
an><span style=3D"font-size:11pt;font-family:Calibri,sans-serif;color:rgb(6=
8,84,106)">hy can=E2=80=99t we just forward </span><span style=3D"font-size=
:11pt;font-family:Calibri,sans-serif;color:rgb(31,73,125)">a </span><span s=
tyle=3D"font-size:11pt;font-family:Calibri,sans-serif;color:rgb(68,84,106)"=
>complete WS frame with </span><span style=3D"font-size:11pt;font-family:Ca=
libri,sans-serif;color:rgb(31,73,125)">its </span><span style=3D"font-size:=
11pt;font-family:Calibri,sans-serif;color:rgb(68,84,106)">full header throu=
gh the HTTP/2</span><span style=3D"font-size:11pt;font-family:Calibri,sans-=
serif;color:rgb(31,73,125)">. If I missed an argument that was put forward =
in a previous discussions please do point me to it. IMO p</span><span style=
=3D"font-size:11pt;font-family:Calibri,sans-serif;color:rgb(68,84,106)">lai=
n tunneling</span><span style=3D"font-size:11pt;font-family:Calibri,sans-se=
rif;color:rgb(31,73,125)">, </span><span style=3D"font-size:11pt;font-famil=
y:Calibri,sans-serif;color:rgb(68,84,106)">just like HTTP </span><span styl=
e=3D"font-size:11pt;font-family:Calibri,sans-serif;color:rgb(31,73,125)">is=
 </span><span style=3D"font-size:11pt;font-family:Calibri,sans-serif;color:=
rgb(68,84,106)">tunneled through TLS</span><span style=3D"font-size:11pt;fo=
nt-family:Calibri,sans-serif;color:rgb(31,73,125)">, would do fine.</span><=
/p>

</div></div></blockquote><div><br></div><div>Please search for Roberto&#39;=
s posts in the HyBi archive [1] and Roberto and Martin&#39;s posts to topic=
s including &quot;WebSocket&quot; in httpbis archive [2]. Basically, for th=
ose who haven&#39;t deployed RFC 6455 based software but (planning to) intr=
oducing HTTP/2.0, inheriting RFC 6455 architecture provides no benefit but =
some design points of RFC 6455 considered to be harmful rather motivate the=
m to throw away RFC 6455 and redesign messaging architecture inside HTTP/2.=
0 that can be applied to the WS API. Another Roberto&#39;s point is that ba=
sic messaging units and meta data should be easily recognized and investiga=
ted by load balancers with least knowledge in addition to HTTP/2.0. Clearly=
 separated thinner layering provides flexibility but also requires more cod=
e and tests.</div>

<div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px =
0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-l=
eft-style:solid;padding-left:1ex"><div lang=3D"EN-US" link=3D"blue" vlink=
=3D"purple">
<div>
<p class=3D"MsoNormal"><span style=3D"font-size:11pt;font-family:Calibri,sa=
ns-serif;color:rgb(68,84,106)"><u></u><u></u></span></p><p class=3D"MsoNorm=
al"><span style=3D"font-size:11pt;font-family:Calibri,sans-serif;color:rgb(=
68,84,106)"><u></u>=C2=A0<u></u></span></p>

<p class=3D"MsoNormal"><span style=3D"font-size:11pt;font-family:Calibri,sa=
ns-serif;color:rgb(68,84,106)">If we strip off almost </span><span style=3D=
"font-size:11pt;font-family:Calibri,sans-serif;color:rgb(31,73,125)">the </=
span><span style=3D"font-size:11pt;font-family:Calibri,sans-serif;color:rgb=
(68,84,106)">entire frame header and </span><span style=3D"font-size:11pt;f=
ont-family:Calibri,sans-serif;color:rgb(31,73,125)">we </span><span style=
=3D"font-size:11pt;font-family:Calibri,sans-serif;color:rgb(68,84,106)">fra=
gment </span><span style=3D"font-size:11pt;font-family:Calibri,sans-serif;c=
olor:rgb(31,73,125)">the </span><span style=3D"font-size:11pt;font-family:C=
alibri,sans-serif;color:rgb(68,84,106)">payload </span><span style=3D"font-=
size:11pt;font-family:Calibri,sans-serif;color:rgb(31,73,125)">so that it i=
s carried by an</span><span style=3D"font-size:11pt;font-family:Calibri,san=
s-serif;color:rgb(68,84,106)"> HTTP/2 frame, we are practically creating a =
new protocol. Such </span><span style=3D"font-size:11pt;font-family:Calibri=
,sans-serif;color:rgb(31,73,125)">a </span><span style=3D"font-size:11pt;fo=
nt-family:Calibri,sans-serif;color:rgb(68,84,106)">protocol doesn=E2=80=99t=
 really need to be related to WebSocket</span><span style=3D"font-size:11pt=
;font-family:Calibri,sans-serif;color:rgb(31,73,125)">, however (if we do c=
reate this new protocol) it </span><span style=3D"font-size:11pt;font-famil=
y:Calibri,sans-serif;color:rgb(68,84,106)">can be </span><span style=3D"fon=
t-size:11pt;font-family:Calibri,sans-serif;color:rgb(31,73,125)">transparen=
tly </span><span style=3D"font-size:11pt;font-family:Calibri,sans-serif;col=
or:rgb(68,84,106)">exposed through the </span><span style=3D"font-size:11pt=
;font-family:Calibri,sans-serif;color:rgb(31,73,125)">same </span><span sty=
le=3D"font-size:11pt;font-family:Calibri,sans-serif;color:rgb(68,84,106)">b=
rowser WS API when HTTP/2 is available.<u></u><u></u></span></p>

<p class=3D"MsoNormal"><span style=3D"font-size:11pt;font-family:Calibri,sa=
ns-serif;color:rgb(68,84,106)"><u></u>=C2=A0</span></p></div></div></blockq=
uote><div><br></div><div>Yes. I just want to deliver some protocol that is =
scalable under the browser WS API. Whether it&#39;s related to the WebSocke=
t Protocol or not does not matter so much. Since we don&#39;t want to imple=
ment two or more protocol in Chromium to achieve the same goal (multiplexed=
 protocol for WS API), we need to choose one.</div>

<div><br></div><div>There seems to be disjoint between the two parties. I u=
nderstand both argument but need to find some landing point.</div><div><br>=
</div><div>[1] <a href=3D"http://www.ietf.org/mail-archive/web/hybi/current=
/maillist.html">http://www.ietf.org/mail-archive/web/hybi/current/maillist.=
html</a></div>

<div>[2]=C2=A0<a href=3D"http://lists.w3.org/Archives/Public/ietf-http-wg/2=
014JanMar/">http://lists.w3.org/Archives/Public/ietf-http-wg/2014JanMar/</a=
></div><div>=C2=A0</div></div></div></div>

--f46d043c07e0bd707504f33fdb59--


From nobody Tue Feb 25 14:05:27 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id D17EC1A07C1 for <hybi@ietfa.amsl.com>; Tue, 25 Feb 2014 14:05:25 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 1.174
X-Spam-Level: *
X-Spam-Status: No, score=1.174 tagged_above=-999 required=5 tests=[BAYES_40=-0.001, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, J_CHICKENPOX_26=0.6, J_CHICKENPOX_52=0.6, RP_MATCHES_RCVD=-0.547, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id FvYOpgGcSsQO for <hybi@ietfa.amsl.com>; Tue, 25 Feb 2014 14:05:24 -0800 (PST)
Received: from mail-we0-x234.google.com (mail-we0-x234.google.com [IPv6:2a00:1450:400c:c03::234]) by ietfa.amsl.com (Postfix) with ESMTP id A2E5F1A07C9 for <hybi@ietf.org>; Tue, 25 Feb 2014 14:05:23 -0800 (PST)
Received: by mail-we0-f180.google.com with SMTP id u57so959176wes.39 for <hybi@ietf.org>; Tue, 25 Feb 2014 14:05:22 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc:content-type; bh=66zdOwURItyDJqQ5+W+vuQ8O8WFy428OvtMHvHZyhAM=; b=eV5dAOISFAHT+sWS6lE60ThMwMW5zBqY1rocA5vYLkJ6G8lBPkLWvj/8x0ms3BDtIc G57JQsuHcYt/rDsnLVbhDpPNjwdGCbl5LERby5tA8N2IWiFfuGo+1ronOoWtYt3iBMuG WTmJaFXSQJ7PeuKCEIVAcHR1C8OA2H1/mou2sXSMMmqYnfBwjIGsgKqk1TJTjAAZs4eB 0vZ2vhE0iwrKtIYnyqjHAwxoQ1WSVHNG/0Rhk0C8OR6XGX8960r5yZs8Fbw8YEPMVC0i wySJTvNzlG7gXuI/yerlyAw2pBhwuEc18+ZxsBxmULXR8ielfxki5X/kmWHYZtRGiQLE xkxw==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc:content-type; bh=66zdOwURItyDJqQ5+W+vuQ8O8WFy428OvtMHvHZyhAM=; b=QEQcfN2eyG4ah/osfSyOz11wlxPGSNrKO/YP7YBsVBa9KU57onOkkzfJtVQXfuzOsi xesgC/5JCOZigKMF7o77e6RALffhNw8IQjwvH1LnQyVOdIDZkIhxyrHs6OZ8mKOdFN7g XfUrPP90YKSswpNMyZqMTqeN4keFTEAcpocQOL7uc/98Kavf7I+hHt2NS9lTWqRMIvwn q4xvH3kzyyRb79f2gO6xDH3fnzznsFn1JhyWiE0BvOZ6/REc+XTSYmnBx7eHg1A8cuZF fuJ5+8xGZkHD0mBhAVdCb/3cENn7CdkIAGlmK6iRQV7i5C9vf92y1HCLvzccStmVWEoi FULg==
X-Gm-Message-State: ALoCoQmPri6x7bl8uDBwHnkePALyqO39cQmzQp9+VtBP6z/0xuZmcyWgmojsLl3t8zeHX80Ohv87PjWNv99dD+FmzoDVf2vuabEMPDxiFg/MFdKUUbG9RQi3Ejyno89caZJ5tKf6v5Yp48IudnZvGE/TFncyS8arOFL3ZNMGqiiSG3bVKd1eCEGjY2Ov3jrBo5dWKY07Ld4b
X-Received: by 10.180.19.138 with SMTP id f10mr5157011wie.11.1393365922101; Tue, 25 Feb 2014 14:05:22 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Tue, 25 Feb 2014 14:05:00 -0800 (PST)
In-Reply-To: <CABihn6GC4VM2AHza-F7ML=FfHLZu7FNqx+BhbuVsfJLWk0P92w@mail.gmail.com>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <CAH9hSJbBmvNPBSSAk-khdWXgWw0GTt0FG3KsdzYeJcfiAPDk0A@mail.gmail.com> <CAHixhFq=wfmYH8-ij_WtsQLN=NUTJwRQ=k8jCPepQDM8V8ZZYA@mail.gmail.com> <CABihn6EN7V6XEwf6NWn78orxvr3XjGHxROJC4JjQ6RYYKEeCug@mail.gmail.com> <CACuKZqHNoR5GQmWyzbXAszZCOT2P4pjSmT3SF6ZG3X7hTY=1xw@mail.gmail.com> <CABihn6GC4VM2AHza-F7ML=FfHLZu7FNqx+BhbuVsfJLWk0P92w@mail.gmail.com>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Wed, 26 Feb 2014 07:05:00 +0900
Message-ID: <CAH9hSJbSfQ2Abp6oLifi0dx4TZENzm2QRn8zMQfAv=vw+H12sw@mail.gmail.com>
To: Yutaka Hirano <yhirano@chromium.org>, Ian Hickson <ian@hixie.ch>
Content-Type: multipart/alternative; boundary=bcaec53d550dd466c004f3424833
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/-hVPKIRA8Zd_7C3w9EmS9M4akJw
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 25 Feb 2014 22:05:25 -0000

--bcaec53d550dd466c004f3424833
Content-Type: text/plain; charset=UTF-8

On Tue, Feb 25, 2014 at 5:27 PM, Yutaka Hirano <yhirano@chromium.org> wrote:

> Shouldn't ws.close() allow queued messages to be flushed before the
>> actual closing?
>
> It is allowed to wait until the endpoint finishes sending the message.
> Is it the desired behavior? I thought it is undesired, but I'm not
> confident now.
>

I think it's underspecified. In the WHATWG WebSocket API spec, close() is
required to "start the WebSocket closing handshake". RFC 6455 neither says
the algorithm may terminate ongoing "Send a WebSocket Message" algorithm
nor says it may not.

----

Even if we allow close to terminate ongoing send, since the sender requests
the termination explicitly and the peer replies to the close with a close
frame after understanding the ongoing send is cancelled, it's fine to call
it as clean close, I think.

There's no way in the WebSocket API to tell that the message it is
receiving has been cancelled to the script, but I don't think it's
problematic. The message sending was clearly cancelled. Both side knows
that. The receiver just won't deliver the cancelled message.

----

But I think there're also needs for graceful close, i.e. to send a message
and do closing handshake after finishing the send. If there's only method
that terminates ongoing send, the user is required to do app level
acknowledgement before calling the method.

My gut feeling is that such a cancellation should be provides as a separate
method and close() should work as a graceful close.

This needs more discussion, but I believe WebSocket users are already used
to graceful behavior.


> On Tue, Feb 25, 2014 at 2:40 PM, Zhong Yu <zhong.j.yu@gmail.com> wrote:
>
>> On Mon, Feb 24, 2014 at 10:58 PM, Yutaka Hirano <yhirano@chromium.org>
>> wrote:
>> >> * Close: Can be used to perform a "clean" close in the middle of
>> sending a
>> >> message. However, I cannot think of any applications that would
>> consider a
>> >> situation in which a in-transit message was aborted to close the
>> connection
>> >> to be a "clean close" scenario.
>> >
>> > An endpoint should be able to cancel the ongoing transmission, i.e. The
>> > close call should abort transferring very_very_long_message in the
>> following
>> > code.
>> >
>> >
>> > ws = new WebSocket(...);
>> > ws.onopen = function() {
>> >   ws.send(very_very_long_message);
>> >   ws.close();
>> > };
>>
>> Shouldn't ws.close() allow queued messages to be flushed before the
>> actual closing?
>
>

--bcaec53d550dd466c004f3424833
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote">On T=
ue, Feb 25, 2014 at 5:27 PM, Yutaka Hirano <span dir=3D"ltr">&lt;<a href=3D=
"mailto:yhirano@chromium.org" target=3D"_blank">yhirano@chromium.org</a>&gt=
;</span> wrote:<br>


<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div dir=3D"ltr"><div><blockquote class=3D"gmail_quote" st=
yle=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb=
(204,204,204);border-left-style:solid;padding-left:1ex">


<span style=3D"font-family:arial,sans-serif;font-size:13px">Shouldn&#39;t w=
s.close() allow queued messages to be flushed before the<br>

</span><span style=3D"font-family:arial,sans-serif;font-size:13px">actual c=
losing?</span></blockquote></div><div>It is allowed to wait until the endpo=
int finishes sending the message.</div><div>Is it the desired behavior? I t=
hought it is undesired, but I&#39;m not confident now.</div>


</div></blockquote><div><br></div><div>I think it&#39;s underspecified. In =
the WHATWG WebSocket API spec, close() is required to &quot;start the WebSo=
cket closing handshake&quot;. RFC 6455 neither says the algorithm may termi=
nate ongoing &quot;Send a WebSocket Message&quot; algorithm nor says it may=
 not.</div>

<div><br></div><div>----</div><div><br></div><div>Even if we allow close to=
 terminate ongoing send, since the sender requests the termination explicit=
ly and the peer replies to the close with a close frame after understanding=
 the ongoing send is cancelled, it&#39;s fine to call it as clean close, I =
think.</div>

<div><br></div><div>There&#39;s no way in the WebSocket API to tell that th=
e message it is receiving has been cancelled to the script, but I don&#39;t=
 think it&#39;s problematic. The message sending was clearly cancelled. Bot=
h side knows that. The receiver just won&#39;t deliver the cancelled messag=
e.</div>

<div><br></div><div>----</div><div><br></div><div>But I think there&#39;re =
also needs for graceful close, i.e. to send a message and do closing handsh=
ake after finishing the send. If there&#39;s only method that terminates on=
going send, the user is required to do app level acknowledgement before cal=
ling the method.</div>

<div><br></div><div>My gut feeling is that such a cancellation should be pr=
ovides as a separate method and close() should work as a graceful close.</d=
iv><div><br></div><div>This needs more discussion, but I believe WebSocket =
users are already used to graceful behavior.</div>

<div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px =
0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-l=
eft-style:solid;padding-left:1ex"><div dir=3D"ltr"><div><div><div></div><di=
v class=3D"gmail_extra">

<div class=3D"gmail_quote">On Tue, Feb 25, 2014 at 2:40 PM, Zhong Yu <span =
dir=3D"ltr">&lt;<a href=3D"mailto:zhong.j.yu@gmail.com" target=3D"_blank">z=
hong.j.yu@gmail.com</a>&gt;</span> wrote:<br>

<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex">
<div>On Mon, Feb 24, 2014 at 10:58 PM, Yutaka Hirano &lt;<a href=3D"mailto:=
yhirano@chromium.org" target=3D"_blank">yhirano@chromium.org</a>&gt; wrote:=
<br>
&gt;&gt; * Close: Can be used to perform a &quot;clean&quot; close in the m=
iddle of sending a<br>
&gt;&gt; message. However, I cannot think of any applications that would co=
nsider a<br>
&gt;&gt; situation in which a in-transit message was aborted to close the c=
onnection<br>
&gt;&gt; to be a &quot;clean close&quot; scenario.<br>
&gt;<br>
&gt; An endpoint should be able to cancel the ongoing transmission, i.e. Th=
e<br>
&gt; close call should abort transferring very_very_long_message in the fol=
lowing<br>
&gt; code.<br>
&gt;<br>
&gt;<br>
&gt; ws =3D new WebSocket(...);<br>
&gt; ws.onopen =3D function() {<br>
&gt; =C2=A0 ws.send(very_very_long_message);<br>
&gt; =C2=A0 ws.close();<br>
&gt; };<br>
<br>
</div>Shouldn&#39;t ws.close() allow queued messages to be flushed before t=
he<br>
actual closing?</blockquote></div></div></div></div></div></blockquote></di=
v><br></div></div>

--bcaec53d550dd466c004f3424833--


From nobody Tue Feb 25 14:09:03 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 1726B1A07B3 for <hybi@ietfa.amsl.com>; Tue, 25 Feb 2014 14:09:01 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.925
X-Spam-Level: 
X-Spam-Status: No, score=-1.925 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.547, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 0vo2KoWIIpn8 for <hybi@ietfa.amsl.com>; Tue, 25 Feb 2014 14:08:58 -0800 (PST)
Received: from mail-we0-x232.google.com (mail-we0-x232.google.com [IPv6:2a00:1450:400c:c03::232]) by ietfa.amsl.com (Postfix) with ESMTP id D93771A0298 for <hybi@ietf.org>; Tue, 25 Feb 2014 14:08:57 -0800 (PST)
Received: by mail-we0-f178.google.com with SMTP id q59so976250wes.9 for <hybi@ietf.org>; Tue, 25 Feb 2014 14:08:56 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc:content-type; bh=9F+kOiHi0C7z0UpQWBlzdHNCt9Fp+iUwhtqduZyc+jY=; b=dvycqk6okVQ/GKnUHEcVMG81au81LuvIS4U80wJE8ZFVRYC2BMEBeetPUI62sWhN3a SSo8dk6nu0F7jYcdx34DIHO61//3u0vdH+YwZMdLP6vokufvjIfG/Sg3ftWU4Znfg2iz RhmVwrDYNmHO5ms5jRf5EVmfbZX9c1j0Fzsw4TxCkVP+xykndeuNreLmoy+3Rhfuh1qG /dcvZHACBhRrtCYYeilkL92h+KTXrLQ9jRkd66KwfuP9bve++wLqftKRDPxU7+gACCvA JBQZ2uDdGiLeJsR1HVN4VgR/3KJmhae7mo8TdyEa4m13jgYpSCZemkZcJj+wG0JQCj61 vbBw==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc:content-type; bh=9F+kOiHi0C7z0UpQWBlzdHNCt9Fp+iUwhtqduZyc+jY=; b=Ci37nAljGQQYC/3lv1uKf9PVnvrV848Ngc2z04STE/9eSAiiraqJGymopMaTSUdB/e stsH+y+yGn8S0J5gfUKZkMV75JzL9ozWL0dy+4Jh5Ip3KtM1p/ZrPehbye2JwasYc3Mk ycpdSRsjxIvSdt5d3Jfn352xwK2+ZqVwoU07eq6hF/04HU1jjr1yNyLXZknsrIvIFsFN v8wiQi4b/cRbzJM4NrVC0xcYDHsFt9Tkdd+HK7Co+AruzjOzfBw5I34nDh4eGrMIRo1g 9hpNs8APcD0OkjJ8Rp/Ol6l3KUL0oqB0nwMFzkm/6sOD+VOQt8fc5uc3wu8l1NcWOxsz 5oqA==
X-Gm-Message-State: ALoCoQlevCRxsQARExHMubzQLE3Gsa4pk7mzv4r54f3rB3S03FnsAW/90hQp7jxHzZ2re5nC8yu3SkMoDVPj7Re+dv7BTeZpNGJvXjhTgV8nNFYUC1j+3yl84ldb/WnaPOXbI2hCyX0EsdcL3QL50IYzdit5QNwZIfXAJY0j1fPGtVqyRMRwIFBNweTrtkT9sTGOY4F62HZM
X-Received: by 10.180.19.138 with SMTP id f10mr5166934wie.11.1393366136420; Tue, 25 Feb 2014 14:08:56 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Tue, 25 Feb 2014 14:08:36 -0800 (PST)
In-Reply-To: <CAH9hSJbSfQ2Abp6oLifi0dx4TZENzm2QRn8zMQfAv=vw+H12sw@mail.gmail.com>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <CAH9hSJbBmvNPBSSAk-khdWXgWw0GTt0FG3KsdzYeJcfiAPDk0A@mail.gmail.com> <CAHixhFq=wfmYH8-ij_WtsQLN=NUTJwRQ=k8jCPepQDM8V8ZZYA@mail.gmail.com> <CABihn6EN7V6XEwf6NWn78orxvr3XjGHxROJC4JjQ6RYYKEeCug@mail.gmail.com> <CACuKZqHNoR5GQmWyzbXAszZCOT2P4pjSmT3SF6ZG3X7hTY=1xw@mail.gmail.com> <CABihn6GC4VM2AHza-F7ML=FfHLZu7FNqx+BhbuVsfJLWk0P92w@mail.gmail.com> <CAH9hSJbSfQ2Abp6oLifi0dx4TZENzm2QRn8zMQfAv=vw+H12sw@mail.gmail.com>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Wed, 26 Feb 2014 07:08:36 +0900
Message-ID: <CAH9hSJZ9jsp4K7vJ1ZuD1o1HXBtFJwCr_oLMr-mrvH9hAUTVDA@mail.gmail.com>
To: Yutaka Hirano <yhirano@chromium.org>, Ian Hickson <ian@hixie.ch>
Content-Type: multipart/alternative; boundary=bcaec53d550d9a8d1404f3425565
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/MgxBf7wOPlebh64Ae4346LNgQPo
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Tue, 25 Feb 2014 22:09:01 -0000

--bcaec53d550d9a8d1404f3425565
Content-Type: text/plain; charset=UTF-8

On Wed, Feb 26, 2014 at 7:05 AM, Takeshi Yoshino <tyoshino@google.com>wrote:

> On Tue, Feb 25, 2014 at 5:27 PM, Yutaka Hirano <yhirano@chromium.org>wrote:
>
>> Shouldn't ws.close() allow queued messages to be flushed before the
>>> actual closing?
>>
>> It is allowed to wait until the endpoint finishes sending the message.
>> Is it the desired behavior? I thought it is undesired, but I'm not
>> confident now.
>>
>
> I think it's underspecified. In the WHATWG WebSocket API spec, close() is
> required to "start the WebSocket closing handshake". RFC 6455 neither says
> the algorithm may terminate ongoing "Send a WebSocket Message" algorithm
> nor says it may not.
>
> ----
>
> Even if we allow close to terminate ongoing send, since the sender
> requests the termination explicitly and the peer replies to the close with
> a close frame after understanding the ongoing send is cancelled, it's fine
> to call it as clean close, I think.
>
> There's no way in the WebSocket API to tell that the message it is
> receiving has been cancelled to the script, but I don't think it's
> problematic. The message sending was clearly cancelled. Both side knows
> that. The receiver just won't deliver the cancelled message.
>

Also note that a close frame may be sent unsolicitedly by the peer and may
result in CloseEvent with wasClean set.

--bcaec53d550d9a8d1404f3425565
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote">On W=
ed, Feb 26, 2014 at 7:05 AM, Takeshi Yoshino <span dir=3D"ltr">&lt;<a href=
=3D"mailto:tyoshino@google.com" target=3D"_blank">tyoshino@google.com</a>&g=
t;</span> wrote:<br>

<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div dir=3D"ltr"><div class=3D"gmail_extra">=
<div class=3D"gmail_quote"><div class=3D"">On Tue, Feb 25, 2014 at 5:27 PM,=
 Yutaka Hirano <span dir=3D"ltr">&lt;<a href=3D"mailto:yhirano@chromium.org=
" target=3D"_blank">yhirano@chromium.org</a>&gt;</span> wrote:<br>



<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div dir=3D"ltr"><div><blockquote class=3D"gmail_quote" st=
yle=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb=
(204,204,204);border-left-style:solid;padding-left:1ex">



<span style=3D"font-family:arial,sans-serif;font-size:13px">Shouldn&#39;t w=
s.close() allow queued messages to be flushed before the<br>

</span><span style=3D"font-family:arial,sans-serif;font-size:13px">actual c=
losing?</span></blockquote></div><div>It is allowed to wait until the endpo=
int finishes sending the message.</div><div>Is it the desired behavior? I t=
hought it is undesired, but I&#39;m not confident now.</div>



</div></blockquote><div><br></div></div><div>I think it&#39;s underspecifie=
d. In the WHATWG WebSocket API spec, close() is required to &quot;start the=
 WebSocket closing handshake&quot;. RFC 6455 neither says the algorithm may=
 terminate ongoing &quot;Send a WebSocket Message&quot; algorithm nor says =
it may not.</div>


<div><br></div><div>----</div><div><br></div><div>Even if we allow close to=
 terminate ongoing send, since the sender requests the termination explicit=
ly and the peer replies to the close with a close frame after understanding=
 the ongoing send is cancelled, it&#39;s fine to call it as clean close, I =
think.</div>


<div><br></div><div>There&#39;s no way in the WebSocket API to tell that th=
e message it is receiving has been cancelled to the script, but I don&#39;t=
 think it&#39;s problematic. The message sending was clearly cancelled. Bot=
h side knows that. The receiver just won&#39;t deliver the cancelled messag=
e.</div>

</div></div></div></blockquote><div><br></div><div>Also note that a close f=
rame may be sent unsolicitedly by the peer and may result in CloseEvent wit=
h wasClean set.</div></div></div></div>

--bcaec53d550d9a8d1404f3425565--


From nobody Tue Feb 25 16:54:27 2014
Return-Path: <zhong.j.yu@gmail.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 9D5A71A0347 for <hybi@ietfa.amsl.com>; Tue, 25 Feb 2014 16:54:18 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 0.6
X-Spam-Level: 
X-Spam-Status: No, score=0.6 tagged_above=-999 required=5 tests=[BAYES_05=-0.5, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FREEMAIL_FROM=0.001, J_CHICKENPOX_26=0.6, J_CHICKENPOX_52=0.6, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id ieX6xrY3DxaY for <hybi@ietfa.amsl.com>; Tue, 25 Feb 2014 16:54:17 -0800 (PST)
Received: from mail-oa0-x22a.google.com (mail-oa0-x22a.google.com [IPv6:2607:f8b0:4003:c02::22a]) by ietfa.amsl.com (Postfix) with ESMTP id 5C4FC1A020A for <hybi@ietf.org>; Tue, 25 Feb 2014 16:54:17 -0800 (PST)
Received: by mail-oa0-f42.google.com with SMTP id j17so107990oag.29 for <hybi@ietf.org>; Tue, 25 Feb 2014 16:54:16 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113;  h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=YLzhFm/xu7N3OF70v4O2+Qv4sXxmYVoyWk3v4Dusxr4=; b=ozpDJat38JQqzHpbQAPd6+mY2c5xKpdkVg48k1OYQbIMCAIQ70NuDGCgHAb8BbvAEU oslBPjkLa6u+KwHs4d/OgjNbLplh37J6sUNN8tQEDKKGpsubxiyHJNPnZW+Es1CkdpCy 3UrRvdErOT70w73lMEWSNYYJMkxGUtFPUPZrfAdrWuDgH3imlVt2WdZe1pdLxaIO+U+1 NFZiIW+BfRjBbe6U55Jgi0l1kkxRK87mp6LmK9exw4SAifnwH2FwEBFwUPJRdoWBghJI AXQFTWDDyWqPFpfNSvZNqmduk7nlGsHA4D/U3BVzVma5tsRM6BvC0z7O60NGNABKHSUz +36w==
MIME-Version: 1.0
X-Received: by 10.60.98.73 with SMTP id eg9mr49145oeb.81.1393376056283; Tue, 25 Feb 2014 16:54:16 -0800 (PST)
Received: by 10.76.106.162 with HTTP; Tue, 25 Feb 2014 16:54:16 -0800 (PST)
In-Reply-To: <CAH9hSJbSfQ2Abp6oLifi0dx4TZENzm2QRn8zMQfAv=vw+H12sw@mail.gmail.com>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <CAH9hSJbBmvNPBSSAk-khdWXgWw0GTt0FG3KsdzYeJcfiAPDk0A@mail.gmail.com> <CAHixhFq=wfmYH8-ij_WtsQLN=NUTJwRQ=k8jCPepQDM8V8ZZYA@mail.gmail.com> <CABihn6EN7V6XEwf6NWn78orxvr3XjGHxROJC4JjQ6RYYKEeCug@mail.gmail.com> <CACuKZqHNoR5GQmWyzbXAszZCOT2P4pjSmT3SF6ZG3X7hTY=1xw@mail.gmail.com> <CABihn6GC4VM2AHza-F7ML=FfHLZu7FNqx+BhbuVsfJLWk0P92w@mail.gmail.com> <CAH9hSJbSfQ2Abp6oLifi0dx4TZENzm2QRn8zMQfAv=vw+H12sw@mail.gmail.com>
Date: Tue, 25 Feb 2014 18:54:16 -0600
Message-ID: <CACuKZqGhBN5SnuUEyRccaqQiBP1tE8r=K+9ioFGikG60a9bFyQ@mail.gmail.com>
From: Zhong Yu <zhong.j.yu@gmail.com>
To: Takeshi Yoshino <tyoshino@google.com>
Content-Type: text/plain; charset=ISO-8859-1
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/vdm2DXn8opPXkivasBXwrbJJex4
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 26 Feb 2014 00:54:19 -0000

To be certain that data is delivered, app needs to perform lingering
close by itself, by polling bufferedAmount, and only call close()
after it reaches 0.

This is strange, due to the flaw that send() is an async method but
without a way to registered a callback.

Zhong Yu


On Tue, Feb 25, 2014 at 4:05 PM, Takeshi Yoshino <tyoshino@google.com> wrote:
> On Tue, Feb 25, 2014 at 5:27 PM, Yutaka Hirano <yhirano@chromium.org> wrote:
>>>
>>> Shouldn't ws.close() allow queued messages to be flushed before the
>>> actual closing?
>>
>> It is allowed to wait until the endpoint finishes sending the message.
>> Is it the desired behavior? I thought it is undesired, but I'm not
>> confident now.
>
>
> I think it's underspecified. In the WHATWG WebSocket API spec, close() is
> required to "start the WebSocket closing handshake". RFC 6455 neither says
> the algorithm may terminate ongoing "Send a WebSocket Message" algorithm nor
> says it may not.
>
> ----
>
> Even if we allow close to terminate ongoing send, since the sender requests
> the termination explicitly and the peer replies to the close with a close
> frame after understanding the ongoing send is cancelled, it's fine to call
> it as clean close, I think.
>
> There's no way in the WebSocket API to tell that the message it is receiving
> has been cancelled to the script, but I don't think it's problematic. The
> message sending was clearly cancelled. Both side knows that. The receiver
> just won't deliver the cancelled message.
>
> ----
>
> But I think there're also needs for graceful close, i.e. to send a message
> and do closing handshake after finishing the send. If there's only method
> that terminates ongoing send, the user is required to do app level
> acknowledgement before calling the method.
>
> My gut feeling is that such a cancellation should be provides as a separate
> method and close() should work as a graceful close.
>
> This needs more discussion, but I believe WebSocket users are already used
> to graceful behavior.
>
>>
>> On Tue, Feb 25, 2014 at 2:40 PM, Zhong Yu <zhong.j.yu@gmail.com> wrote:
>>>
>>> On Mon, Feb 24, 2014 at 10:58 PM, Yutaka Hirano <yhirano@chromium.org>
>>> wrote:
>>> >> * Close: Can be used to perform a "clean" close in the middle of
>>> >> sending a
>>> >> message. However, I cannot think of any applications that would
>>> >> consider a
>>> >> situation in which a in-transit message was aborted to close the
>>> >> connection
>>> >> to be a "clean close" scenario.
>>> >
>>> > An endpoint should be able to cancel the ongoing transmission, i.e. The
>>> > close call should abort transferring very_very_long_message in the
>>> > following
>>> > code.
>>> >
>>> >
>>> > ws = new WebSocket(...);
>>> > ws.onopen = function() {
>>> >   ws.send(very_very_long_message);
>>> >   ws.close();
>>> > };
>>>
>>> Shouldn't ws.close() allow queued messages to be flushed before the
>>> actual closing?
>
>


From nobody Tue Feb 25 17:38:18 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id A31FA1A0834 for <hybi@ietfa.amsl.com>; Tue, 25 Feb 2014 17:38:10 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 0.675
X-Spam-Level: 
X-Spam-Status: No, score=0.675 tagged_above=-999 required=5 tests=[BAYES_05=-0.5, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, J_CHICKENPOX_26=0.6, J_CHICKENPOX_52=0.6, RP_MATCHES_RCVD=-0.547, SPF_PASS=-0.001] autolearn=no
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 7J4dRVq7dKkT for <hybi@ietfa.amsl.com>; Tue, 25 Feb 2014 17:38:08 -0800 (PST)
Received: from mail-wi0-x22b.google.com (mail-wi0-x22b.google.com [IPv6:2a00:1450:400c:c05::22b]) by ietfa.amsl.com (Postfix) with ESMTP id 21D7C1A0313 for <hybi@ietf.org>; Tue, 25 Feb 2014 17:38:07 -0800 (PST)
Received: by mail-wi0-f171.google.com with SMTP id cc10so5292474wib.10 for <hybi@ietf.org>; Tue, 25 Feb 2014 17:38:06 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc:content-type; bh=/CJTx7dLmpyofTE9UuF5mV8q3NNAIN/g/h2HBqqrxQ0=; b=Qdfq8FPen6za49v88uP/ezBVZ27HiydxcdtNf1HPQgm4tvx7JE4RIX+zWf0nYcpMI+ RN8cDOcRlJ6/fm8TUscwObuZJvPw6TBUEVgCAsJhsU02vyvzSvrNeP1lFuGMgUL5Wk4Y QwxbugdfXjP4luGeostqDU8hwPxQGF1lwbnLFYJ7JP8dGzXnVz5YJPxAA58bJfYOn225 veTpEjnJykhj3XdAzGVfNjIhbTkXC4tTT0y08+9QJQqYAoQHlw+IcUrag5As0o7UKcA4 ObIIEjXDoo+nWmYeU+DE+Y6S+iq2IpdA4Gp3hYiV9c//af66ZewQSQmvbnJ7Ly1DwyQG qUSA==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc:content-type; bh=/CJTx7dLmpyofTE9UuF5mV8q3NNAIN/g/h2HBqqrxQ0=; b=NTp2hl8sSMulVTukMBHhg0nZuqS4kov15WCTesPzS1lABZMleKOANU9wmq/5BBYadB Y8wqLcgB8Qs+2KWqnWMYbjgsn2QZ2pnHQP0dVtbtlrnOIsqwaVViwW4Q6QbZZgfwkuLB hLm1Qh347X/rFh0u2zxIt6wXxzK7tv3ZoH7M90RW/GNEiAWszJPVJJo3Q9qCoPfZPLl7 oOdk/DMR9X4xd+4bnq07poYhRmg+XZyGdGCwxIN47/xdlYHTi1PW2kWUBipBtGmpPUQ8 tnyL11P4bTCPiy+E78PnNaScM+raLJRgsPiYgvsgnRtW+rirH2LFRPzauhi2W15S30kt 0ATQ==
X-Gm-Message-State: ALoCoQkvlANCcUw6QPtTsdYJflu63dZ769da5gjG/z+faBVkiJ9dFxG0ntqG47llqJ8n+SMSvs/QmQHTIiEwfmR0xfOJcr/GBgIDx/z668bVeGbYmIwTjwwtjyZOG+hYYEk2CXgti0Ozqkim9npP+4jsrydgUXG6kbKrO5OMNINS73x+5hEdM1EJbhteRUfaAChQHsPQ976S
X-Received: by 10.180.221.68 with SMTP id qc4mr5557400wic.30.1393378686553; Tue, 25 Feb 2014 17:38:06 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Tue, 25 Feb 2014 17:37:46 -0800 (PST)
In-Reply-To: <CACuKZqGhBN5SnuUEyRccaqQiBP1tE8r=K+9ioFGikG60a9bFyQ@mail.gmail.com>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <CAH9hSJbBmvNPBSSAk-khdWXgWw0GTt0FG3KsdzYeJcfiAPDk0A@mail.gmail.com> <CAHixhFq=wfmYH8-ij_WtsQLN=NUTJwRQ=k8jCPepQDM8V8ZZYA@mail.gmail.com> <CABihn6EN7V6XEwf6NWn78orxvr3XjGHxROJC4JjQ6RYYKEeCug@mail.gmail.com> <CACuKZqHNoR5GQmWyzbXAszZCOT2P4pjSmT3SF6ZG3X7hTY=1xw@mail.gmail.com> <CABihn6GC4VM2AHza-F7ML=FfHLZu7FNqx+BhbuVsfJLWk0P92w@mail.gmail.com> <CAH9hSJbSfQ2Abp6oLifi0dx4TZENzm2QRn8zMQfAv=vw+H12sw@mail.gmail.com> <CACuKZqGhBN5SnuUEyRccaqQiBP1tE8r=K+9ioFGikG60a9bFyQ@mail.gmail.com>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Wed, 26 Feb 2014 10:37:46 +0900
Message-ID: <CAH9hSJZh1KwTPqrn+vvTOG4878foaXChPrgTf01zaJPRmsKdsw@mail.gmail.com>
To: Zhong Yu <zhong.j.yu@gmail.com>
Content-Type: multipart/alternative; boundary=001a1134d69ca672fb04f345410c
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/BzNTxUsRCNf23ZHRqa5PWit6ACs
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 26 Feb 2014 01:38:10 -0000

--001a1134d69ca672fb04f345410c
Content-Type: text/plain; charset=UTF-8

On Wed, Feb 26, 2014 at 9:54 AM, Zhong Yu <zhong.j.yu@gmail.com> wrote:

> To be certain that data is delivered, app needs to perform lingering
> close by itself, by polling bufferedAmount, and only call close()
> after it reaches 0.
>

The "queue" referred by some texts in the WebSocket API spec is also
underspecified. It doesn't say the queue is not used by close frame
sending. Chromium uses the same queue. There can be implementations that
allow a close frame overtake buffered data. Then, you're right.


> This is strange, due to the flaw that send() is an async method but
> without a way to registered a callback.
>

The API should have some method to get notified of full/partial drain of
the buffer. Also, it's said to be an issue that there's no way to know how
much the upper limit of buffer size is on the UA in use.


>
> Zhong Yu
>
>
> On Tue, Feb 25, 2014 at 4:05 PM, Takeshi Yoshino <tyoshino@google.com>
> wrote:
> > On Tue, Feb 25, 2014 at 5:27 PM, Yutaka Hirano <yhirano@chromium.org>
> wrote:
> >>>
> >>> Shouldn't ws.close() allow queued messages to be flushed before the
> >>> actual closing?
> >>
> >> It is allowed to wait until the endpoint finishes sending the message.
> >> Is it the desired behavior? I thought it is undesired, but I'm not
> >> confident now.
> >
> >
> > I think it's underspecified. In the WHATWG WebSocket API spec, close() is
> > required to "start the WebSocket closing handshake". RFC 6455 neither
> says
> > the algorithm may terminate ongoing "Send a WebSocket Message" algorithm
> nor
> > says it may not.
> >
> > ----
> >
> > Even if we allow close to terminate ongoing send, since the sender
> requests
> > the termination explicitly and the peer replies to the close with a close
> > frame after understanding the ongoing send is cancelled, it's fine to
> call
> > it as clean close, I think.
> >
> > There's no way in the WebSocket API to tell that the message it is
> receiving
> > has been cancelled to the script, but I don't think it's problematic. The
> > message sending was clearly cancelled. Both side knows that. The receiver
> > just won't deliver the cancelled message.
> >
> > ----
> >
> > But I think there're also needs for graceful close, i.e. to send a
> message
> > and do closing handshake after finishing the send. If there's only method
> > that terminates ongoing send, the user is required to do app level
> > acknowledgement before calling the method.
> >
> > My gut feeling is that such a cancellation should be provides as a
> separate
> > method and close() should work as a graceful close.
> >
> > This needs more discussion, but I believe WebSocket users are already
> used
> > to graceful behavior.
> >
> >>
> >> On Tue, Feb 25, 2014 at 2:40 PM, Zhong Yu <zhong.j.yu@gmail.com> wrote:
> >>>
> >>> On Mon, Feb 24, 2014 at 10:58 PM, Yutaka Hirano <yhirano@chromium.org>
> >>> wrote:
> >>> >> * Close: Can be used to perform a "clean" close in the middle of
> >>> >> sending a
> >>> >> message. However, I cannot think of any applications that would
> >>> >> consider a
> >>> >> situation in which a in-transit message was aborted to close the
> >>> >> connection
> >>> >> to be a "clean close" scenario.
> >>> >
> >>> > An endpoint should be able to cancel the ongoing transmission, i.e.
> The
> >>> > close call should abort transferring very_very_long_message in the
> >>> > following
> >>> > code.
> >>> >
> >>> >
> >>> > ws = new WebSocket(...);
> >>> > ws.onopen = function() {
> >>> >   ws.send(very_very_long_message);
> >>> >   ws.close();
> >>> > };
> >>>
> >>> Shouldn't ws.close() allow queued messages to be flushed before the
> >>> actual closing?
> >
> >
>

--001a1134d69ca672fb04f345410c
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote">On W=
ed, Feb 26, 2014 at 9:54 AM, Zhong Yu <span dir=3D"ltr">&lt;<a href=3D"mail=
to:zhong.j.yu@gmail.com" target=3D"_blank">zhong.j.yu@gmail.com</a>&gt;</sp=
an> wrote:<br>

<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">To be certain that data is delivered, app ne=
eds to perform lingering<br>
close by itself, by polling bufferedAmount, and only call close()<br>
after it reaches 0.<br></blockquote><div><br></div><div>The &quot;queue&quo=
t; referred by some texts in the WebSocket API spec is also underspecified.=
 It doesn&#39;t say the queue is not used by close frame sending. Chromium =
uses the same queue. There can be implementations that allow a close frame =
overtake buffered data. Then, you&#39;re right.</div>

<div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8=
ex;border-left:1px #ccc solid;padding-left:1ex">This is strange, due to the=
 flaw that send() is an async method but<br>
without a way to registered a callback.<br></blockquote><div><br></div><div=
>The API should have some method to get notified of full/partial drain of t=
he buffer. Also, it&#39;s said to be an issue that there&#39;s no way to kn=
ow how much the upper limit of buffer size is on the UA in use.</div>

<div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8=
ex;border-left:1px #ccc solid;padding-left:1ex">
<span class=3D"HOEnZb"><font color=3D"#888888"><br>
Zhong Yu<br>
</font></span><div class=3D"HOEnZb"><div class=3D"h5"><br>
<br>
On Tue, Feb 25, 2014 at 4:05 PM, Takeshi Yoshino &lt;<a href=3D"mailto:tyos=
hino@google.com">tyoshino@google.com</a>&gt; wrote:<br>
&gt; On Tue, Feb 25, 2014 at 5:27 PM, Yutaka Hirano &lt;<a href=3D"mailto:y=
hirano@chromium.org">yhirano@chromium.org</a>&gt; wrote:<br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt; Shouldn&#39;t ws.close() allow queued messages to be flushed b=
efore the<br>
&gt;&gt;&gt; actual closing?<br>
&gt;&gt;<br>
&gt;&gt; It is allowed to wait until the endpoint finishes sending the mess=
age.<br>
&gt;&gt; Is it the desired behavior? I thought it is undesired, but I&#39;m=
 not<br>
&gt;&gt; confident now.<br>
&gt;<br>
&gt;<br>
&gt; I think it&#39;s underspecified. In the WHATWG WebSocket API spec, clo=
se() is<br>
&gt; required to &quot;start the WebSocket closing handshake&quot;. RFC 645=
5 neither says<br>
&gt; the algorithm may terminate ongoing &quot;Send a WebSocket Message&quo=
t; algorithm nor<br>
&gt; says it may not.<br>
&gt;<br>
&gt; ----<br>
&gt;<br>
&gt; Even if we allow close to terminate ongoing send, since the sender req=
uests<br>
&gt; the termination explicitly and the peer replies to the close with a cl=
ose<br>
&gt; frame after understanding the ongoing send is cancelled, it&#39;s fine=
 to call<br>
&gt; it as clean close, I think.<br>
&gt;<br>
&gt; There&#39;s no way in the WebSocket API to tell that the message it is=
 receiving<br>
&gt; has been cancelled to the script, but I don&#39;t think it&#39;s probl=
ematic. The<br>
&gt; message sending was clearly cancelled. Both side knows that. The recei=
ver<br>
&gt; just won&#39;t deliver the cancelled message.<br>
&gt;<br>
&gt; ----<br>
&gt;<br>
&gt; But I think there&#39;re also needs for graceful close, i.e. to send a=
 message<br>
&gt; and do closing handshake after finishing the send. If there&#39;s only=
 method<br>
&gt; that terminates ongoing send, the user is required to do app level<br>
&gt; acknowledgement before calling the method.<br>
&gt;<br>
&gt; My gut feeling is that such a cancellation should be provides as a sep=
arate<br>
&gt; method and close() should work as a graceful close.<br>
&gt;<br>
&gt; This needs more discussion, but I believe WebSocket users are already =
used<br>
&gt; to graceful behavior.<br>
&gt;<br>
&gt;&gt;<br>
&gt;&gt; On Tue, Feb 25, 2014 at 2:40 PM, Zhong Yu &lt;<a href=3D"mailto:zh=
ong.j.yu@gmail.com">zhong.j.yu@gmail.com</a>&gt; wrote:<br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt; On Mon, Feb 24, 2014 at 10:58 PM, Yutaka Hirano &lt;<a href=3D=
"mailto:yhirano@chromium.org">yhirano@chromium.org</a>&gt;<br>
&gt;&gt;&gt; wrote:<br>
&gt;&gt;&gt; &gt;&gt; * Close: Can be used to perform a &quot;clean&quot; c=
lose in the middle of<br>
&gt;&gt;&gt; &gt;&gt; sending a<br>
&gt;&gt;&gt; &gt;&gt; message. However, I cannot think of any applications =
that would<br>
&gt;&gt;&gt; &gt;&gt; consider a<br>
&gt;&gt;&gt; &gt;&gt; situation in which a in-transit message was aborted t=
o close the<br>
&gt;&gt;&gt; &gt;&gt; connection<br>
&gt;&gt;&gt; &gt;&gt; to be a &quot;clean close&quot; scenario.<br>
&gt;&gt;&gt; &gt;<br>
&gt;&gt;&gt; &gt; An endpoint should be able to cancel the ongoing transmis=
sion, i.e. The<br>
&gt;&gt;&gt; &gt; close call should abort transferring very_very_long_messa=
ge in the<br>
&gt;&gt;&gt; &gt; following<br>
&gt;&gt;&gt; &gt; code.<br>
&gt;&gt;&gt; &gt;<br>
&gt;&gt;&gt; &gt;<br>
&gt;&gt;&gt; &gt; ws =3D new WebSocket(...);<br>
&gt;&gt;&gt; &gt; ws.onopen =3D function() {<br>
&gt;&gt;&gt; &gt; =C2=A0 ws.send(very_very_long_message);<br>
&gt;&gt;&gt; &gt; =C2=A0 ws.close();<br>
&gt;&gt;&gt; &gt; };<br>
&gt;&gt;&gt;<br>
&gt;&gt;&gt; Shouldn&#39;t ws.close() allow queued messages to be flushed b=
efore the<br>
&gt;&gt;&gt; actual closing?<br>
&gt;<br>
&gt;<br>
</div></div></blockquote></div><br></div></div>

--001a1134d69ca672fb04f345410c--


From nobody Wed Feb 26 04:50:15 2014
Return-Path: <tobias.oberstein@tavendo.de>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 657A51A02B6 for <hybi@ietfa.amsl.com>; Wed, 26 Feb 2014 04:50:08 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 0
X-Spam-Level: 
X-Spam-Status: No, score=0 tagged_above=-999 required=5 tests=[BAYES_20=-0.001, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_NONE=-0.0001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 9S2LikvrXo8n for <hybi@ietfa.amsl.com>; Wed, 26 Feb 2014 04:50:03 -0800 (PST)
Received: from EXHUB020-1.exch020.serverdata.net (exhub020-1.exch020.serverdata.net [206.225.164.28]) by ietfa.amsl.com (Postfix) with ESMTP id 5DF701A02EE for <hybi@ietf.org>; Wed, 26 Feb 2014 04:50:02 -0800 (PST)
Received: from EXVMBX020-12.exch020.serverdata.net ([169.254.3.76]) by EXHUB020-1.exch020.serverdata.net ([206.225.164.28]) with mapi; Wed, 26 Feb 2014 04:50:00 -0800
From: Tobias Oberstein <tobias.oberstein@tavendo.de>
To: Takeshi Yoshino <tyoshino@google.com>
Date: Wed, 26 Feb 2014 04:49:57 -0800
Thread-Topic: Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: [hybi] Discontinuation of mux ...)
Thread-Index: Ac8xsM59n4hhiFqCTMaLHxyGygBHRgBOnt7Q
Message-ID: <634914A010D0B943A035D226786325D4447358E224@EXVMBX020-12.exch020.serverdata.net>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <634914A010D0B943A035D226786325D4446C537686@EXVMBX020-12.exch020.serverdata.net> <CAH9hSJaALTNiC9rUC-qVYKK74_YJA5psvGhbsFegTwaLsg9Dug@mail.gmail.com>
In-Reply-To: <CAH9hSJaALTNiC9rUC-qVYKK74_YJA5psvGhbsFegTwaLsg9Dug@mail.gmail.com>
Accept-Language: de-DE, en-US
Content-Language: de-DE
X-MS-Has-Attach: 
X-MS-TNEF-Correlator: 
acceptlanguage: de-DE, en-US
Content-Type: multipart/alternative; boundary="_000_634914A010D0B943A035D226786325D4447358E224EXVMBX02012ex_"
MIME-Version: 1.0
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/Jiv7B9WfWvm-flMWo3A-mb3QnYo
Cc: "hybi@ietf.org" <hybi@ietf.org>, Yutaka Hirano <yhirano@google.com>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 26 Feb 2014 12:50:08 -0000

--_000_634914A010D0B943A035D226786325D4447358E224EXVMBX02012ex_
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: base64
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--_000_634914A010D0B943A035D226786325D4447358E224EXVMBX02012ex_
Content-Type: text/html; charset="utf-8"
Content-Transfer-Encoding: base64
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--_000_634914A010D0B943A035D226786325D4447358E224EXVMBX02012ex_--


From nobody Wed Feb 26 09:35:23 2014
Return-Path: <ian@hixie.ch>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 788A11A00E8 for <hybi@ietfa.amsl.com>; Wed, 26 Feb 2014 09:35:16 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2
X-Spam-Level: 
X-Spam-Status: No, score=-2 tagged_above=-999 required=5 tests=[BAYES_00=-1.9,  DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id DRRYAm9OcHyO for <hybi@ietfa.amsl.com>; Wed, 26 Feb 2014 09:35:11 -0800 (PST)
Received: from homiemail-a45.g.dreamhost.com (caibbdcaacfa.dreamhost.com [208.113.200.250]) by ietfa.amsl.com (Postfix) with ESMTP id D9A621A0074 for <hybi@ietf.org>; Wed, 26 Feb 2014 09:35:11 -0800 (PST)
Received: from homiemail-a45.g.dreamhost.com (localhost [127.0.0.1]) by homiemail-a45.g.dreamhost.com (Postfix) with ESMTP id BA7F3480D0; Wed, 26 Feb 2014 09:35:10 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha1; c=relaxed; d=hixie.ch; h=date:from:to :cc:subject:in-reply-to:message-id:references:mime-version: content-type; s=hixie.ch; bh=dPDqUcDW+ykRVL7plGasjA8nGF8=; b=yue /0v5MQQ8dtnpRUSSWy7LFOx5z4akem/rzh1aSXlly/Ymkjdobm01sM1SqailSPFa hMsZ3Yzyx8u1+9CMBwfT6ue8KSZf2KzEhR7MMul1K/ssx0CHKCTAfE+eGgQ8sNAA HpISYv/o4meEgO9UAbT2pAeSg+CIEJJeJCz+Osys=
Received: from ps20323.dreamhostps.com (ps20323.dreamhost.com [208.113.236.128]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) (Authenticated sender: internal@index.hixie.ch) by homiemail-a45.g.dreamhost.com (Postfix) with ESMTPSA id AA5B14809C; Wed, 26 Feb 2014 09:35:10 -0800 (PST)
Date: Wed, 26 Feb 2014 17:35:10 +0000 (UTC)
From: Ian Hickson <ian@hixie.ch>
To: Takeshi Yoshino <tyoshino@google.com>
In-Reply-To: <CAH9hSJbSfQ2Abp6oLifi0dx4TZENzm2QRn8zMQfAv=vw+H12sw@mail.gmail.com>
Message-ID: <alpine.DEB.2.00.1402261734120.31525@ps20323.dreamhostps.com>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <CAH9hSJbBmvNPBSSAk-khdWXgWw0GTt0FG3KsdzYeJcfiAPDk0A@mail.gmail.com> <CAHixhFq=wfmYH8-ij_WtsQLN=NUTJwRQ=k8jCPepQDM8V8ZZYA@mail.gmail.com> <CABihn6EN7V6XEwf6NWn78orxvr3XjGHxROJC4JjQ6RYYKEeCug@mail.gmail.com> <CACuKZqHNoR5GQmWyzbXAszZCOT2P4pjSmT3SF6ZG3X7hTY=1xw@mail.gmail.com> <CABihn6GC4VM2AHza-F7ML=FfHLZu7FNqx+BhbuVsfJLWk0P92w@mail.gmail.com> <CAH9hSJbSfQ2Abp6oLifi0dx4TZENzm2QRn8zMQfAv=vw+H12sw@mail.gmail.com>
User-Agent: Alpine 2.00 (DEB 1167 2008-08-23)
Content-Language: en-GB-hixie
Content-Style-Type: text/css
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/3jJY-6cTHtAY0Dd3ekKUcZlD2ZM
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 26 Feb 2014 17:35:16 -0000

On Wed, 26 Feb 2014, Takeshi Yoshino wrote:
> 
> I think it's underspecified. In the WHATWG WebSocket API spec, close() 
> is required to "start the WebSocket closing handshake". RFC 6455 neither 
> says the algorithm may terminate ongoing "Send a WebSocket Message" 
> algorithm nor says it may not.

If it doesn't allow it, then it's not allowed.

If you send a bunch of messages, then start the closing handshake, then 
the messages should all be sent. I don't think there's anything in either 
spec that would allow the handshake to jump ahead.

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'


From nobody Wed Feb 26 09:53:21 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 85D341A0761 for <hybi@ietfa.amsl.com>; Wed, 26 Feb 2014 09:53:12 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.925
X-Spam-Level: 
X-Spam-Status: No, score=-1.925 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.547, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id LMb4815399HH for <hybi@ietfa.amsl.com>; Wed, 26 Feb 2014 09:53:08 -0800 (PST)
Received: from mail-wi0-x230.google.com (mail-wi0-x230.google.com [IPv6:2a00:1450:400c:c05::230]) by ietfa.amsl.com (Postfix) with ESMTP id 5A79F1A0739 for <hybi@ietf.org>; Wed, 26 Feb 2014 09:53:08 -0800 (PST)
Received: by mail-wi0-f176.google.com with SMTP id hi5so6229167wib.3 for <hybi@ietf.org>; Wed, 26 Feb 2014 09:53:06 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc:content-type; bh=R1J6QjTRPLxq1PO/NBAZD0QSQDfsMWElpPPYCH5Jfq8=; b=M6h48IVy6BSkVnCJ4TSq6+YnMXQm0kOkfg+Gn/5HZtbTNRe5fE0vLjeqCiAT5afTxP kv1anY+oqUQ/bzUmtnnfr+oHr/borq30/iTaEutasgtDwEE5u47qqLdO39mnULeiOkqc S8ZvgE7xj3sGuNxtN67YcU4KgtwhhIt2GY8V92GisWs657XMqoEs6vROho7vET5pFhi8 Gnpq6mshkGLw2ejYBnlsNHARTjlHsBPh748R4tvCjB1+xNpuHq1oE6o6V9t5TilV5A72 2KgH01msK4Ir/UnEsg2WDzWSp5pcqkHyAfwOYlYgtz3npFA5bBuWeEMihYmBRNLiKdvA mr3g==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc:content-type; bh=R1J6QjTRPLxq1PO/NBAZD0QSQDfsMWElpPPYCH5Jfq8=; b=OfDpP7+4S0EcggiVPBeU8jDTo503oTlPbnoM/Z+YpAOrhNCFWaFpRiwtYPN+B3wAmF RDTwB00v05+S+4/8un1EsywaCmGFUNpAxMyq6BfGQKzsOVFf4Q3g6QlK9iS8IeHWptxL mYS6jxaoC6IdFsyt9SqRBqc8VBPj6IUioQePRICz+Uoc00j7dn68X0WHHKJSI7R6A6k6 GU+/S0SZko6Pf12xC4IlEdtN4rAgvv+c8JG2AlgziQ/XTRhN6sUyR0gjtic0FSw3HK9T 3SR8cjTUITXhbyt4VgPVqZNG00O9aMJ67SNQOcLwvLJEes5eAlZI8zQU2+RI3gSo1hjc gBfA==
X-Gm-Message-State: ALoCoQls4nidp1Jbt2R/t+T0ZUZNIQ7IEDo0WXEx1Fc+6XB+MVINZYCvPZZVUVFJ0jpKPeI7cGu0b/D2Ld/9ebS3hlyRI+WZcdKUKydTYWJoCGy+C9yZrYbOFxx+jaOJEd0QjupdrNuf2DGnOFEosfWIE08xCbJxjF11Bb4nzi3Y0oblhM/B3YlFm0JRm/r5CT1Hd8CNfzTr
X-Received: by 10.194.6.164 with SMTP id c4mr3674738wja.38.1393437186368; Wed, 26 Feb 2014 09:53:06 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Wed, 26 Feb 2014 09:52:46 -0800 (PST)
In-Reply-To: <alpine.DEB.2.00.1402261734120.31525@ps20323.dreamhostps.com>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <CAH9hSJbBmvNPBSSAk-khdWXgWw0GTt0FG3KsdzYeJcfiAPDk0A@mail.gmail.com> <CAHixhFq=wfmYH8-ij_WtsQLN=NUTJwRQ=k8jCPepQDM8V8ZZYA@mail.gmail.com> <CABihn6EN7V6XEwf6NWn78orxvr3XjGHxROJC4JjQ6RYYKEeCug@mail.gmail.com> <CACuKZqHNoR5GQmWyzbXAszZCOT2P4pjSmT3SF6ZG3X7hTY=1xw@mail.gmail.com> <CABihn6GC4VM2AHza-F7ML=FfHLZu7FNqx+BhbuVsfJLWk0P92w@mail.gmail.com> <CAH9hSJbSfQ2Abp6oLifi0dx4TZENzm2QRn8zMQfAv=vw+H12sw@mail.gmail.com> <alpine.DEB.2.00.1402261734120.31525@ps20323.dreamhostps.com>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Thu, 27 Feb 2014 02:52:46 +0900
Message-ID: <CAH9hSJZs01ZuimkLWat8Au+HZ5MPp33x+V5CTbG9OUia=S-8Jw@mail.gmail.com>
To: Ian Hickson <ian@hixie.ch>
Content-Type: multipart/alternative; boundary=047d7b5d256a82c0c804f352e028
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/mktsYvQgXAGb3o9BtcG4aHkXKQg
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 26 Feb 2014 17:53:12 -0000

--047d7b5d256a82c0c804f352e028
Content-Type: text/plain; charset=UTF-8

On Thu, Feb 27, 2014 at 2:35 AM, Ian Hickson <ian@hixie.ch> wrote:

> On Wed, 26 Feb 2014, Takeshi Yoshino wrote:
> >
> > I think it's underspecified. In the WHATWG WebSocket API spec, close()
> > is required to "start the WebSocket closing handshake". RFC 6455 neither
> > says the algorithm may terminate ongoing "Send a WebSocket Message"
> > algorithm nor says it may not.
>
> If it doesn't allow it, then it's not allowed.
>

Yeah, it's kinda twist.

But also it's not strange that people wonder if close frame interjection
like this is ok or not. Actually, even ping/pong sending are not explicitly
allowed to be inserted between "Send a WebSocket Message" algorithm.

The only normative text talking about interjection are this one in
https://tools.ietf.org/html/rfc6455#section-5.5
"Control frames can be interjected in the middle of a fragmented message."
and this in https://tools.ietf.org/html/rfc6455#section-5.4
"Control frames (see Section 5.5) MAY be injected in the middle of a
fragmented message."


>
> If you send a bunch of messages, then start the closing handshake, then
> the messages should all be sent. I don't think there's anything in either
> spec that would allow the handshake to jump ahead.
>

It would be nice if you add some note to the API spec to say "queuing"
doesn't have any implication that the closing handshake can jump in in
response to Zhong's question above.

--047d7b5d256a82c0c804f352e028
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote">On T=
hu, Feb 27, 2014 at 2:35 AM, Ian Hickson <span dir=3D"ltr">&lt;<a href=3D"m=
ailto:ian@hixie.ch" target=3D"_blank">ian@hixie.ch</a>&gt;</span> wrote:<br=
><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border=
-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;=
padding-left:1ex">

<div class=3D"">On Wed, 26 Feb 2014, Takeshi Yoshino wrote:<br>
&gt;<br>
&gt; I think it&#39;s underspecified. In the WHATWG WebSocket API spec, clo=
se()<br>
&gt; is required to &quot;start the WebSocket closing handshake&quot;. RFC =
6455 neither<br>
&gt; says the algorithm may terminate ongoing &quot;Send a WebSocket Messag=
e&quot;<br>
&gt; algorithm nor says it may not.<br>
<br>
</div>If it doesn&#39;t allow it, then it&#39;s not allowed.<br></blockquot=
e><div><br></div><div>Yeah, it&#39;s kinda twist.</div><div><br></div><div>=
But also it&#39;s not strange that people wonder if close frame interjectio=
n like this is ok or not. Actually, even ping/pong sending are not explicit=
ly allowed to be inserted between &quot;Send a WebSocket Message&quot; algo=
rithm.</div>

<div><br></div><div>The only normative text talking about interjection are =
this one in <a href=3D"https://tools.ietf.org/html/rfc6455#section-5.5">htt=
ps://tools.ietf.org/html/rfc6455#section-5.5</a></div><div><div>&quot;Contr=
ol frames can be interjected in the middle of a fragmented message.&quot;</=
div>

</div><div>and this in=C2=A0<a href=3D"https://tools.ietf.org/html/rfc6455#=
section-5.4">https://tools.ietf.org/html/rfc6455#section-5.4</a></div><div>=
<div>&quot;Control frames (see Section 5.5) MAY be injected in the middle o=
f a fragmented message.&quot;</div>

</div><div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0p=
x 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);bo=
rder-left-style:solid;padding-left:1ex">
<br>
If you send a bunch of messages, then start the closing handshake, then<br>
the messages should all be sent. I don&#39;t think there&#39;s anything in =
either<br>
spec that would allow the handshake to jump ahead.<br></blockquote><div><br=
></div><div>It would be nice if you add some note to the API spec to say &q=
uot;queuing&quot; doesn&#39;t have any implication that the closing handsha=
ke can jump in in response to Zhong&#39;s question above.</div>

</div></div></div>

--047d7b5d256a82c0c804f352e028--


From nobody Wed Feb 26 09:58:30 2014
Return-Path: <ian@hixie.ch>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 80ACD1A0788 for <hybi@ietfa.amsl.com>; Wed, 26 Feb 2014 09:58:20 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2
X-Spam-Level: 
X-Spam-Status: No, score=-2 tagged_above=-999 required=5 tests=[BAYES_00=-1.9,  DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id pz1EK9CUuOoS for <hybi@ietfa.amsl.com>; Wed, 26 Feb 2014 09:58:18 -0800 (PST)
Received: from homiemail-a51.g.dreamhost.com (caibbdcaaagc.dreamhost.com [208.113.200.62]) by ietfa.amsl.com (Postfix) with ESMTP id DFBF71A0787 for <hybi@ietf.org>; Wed, 26 Feb 2014 09:58:18 -0800 (PST)
Received: from homiemail-a51.g.dreamhost.com (localhost [127.0.0.1]) by homiemail-a51.g.dreamhost.com (Postfix) with ESMTP id C78FC2E807F; Wed, 26 Feb 2014 09:58:17 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha1; c=relaxed; d=hixie.ch; h=date:from:to :cc:subject:in-reply-to:message-id:references:mime-version: content-type; s=hixie.ch; bh=LsnSJA6Z/LhyYUHc6wKwiDzbFRA=; b=dPb Co1Uy4aBExZkQ0T1h3gWCSg5K9PhSqtF3nwWL1MVcsGGQ6N7LcfIuxXSMKe2pCH7 D8zbaSM7bGrN8m8YoqicLX1tGNPa/xrTti1/3WTGWPh3893QOlq7xQBaG8cThaTI nmb2ZQHZ2sqot/l/uGZD8vmjV9wnMyLeF/mTEcg4=
Received: from ps20323.dreamhostps.com (ps20323.dreamhost.com [208.113.236.128]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) (Authenticated sender: internal@index.hixie.ch) by homiemail-a51.g.dreamhost.com (Postfix) with ESMTPSA id A20852E8079; Wed, 26 Feb 2014 09:58:17 -0800 (PST)
Date: Wed, 26 Feb 2014 17:58:17 +0000 (UTC)
From: Ian Hickson <ian@hixie.ch>
To: Takeshi Yoshino <tyoshino@google.com>
In-Reply-To: <CAH9hSJZs01ZuimkLWat8Au+HZ5MPp33x+V5CTbG9OUia=S-8Jw@mail.gmail.com>
Message-ID: <alpine.DEB.2.00.1402261756240.31525@ps20323.dreamhostps.com>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <CAH9hSJbBmvNPBSSAk-khdWXgWw0GTt0FG3KsdzYeJcfiAPDk0A@mail.gmail.com> <CAHixhFq=wfmYH8-ij_WtsQLN=NUTJwRQ=k8jCPepQDM8V8ZZYA@mail.gmail.com> <CABihn6EN7V6XEwf6NWn78orxvr3XjGHxROJC4JjQ6RYYKEeCug@mail.gmail.com> <CACuKZqHNoR5GQmWyzbXAszZCOT2P4pjSmT3SF6ZG3X7hTY=1xw@mail.gmail.com> <CABihn6GC4VM2AHza-F7ML=FfHLZu7FNqx+BhbuVsfJLWk0P92w@mail.gmail.com> <CAH9hSJbSfQ2Abp6oLifi0dx4TZENzm2QRn8zMQfAv=vw+H12sw@mail.gmail.com> <alpine.DEB.2.00.1402261734120.31525@ps20323.dreamhostps.com> <CAH9hSJZs01ZuimkLWat8Au+HZ5MPp33x+V5CTbG9OUia=S-8Jw@mail.gmail.com>
User-Agent: Alpine 2.00 (DEB 1167 2008-08-23)
Content-Language: en-GB-hixie
Content-Style-Type: text/css
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/CXLyx-z9sdRXfcAQoUfDA2eIupw
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 26 Feb 2014 17:58:20 -0000

On Thu, 27 Feb 2014, Takeshi Yoshino wrote:
> 
> Yeah, it's kinda twist.
> 
> But also it's not strange that people wonder if close frame interjection 
> like this is ok or not. Actually, even ping/pong sending are not 
> explicitly allowed to be inserted between "Send a WebSocket Message" 
> algorithm.

If it's unclear, we should update the protocol spec to make it clear.


> > If you send a bunch of messages, then start the closing handshake, 
> > then the messages should all be sent. I don't think there's anything 
> > in either spec that would allow the handshake to jump ahead.
> 
> It would be nice if you add some note to the API spec to say "queuing" 
> doesn't have any implication that the closing handshake can jump in in 
> response to Zhong's question above.

Can you file a bug explaining specifically what needs clarifying at the 
API level? I'll get to it ASAP (I'm in the middle of an edit on focus 
navigation, so I can't fix it right now). http://whatwg.org/newbug

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'


From nobody Wed Feb 26 10:09:16 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id C30FB1A01E2 for <hybi@ietfa.amsl.com>; Wed, 26 Feb 2014 10:09:11 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.925
X-Spam-Level: 
X-Spam-Status: No, score=-1.925 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.547, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id v_uvVhvxu6bp for <hybi@ietfa.amsl.com>; Wed, 26 Feb 2014 10:09:07 -0800 (PST)
Received: from mail-wi0-x22c.google.com (mail-wi0-x22c.google.com [IPv6:2a00:1450:400c:c05::22c]) by ietfa.amsl.com (Postfix) with ESMTP id C1A441A00E8 for <hybi@ietf.org>; Wed, 26 Feb 2014 10:09:06 -0800 (PST)
Received: by mail-wi0-f172.google.com with SMTP id e4so6250956wiv.17 for <hybi@ietf.org>; Wed, 26 Feb 2014 10:09:05 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc:content-type; bh=EmIPM6OiMT5xW5zFvtLVzfEVT/kI19cfvx7B+LLsvvI=; b=YwSH694iBiWiPka7FvfRuIBIfjDsBX7WWtnyZ3d2JmJ+QWygTWS708i1AIVYKkcIUz uGDj8xvsiM5YesmDs7I2bUmH3ZRAdbh03KgiD/wDivz4ZCALi9nB1Q6+Nkwf0kpgC3LU EabWgd9KKbZGTrxpkJ0HPDPBXyy7TacFBhKRv43aNURUZ6OSHxXSjvac95SIVIFAyyRD 1oUY3HOlBLvlQB24n7qHlr+Wo1aP8CrzNaZa74Ju5BRL3IszGct3oM2o3QiPLwCkEO88 3raxQN7KSY3oRv0blY4k/FPnF4emIwyF8Il89tPOHU19OI1lk+NZAHM5KHqLojycrVqA O4Vg==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc:content-type; bh=EmIPM6OiMT5xW5zFvtLVzfEVT/kI19cfvx7B+LLsvvI=; b=I+w1ck41X4JHzriTsGHapkbaBX/sMZtSj3QcxsbuxiTr8gpV1t+KSQSekR5mBPClKx CEo9gUwOvGCSwWJN+oNrHNGF/myRp/1ry3bcZTlK3H0r4Rel6d+tcDwIAeFMN1pMoDIz rhd9XltlSM26VasG6nFcFir2HQcEd/IUoNPTRftps3aAqgziJtmrWkrNbmxrctBzbVdi ofWsY00NFfUBihLsW/Li+3APwoM3+876vVqU8tjqH6xwHcHnpKlfxCIGpamjhrHHD/H+ yJj27TeuRJq/Mf7Ezso99DReZZSSFByEV3xiL0OPJ9TxsNgF+OQ45Elai49XTqwzW4kU F7gQ==
X-Gm-Message-State: ALoCoQnNy9sD6RNSEeGL4EnSIEY/2yAuTK2DHPYan6i3pUhaUcqI529mo1DJdrLlla0YJOd2lNjYcJFXDxmDJBEzLm2owRbCnhsL3Iv/SxkQ2tEyTDtcxeOjybDhwKJSVWZbuL6UWjnoXuaMP9RdEbOj890lgB2Yhc/8340JWokWQGOox31vFUWoza+LgJQ3hB1gt9DmHInC
X-Received: by 10.180.73.173 with SMTP id m13mr5978745wiv.52.1393438145010; Wed, 26 Feb 2014 10:09:05 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Wed, 26 Feb 2014 10:08:44 -0800 (PST)
In-Reply-To: <alpine.DEB.2.00.1402261756240.31525@ps20323.dreamhostps.com>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <CAH9hSJbBmvNPBSSAk-khdWXgWw0GTt0FG3KsdzYeJcfiAPDk0A@mail.gmail.com> <CAHixhFq=wfmYH8-ij_WtsQLN=NUTJwRQ=k8jCPepQDM8V8ZZYA@mail.gmail.com> <CABihn6EN7V6XEwf6NWn78orxvr3XjGHxROJC4JjQ6RYYKEeCug@mail.gmail.com> <CACuKZqHNoR5GQmWyzbXAszZCOT2P4pjSmT3SF6ZG3X7hTY=1xw@mail.gmail.com> <CABihn6GC4VM2AHza-F7ML=FfHLZu7FNqx+BhbuVsfJLWk0P92w@mail.gmail.com> <CAH9hSJbSfQ2Abp6oLifi0dx4TZENzm2QRn8zMQfAv=vw+H12sw@mail.gmail.com> <alpine.DEB.2.00.1402261734120.31525@ps20323.dreamhostps.com> <CAH9hSJZs01ZuimkLWat8Au+HZ5MPp33x+V5CTbG9OUia=S-8Jw@mail.gmail.com> <alpine.DEB.2.00.1402261756240.31525@ps20323.dreamhostps.com>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Thu, 27 Feb 2014 03:08:44 +0900
Message-ID: <CAH9hSJbEJpUeDjpNsroUZ1CuD+Jifq5qakCjmfviZjSgvHBmrA@mail.gmail.com>
To: Ian Hickson <ian@hixie.ch>, salvatore loreto <salvatore.loreto@ericsson.com>
Content-Type: multipart/alternative; boundary=f46d043c07e0a688b804f3531914
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/HexQz164oC9T7xCVv320rDkEnQE
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 26 Feb 2014 18:09:12 -0000

--f46d043c07e0a688b804f3531914
Content-Type: text/plain; charset=UTF-8

On Thu, Feb 27, 2014 at 2:58 AM, Ian Hickson <ian@hixie.ch> wrote:

> On Thu, 27 Feb 2014, Takeshi Yoshino wrote:
> >
> > Yeah, it's kinda twist.
> >
> > But also it's not strange that people wonder if close frame interjection
> > like this is ok or not. Actually, even ping/pong sending are not
> > explicitly allowed to be inserted between "Send a WebSocket Message"
> > algorithm.
>
> If it's unclear, we should update the protocol spec to make it clear.
>
>
I can't file a ticket.

Sal, could you file a ticket for this for update in the future?


>
> > > If you send a bunch of messages, then start the closing handshake,
> > > then the messages should all be sent. I don't think there's anything
> > > in either spec that would allow the handshake to jump ahead.
> >
> > It would be nice if you add some note to the API spec to say "queuing"
> > doesn't have any implication that the closing handshake can jump in in
> > response to Zhong's question above.
>
> Can you file a bug explaining specifically what needs clarifying at the
> API level? I'll get to it ASAP (I'm in the middle of an edit on focus
> navigation, so I can't fix it right now). http://whatwg.org/newbug


Thanks. Filed.
https://www.w3.org/Bugs/Public/show_bug.cgi?id=24824

--f46d043c07e0a688b804f3531914
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote">On T=
hu, Feb 27, 2014 at 2:58 AM, Ian Hickson <span dir=3D"ltr">&lt;<a href=3D"m=
ailto:ian@hixie.ch" target=3D"_blank">ian@hixie.ch</a>&gt;</span> wrote:<br=
><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border=
-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;=
padding-left:1ex">

<div class=3D"">On Thu, 27 Feb 2014, Takeshi Yoshino wrote:<br>
&gt;<br>
&gt; Yeah, it&#39;s kinda twist.<br>
&gt;<br>
&gt; But also it&#39;s not strange that people wonder if close frame interj=
ection<br>
&gt; like this is ok or not. Actually, even ping/pong sending are not<br>
&gt; explicitly allowed to be inserted between &quot;Send a WebSocket Messa=
ge&quot;<br>
&gt; algorithm.<br>
<br>
</div>If it&#39;s unclear, we should update the protocol spec to make it cl=
ear.<br>
<div class=3D""><br></div></blockquote><div><br></div><div>I can&#39;t file=
 a ticket.</div><div><br></div><div>Sal, could you file a ticket for this f=
or update in the future?</div><div>=C2=A0</div><blockquote class=3D"gmail_q=
uote" style=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-c=
olor:rgb(204,204,204);border-left-style:solid;padding-left:1ex">

<div class=3D"">
<br>
&gt; &gt; If you send a bunch of messages, then start the closing handshake=
,<br>
&gt; &gt; then the messages should all be sent. I don&#39;t think there&#39=
;s anything<br>
&gt; &gt; in either spec that would allow the handshake to jump ahead.<br>
&gt;<br>
&gt; It would be nice if you add some note to the API spec to say &quot;que=
uing&quot;<br>
&gt; doesn&#39;t have any implication that the closing handshake can jump i=
n in<br>
&gt; response to Zhong&#39;s question above.<br>
<br>
</div>Can you file a bug explaining specifically what needs clarifying at t=
he<br>
API level? I&#39;ll get to it ASAP (I&#39;m in the middle of an edit on foc=
us<br>
navigation, so I can&#39;t fix it right now). <a href=3D"http://whatwg.org/=
newbug" target=3D"_blank">http://whatwg.org/newbug</a></blockquote><div><br=
></div><div>Thanks. Filed.</div><div><a href=3D"https://www.w3.org/Bugs/Pub=
lic/show_bug.cgi?id=3D24824">https://www.w3.org/Bugs/Public/show_bug.cgi?id=
=3D24824</a>=C2=A0</div>

</div></div></div>

--f46d043c07e0a688b804f3531914--


From nobody Wed Feb 26 11:47:55 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 9B4921A01D3 for <hybi@ietfa.amsl.com>; Wed, 26 Feb 2014 11:47:45 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.925
X-Spam-Level: 
X-Spam-Status: No, score=-1.925 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.547, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id SwlyBlutSSXx for <hybi@ietfa.amsl.com>; Wed, 26 Feb 2014 11:47:43 -0800 (PST)
Received: from mail-wi0-x230.google.com (mail-wi0-x230.google.com [IPv6:2a00:1450:400c:c05::230]) by ietfa.amsl.com (Postfix) with ESMTP id ADDC91A0139 for <hybi@ietf.org>; Wed, 26 Feb 2014 11:47:42 -0800 (PST)
Received: by mail-wi0-f176.google.com with SMTP id hi5so6358325wib.3 for <hybi@ietf.org>; Wed, 26 Feb 2014 11:47:41 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc:content-type; bh=RcRT/qlHYl0cQpuxwXUsg2y19sMy/OH8ihjOI79bC/I=; b=aG+lC1ZF1I4fvZH3WA/JRLPGAjIxjl/EHobsFUEQarpsDEsdMc1XCF/xoEj952FG2c UFG0Ir9yvm5xrjASGZE5yG9bgQf8/+GSpWX7YDoCuKrYfWo7+eGk1R6I2X5vHlbTc84S xdPlZKLnjGxxdh4YlMNhEQin6sH+3Ac15uaifuR9bHVwU8pjOZoJY9A9dAVNLuUt9K/d aZ3Mt7xahe0Nt9m0A6Nq3yiv9HjxXNYEjJfzX1EGklUzFhN63p6U9cGpTVXXKUcrByeT YJO7qS7m5iW+/Xz0g1ABXnVmR4CrGY3BYEQHzq/ZarPhnjcCX77qsABJWUKFoynnCnXs WmHg==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc:content-type; bh=RcRT/qlHYl0cQpuxwXUsg2y19sMy/OH8ihjOI79bC/I=; b=MTOMWMiXy4g+JpK1zpaEm97mNSHQ+Zmx0SX8s9/BoQfyufCwGZ8o60FHorWc3wNBd5 lZ8jmGACl6SzI669y+F57IuX41UGkjmmQAQXm68c5gqPZmMVI2Bb3UFNNAAbIPdtiuRs oaNOiNbclfTlBSx3fc+xWJMaz7+BewvA4F4yQUfZXTJemZ0okvrPV9/CUU11+L99uddj e9jjfkPR7MNYHW/4kjBAA5z98bj8gAEMXS2F60Z4cDAVNfQvjyofRMDeHDqFIVaTecXc wrRLAeQhD8+bq1E9l+qQFtMtUm7xx6xQfR4cpVWbjZzxTMmofcAuYWzyBem480SN8I2n BLxA==
X-Gm-Message-State: ALoCoQlqZ0gdYVRrVVs+XEstB0tjGsTEuh0iPYqvuTHbXnJMVN6N3Z+P/etifexdZOCWqSoZC6QdKvw/zngY7qseeOVbHlJUknQTIoqsNKTJN25M4WQ/o8vrbZqWcouYOjgetG0CWJUE+JuIk6SqlWozJVoZUo6KpAAbLnqX5m/juGwCu7MhXMImrUrGoAysYCTfyD9w2U2+
X-Received: by 10.194.62.206 with SMTP id a14mr4182501wjs.26.1393444060856; Wed, 26 Feb 2014 11:47:40 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Wed, 26 Feb 2014 11:47:20 -0800 (PST)
In-Reply-To: <634914A010D0B943A035D226786325D4447358E224@EXVMBX020-12.exch020.serverdata.net>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <634914A010D0B943A035D226786325D4446C537686@EXVMBX020-12.exch020.serverdata.net> <CAH9hSJaALTNiC9rUC-qVYKK74_YJA5psvGhbsFegTwaLsg9Dug@mail.gmail.com> <634914A010D0B943A035D226786325D4447358E224@EXVMBX020-12.exch020.serverdata.net>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Thu, 27 Feb 2014 04:47:20 +0900
Message-ID: <CAH9hSJZzhePkezJvUQ=OuFXBOojtHWbtr70uHEMn0kn7WM1gog@mail.gmail.com>
To: Tobias Oberstein <tobias.oberstein@tavendo.de>
Content-Type: multipart/alternative; boundary=047d7b66f8e143235404f3547a4c
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/KSjcFj4m_fdgxRHqw2_K0Fl0LKI
Cc: "hybi@ietf.org" <hybi@ietf.org>, Yutaka Hirano <yhirano@google.com>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Wed, 26 Feb 2014 19:47:45 -0000

--047d7b66f8e143235404f3547a4c
Content-Type: text/plain; charset=UTF-8

On Wed, Feb 26, 2014 at 9:49 PM, Tobias Oberstein <
tobias.oberstein@tavendo.de> wrote:

> Takeshi,
>
>
>
> it's my impression that this whole thread has reached some kind of dead
> end.
>
>
>
> There is one camp of people which would see "Plan D" (that is pure/raw
> tunneling of WS over HTTP2) as both sufficient and suitable.
>
>
>
> "Plan D" can provide a "scalable transport for the WebSocket API", and can
> also be actually called "WebSocket tunneling", since it maintains the whole
> WS semantics.
>
>
>
> "Plan D" also does not restrict RFC6455 design space (including any
> extensions) in any way.
>
>
>
> And this, I _do_ care about: the integrity and design freedom of the
> WebSocket protocol.
>
>
>
> So from a Chromium developers point of view, what is missing in "Plan D"?
>

I basically agree with you for most of the points here, but cannot ignore
input from Roberto. I've been also struggling to adopt WebSocket for Google
infrastructure and facing various difficulties which includes additional
complexity for introducing new protocol. I'm not saying Google is the only
target, but we need to respect large scale loadbalancer developers'
perspective.


> FWIW, we are interested in developing a WS extension
> ("permessage-priority") that supports message priorization (and could also
> support multiple logical channels).
>
>
>
> From a Chromium developers point of view: are you open to consider other
> options of reaching your goal: "scalable transport for the WebSocket API"?
>

Yes. But sorry I haven't got a chance to read through your proposal yet.


>
>
> /Tobias
>
>
>
>
>
>
>
> *Von:* Takeshi Yoshino [mailto:tyoshino@google.com]
> *Gesendet:* Montag, 24. Februar 2014 23:36
> *An:* Tobias Oberstein
> *Cc:* Yutaka Hirano; Peter Thorson; Roberto Peon; hybi@ietf.org
> *Betreff:* Re: Straw poll: Do you miss interjectable WebSocket level
> control frame? (was: Re: [hybi] Discontinuation of mux ...)
>
>
>
> On Wed, Feb 19, 2014 at 8:45 PM, Tobias Oberstein <
> tobias.oberstein@tavendo.de> wrote:
>
> I don't agree with that approach. I think HTTP2 should take WebSocket "as
> is", if it wants to act as a transport instead of attempting to redefine
> it. Or define it's own messaging (not bound by WS).
>
>
>
> I want to understand what you mean by this, Tobias, and maybe Peter.
>
>
>
> As one of Chromium developers, I just want to build and use a scalable
> transport for the WebSocket API. For this goal, it doesn't matter whether
> HTTP/2.0 works as a transparent transport or not. There's no such
> implication in the term "WS/HTTP/2.0" at least when used by me.
>
>
>
> What Roberto and Martin are suggesting sounds like one to be categorized
> as "define it's own messaging". Some of Yutaka's plans, too. What's your
> actual position for them. Not the best option for you but not against?
>
>
>

--047d7b66f8e143235404f3547a4c
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote">On W=
ed, Feb 26, 2014 at 9:49 PM, Tobias Oberstein <span dir=3D"ltr">&lt;<a href=
=3D"mailto:tobias.oberstein@tavendo.de" target=3D"_blank">tobias.oberstein@=
tavendo.de</a>&gt;</span> wrote:<br>


<blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-=
left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;p=
adding-left:1ex"><div lang=3D"DE" link=3D"blue" vlink=3D"purple"><p class=
=3D"MsoNormal">


<span style=3D"font-size:11pt;font-family:Calibri,sans-serif;color:rgb(31,7=
3,125)">Takeshi,<u></u><u></u></span></p><p class=3D"MsoNormal"><span style=
=3D"font-size:11pt;font-family:Calibri,sans-serif;color:rgb(31,73,125)"><u>=
</u>=C2=A0<u></u></span></p>


<p class=3D"MsoNormal"><span style=3D"font-size:11pt;font-family:Calibri,sa=
ns-serif;color:rgb(31,73,125)">it&#39;s my impression that this whole threa=
d has reached some kind of dead end.<u></u><u></u></span></p><p class=3D"Ms=
oNormal">


<span style=3D"font-size:11pt;font-family:Calibri,sans-serif;color:rgb(31,7=
3,125)"><u></u>=C2=A0<u></u></span></p><p class=3D"MsoNormal"><span style=
=3D"font-size:11pt;font-family:Calibri,sans-serif;color:rgb(31,73,125)">The=
re is one camp of people which would see &quot;Plan D&quot; (that is pure/r=
aw tunneling of WS over HTTP2) as both sufficient and suitable.<u></u><u></=
u></span></p>


<p class=3D"MsoNormal"><span style=3D"font-size:11pt;font-family:Calibri,sa=
ns-serif;color:rgb(31,73,125)"><u></u>=C2=A0<u></u></span></p><p class=3D"M=
soNormal"><span style=3D"font-size:11pt;font-family:Calibri,sans-serif;colo=
r:rgb(31,73,125)">&quot;Plan D&quot; can provide a &quot;scalable transport=
 for the WebSocket API&quot;, and can also be actually called &quot;WebSock=
et tunneling&quot;, since it maintains the whole WS semantics.<u></u><u></u=
></span></p>


<p class=3D"MsoNormal"><span style=3D"font-size:11pt;font-family:Calibri,sa=
ns-serif;color:rgb(31,73,125)"><u></u>=C2=A0<u></u></span></p><p class=3D"M=
soNormal"><span style=3D"font-size:11pt;font-family:Calibri,sans-serif;colo=
r:rgb(31,73,125)">&quot;Plan D&quot; also does not restrict RFC6455 design =
space (including any extensions) in any way.<u></u><u></u></span></p>


<p class=3D"MsoNormal"><span style=3D"font-size:11pt;font-family:Calibri,sa=
ns-serif;color:rgb(31,73,125)"><u></u>=C2=A0<u></u></span></p><p class=3D"M=
soNormal"><span style=3D"font-size:11pt;font-family:Calibri,sans-serif;colo=
r:rgb(31,73,125)">And this, I _do_ care about: the integrity and design fre=
edom of the WebSocket protocol.<u></u><u></u></span></p>


<p class=3D"MsoNormal"><span style=3D"font-size:11pt;font-family:Calibri,sa=
ns-serif;color:rgb(31,73,125)"><u></u>=C2=A0<u></u></span></p><p class=3D"M=
soNormal"><span style=3D"font-size:11pt;font-family:Calibri,sans-serif;colo=
r:rgb(31,73,125)">So from a Chromium developers point of view, what is miss=
ing in &quot;Plan D&quot;?</span></p>


</div></blockquote><div><br></div><div>I basically agree with you for most =
of the points here, but cannot ignore input from Roberto. I&#39;ve been als=
o struggling to adopt WebSocket for Google infrastructure and facing variou=
s difficulties which includes additional complexity for introducing new pro=
tocol. I&#39;m not saying Google is the only target, but we need to respect=
 large scale loadbalancer developers&#39; perspective.<br>


</div><div>=C2=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0p=
x 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);bo=
rder-left-style:solid;padding-left:1ex"><div lang=3D"DE" link=3D"blue" vlin=
k=3D"purple">


<div><p class=3D"MsoNormal"><span style=3D"font-size:11pt;font-family:Calib=
ri,sans-serif;color:rgb(31,73,125)"><u></u></span></p><p class=3D"MsoNormal=
"><span style=3D"font-size:11pt;font-family:Calibri,sans-serif;color:rgb(31=
,73,125)">FWIW, we are interested in developing a WS extension (&quot;perme=
ssage-priority&quot;) that supports message priorization (and could also su=
pport multiple logical channels).<u></u><u></u></span></p>


<p class=3D"MsoNormal"><span style=3D"font-size:11pt;font-family:Calibri,sa=
ns-serif;color:rgb(31,73,125)"><u></u>=C2=A0<u></u></span></p><p class=3D"M=
soNormal"><span style=3D"font-size:11pt;font-family:Calibri,sans-serif;colo=
r:rgb(31,73,125)">From a Chromium developers point of view: are you open to=
 consider other options of reaching your goal: &quot;scalable transport for=
 the WebSocket API&quot;?</span></p>


</div></div></blockquote><div><br></div><div>Yes. But sorry I haven&#39;t g=
ot a chance to read through your proposal yet.</div><div>=C2=A0</div><block=
quote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border-left-w=
idth:1px;border-left-color:rgb(204,204,204);border-left-style:solid;padding=
-left:1ex">


<div lang=3D"DE" link=3D"blue" vlink=3D"purple"><div><p class=3D"MsoNormal"=
><span style=3D"font-size:11pt;font-family:Calibri,sans-serif;color:rgb(31,=
73,125)"><u></u><u></u></span></p><p class=3D"MsoNormal"><span style=3D"fon=
t-size:11pt;font-family:Calibri,sans-serif;color:rgb(31,73,125)"><u></u>=C2=
=A0<u></u></span></p>


<p class=3D"MsoNormal"><span style=3D"font-size:11pt;font-family:Calibri,sa=
ns-serif;color:rgb(31,73,125)">/Tobias<u></u><u></u></span></p><p class=3D"=
MsoNormal"><span style=3D"font-size:11pt;font-family:Calibri,sans-serif;col=
or:rgb(31,73,125)"><u></u>=C2=A0<u></u></span></p>


<p class=3D"MsoNormal"><span style=3D"font-size:11pt;font-family:Calibri,sa=
ns-serif;color:rgb(31,73,125)"><u></u>=C2=A0<u></u></span></p><p class=3D"M=
soNormal"><span style=3D"font-size:11pt;font-family:Calibri,sans-serif;colo=
r:rgb(31,73,125)"><u></u>=C2=A0<u></u></span></p>


<div style=3D"border-style:none none none solid;border-left-color:blue;bord=
er-left-width:1.5pt;padding:0cm 0cm 0cm 4pt"><div><div style=3D"border-styl=
e:solid none none;border-top-color:rgb(225,225,225);border-top-width:1pt;pa=
dding:3pt 0cm 0cm">


<p class=3D"MsoNormal"><b><span style=3D"font-size:11pt;font-family:Calibri=
,sans-serif">Von:</span></b><span style=3D"font-size:11pt;font-family:Calib=
ri,sans-serif"> Takeshi Yoshino [mailto:<a href=3D"mailto:tyoshino@google.c=
om" target=3D"_blank">tyoshino@google.com</a>] <br>


<b>Gesendet:</b> Montag, 24. Februar 2014 23:36<br><b>An:</b> Tobias Oberst=
ein<br><b>Cc:</b> Yutaka Hirano; Peter Thorson; Roberto Peon; <a href=3D"ma=
ilto:hybi@ietf.org" target=3D"_blank">hybi@ietf.org</a><br><b>Betreff:</b> =
Re: Straw poll: Do you miss interjectable WebSocket level control frame? (w=
as: Re: [hybi] Discontinuation of mux ...)<u></u><u></u></span></p>


</div></div><div><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p><div><=
div><div><p class=3D"MsoNormal">On Wed, Feb 19, 2014 at 8:45 PM, Tobias Obe=
rstein &lt;<a href=3D"mailto:tobias.oberstein@tavendo.de" target=3D"_blank"=
>tobias.oberstein@tavendo.de</a>&gt; wrote:<u></u><u></u></p>


<blockquote style=3D"border-style:none none none solid;border-left-color:rg=
b(204,204,204);border-left-width:1pt;padding:0cm 0cm 0cm 6pt;margin-left:4.=
8pt;margin-right:0cm"><div><p class=3D"MsoNormal">I don&#39;t agree with th=
at approach. I think HTTP2 should take WebSocket &quot;as is&quot;, if it w=
ants to act as a transport instead of attempting to redefine it. Or define =
it&#39;s own messaging (not bound by WS).<u></u><u></u></p>


</div></blockquote><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></di=
v><div><p class=3D"MsoNormal">I want to understand what you mean by this, T=
obias, and maybe Peter.<u></u><u></u></p></div><div><p class=3D"MsoNormal">=
<u></u>=C2=A0<u></u></p>


</div><div><p class=3D"MsoNormal">As one of Chromium developers, I just wan=
t to build and use a scalable transport for the WebSocket API. For this goa=
l, it doesn&#39;t matter whether HTTP/2.0 works as a transparent transport =
or not. There&#39;s no such implication in the term &quot;WS/HTTP/2.0&quot;=
 at least when used by me.<u></u><u></u></p>


</div><div><p class=3D"MsoNormal"><u></u>=C2=A0<u></u></p></div><div><p cla=
ss=3D"MsoNormal">What Roberto and Martin are suggesting sounds like one to =
be categorized as &quot;define it&#39;s own messaging&quot;. Some of Yutaka=
&#39;s plans, too. What&#39;s your actual position for them. Not the best o=
ption for you but not against?<u></u><u></u></p>


</div><div><p class=3D"MsoNormal">=C2=A0<u></u><u></u></p></div></div></div=
></div></div></div></div></div></div></blockquote></div><br></div></div>

--047d7b66f8e143235404f3547a4c--


From nobody Fri Feb 28 08:17:06 2014
Return-Path: <webmaster@zaphoyd.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id CE7091A02BE for <hybi@ietfa.amsl.com>; Fri, 28 Feb 2014 08:17:03 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: 0
X-Spam-Level: 
X-Spam-Status: No, score=0 tagged_above=-999 required=5 tests=[BAYES_40=-0.001, HTML_MESSAGE=0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id fx3JUVrzI8u8 for <hybi@ietfa.amsl.com>; Fri, 28 Feb 2014 08:17:01 -0800 (PST)
Received: from authsmtp00.uchicago.edu (authsmtp00.uchicago.edu [128.135.12.120]) by ietfa.amsl.com (Postfix) with ESMTP id 1788E1A0325 for <hybi@ietf.org>; Fri, 28 Feb 2014 08:17:01 -0800 (PST)
Received: from ranna.uchicago.edu (ranna.uchicago.edu [128.135.45.206]) (Authenticated sender: zaphoyd) by authsmtp00.uchicago.edu (Postfix) with ESMTP id A258449805E; Fri, 28 Feb 2014 10:16:58 -0600 (CST)
Content-Type: multipart/alternative; boundary="Apple-Mail=_12790FBE-C3CA-41F4-85BB-B445A01FCE34"
Mime-Version: 1.0 (Mac OS X Mail 7.2 \(1874\))
From: Peter Thorson <webmaster@zaphoyd.com>
In-Reply-To: <CAH9hSJbEJpUeDjpNsroUZ1CuD+Jifq5qakCjmfviZjSgvHBmrA@mail.gmail.com>
Date: Fri, 28 Feb 2014 10:16:57 -0600
Message-Id: <AE8FD8AD-C52F-4E6C-9229-7990402C9414@zaphoyd.com>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <CAH9hSJbBmvNPBSSAk-khdWXgWw0GTt0FG3KsdzYeJcfiAPDk0A@mail.gmail.com> <CAHixhFq=wfmYH8-ij_WtsQLN=NUTJwRQ=k8jCPepQDM8V8ZZYA@mail.gmail.com> <CABihn6EN7V6XEwf6NWn78orxvr3XjGHxROJC4JjQ6RYYKEeCug@mail.gmail.com> <CACuKZqHNoR5GQmWyzbXAszZCOT2P4pjSmT3SF6ZG3X7hTY=1xw@mail.gmail.com> <CABihn6GC4VM2AHza-F7ML=FfHLZu7FNqx+BhbuVsfJLWk0P92w@mail.gmail.com> <CAH9hSJbSfQ2Abp6oLifi0dx4TZENzm2QRn8zMQfAv=vw+H12sw@mail.gmail.com> <alpine.DEB.2.00.1402261734120.31525@ps20323.dreamhostps.com> <CAH9hSJZs01ZuimkLWat8Au+HZ5MPp33x+V5CTbG9OUia=S-8Jw@mail.gmail.com> <alpine.DEB.2.00.1402261756240.31525@ps20323.dreamhostps.com> <CAH9hSJbEJpUeDjpNsroUZ1CuD+Jifq5qakCjmfviZjSgvHBmrA@mail.gmail.com>
To: Takeshi Yoshino <tyoshino@google.com>
X-Mailer: Apple Mail (2.1874)
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/cxMhoNpTL-8ta9Inh6PsB34l8PQ
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 28 Feb 2014 16:17:04 -0000

--Apple-Mail=_12790FBE-C3CA-41F4-85BB-B445A01FCE34
Content-Transfer-Encoding: quoted-printable
Content-Type: text/plain;
	charset=us-ascii

To clarify here. This is a decision made that affects the W3C browser =
API only, correct? It does not imply that injecting a close frame into =
the outgoing message to cleanly cancel and exit is violating RFC6455?

On Feb 26, 2014, at 12:08 , Takeshi Yoshino <tyoshino@google.com> wrote:

> On Thu, Feb 27, 2014 at 2:58 AM, Ian Hickson <ian@hixie.ch> wrote:
> On Thu, 27 Feb 2014, Takeshi Yoshino wrote:
> >
> > Yeah, it's kinda twist.
> >
> > But also it's not strange that people wonder if close frame =
interjection
> > like this is ok or not. Actually, even ping/pong sending are not
> > explicitly allowed to be inserted between "Send a WebSocket Message"
> > algorithm.
>=20
> If it's unclear, we should update the protocol spec to make it clear.
>=20
>=20
> I can't file a ticket.
>=20
> Sal, could you file a ticket for this for update in the future?
> =20
>=20
> > > If you send a bunch of messages, then start the closing handshake,
> > > then the messages should all be sent. I don't think there's =
anything
> > > in either spec that would allow the handshake to jump ahead.
> >
> > It would be nice if you add some note to the API spec to say =
"queuing"
> > doesn't have any implication that the closing handshake can jump in =
in
> > response to Zhong's question above.
>=20
> Can you file a bug explaining specifically what needs clarifying at =
the
> API level? I'll get to it ASAP (I'm in the middle of an edit on focus
> navigation, so I can't fix it right now). http://whatwg.org/newbug
>=20
> Thanks. Filed.
> https://www.w3.org/Bugs/Public/show_bug.cgi?id=3D24824=20


--Apple-Mail=_12790FBE-C3CA-41F4-85BB-B445A01FCE34
Content-Transfer-Encoding: quoted-printable
Content-Type: text/html;
	charset=us-ascii

<html><head><meta http-equiv=3D"Content-Type" content=3D"text/html =
charset=3Dus-ascii"></head><body style=3D"word-wrap: break-word; =
-webkit-nbsp-mode: space; -webkit-line-break: =
after-white-space;"><div>To clarify here. This is a decision made that =
affects the W3C browser API only, correct? It does not imply that =
injecting a close frame into the outgoing message to cleanly cancel and =
exit is violating RFC6455?</div><br><div><div>On Feb 26, 2014, at 12:08 =
, Takeshi Yoshino &lt;<a =
href=3D"mailto:tyoshino@google.com">tyoshino@google.com</a>&gt; =
wrote:</div><br class=3D"Apple-interchange-newline"><blockquote =
type=3D"cite"><div dir=3D"ltr"><div class=3D"gmail_extra"><div =
class=3D"gmail_quote">On Thu, Feb 27, 2014 at 2:58 AM, Ian Hickson <span =
dir=3D"ltr">&lt;<a href=3D"mailto:ian@hixie.ch" =
target=3D"_blank">ian@hixie.ch</a>&gt;</span> wrote:<br><blockquote =
class=3D"gmail_quote" style=3D"margin:0px 0px 0px =
0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left=
-style:solid;padding-left:1ex">

<div class=3D"">On Thu, 27 Feb 2014, Takeshi Yoshino wrote:<br>
&gt;<br>
&gt; Yeah, it's kinda twist.<br>
&gt;<br>
&gt; But also it's not strange that people wonder if close frame =
interjection<br>
&gt; like this is ok or not. Actually, even ping/pong sending are =
not<br>
&gt; explicitly allowed to be inserted between "Send a WebSocket =
Message"<br>
&gt; algorithm.<br>
<br>
</div>If it's unclear, we should update the protocol spec to make it =
clear.<br>
<div class=3D""><br></div></blockquote><div><br></div><div>I can't file =
a ticket.</div><div><br></div><div>Sal, could you file a ticket for this =
for update in the future?</div><div>&nbsp;</div><blockquote =
class=3D"gmail_quote" style=3D"margin:0px 0px 0px =
0.8ex;border-left-width:1px;border-left-color:rgb(204,204,204);border-left=
-style:solid;padding-left:1ex">

<div class=3D"">
<br>
&gt; &gt; If you send a bunch of messages, then start the closing =
handshake,<br>
&gt; &gt; then the messages should all be sent. I don't think there's =
anything<br>
&gt; &gt; in either spec that would allow the handshake to jump =
ahead.<br>
&gt;<br>
&gt; It would be nice if you add some note to the API spec to say =
"queuing"<br>
&gt; doesn't have any implication that the closing handshake can jump in =
in<br>
&gt; response to Zhong's question above.<br>
<br>
</div>Can you file a bug explaining specifically what needs clarifying =
at the<br>
API level? I'll get to it ASAP (I'm in the middle of an edit on =
focus<br>
navigation, so I can't fix it right now). <a =
href=3D"http://whatwg.org/newbug" =
target=3D"_blank">http://whatwg.org/newbug</a></blockquote><div><br></div>=
<div>Thanks. Filed.</div><div><a =
href=3D"https://www.w3.org/Bugs/Public/show_bug.cgi?id=3D24824">https://ww=
w.w3.org/Bugs/Public/show_bug.cgi?id=3D24824</a>&nbsp;</div>

</div></div></div>
</blockquote></div><br></body></html>=

--Apple-Mail=_12790FBE-C3CA-41F4-85BB-B445A01FCE34--


From nobody Fri Feb 28 10:00:58 2014
Return-Path: <tyoshino@google.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 4E8E61A00FF for <hybi@ietfa.amsl.com>; Fri, 28 Feb 2014 10:00:57 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -1.925
X-Spam-Level: 
X-Spam-Status: No, score=-1.925 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FM_FORGED_GMAIL=0.622, HTML_MESSAGE=0.001, RP_MATCHES_RCVD=-0.547, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 5qzV2YDLZEDf for <hybi@ietfa.amsl.com>; Fri, 28 Feb 2014 10:00:55 -0800 (PST)
Received: from mail-wg0-x22e.google.com (mail-wg0-x22e.google.com [IPv6:2a00:1450:400c:c00::22e]) by ietfa.amsl.com (Postfix) with ESMTP id CB1171A00D9 for <hybi@ietf.org>; Fri, 28 Feb 2014 10:00:54 -0800 (PST)
Received: by mail-wg0-f46.google.com with SMTP id z12so856771wgg.29 for <hybi@ietf.org>; Fri, 28 Feb 2014 10:00:52 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=google.com; s=20120113; h=mime-version:in-reply-to:references:from:date:message-id:subject:to :cc:content-type; bh=7bY9kfNt9ojJQjSnBpd/+B1vDwa++/j2x05KmIUzh6M=; b=LlCxt1UY1DVDsFygZljpn5ac1fvfFfTgIwQbyFZY2uu1FEcFq8d8no+LzD3cyjWIXm G+nthqrPnoZK0CGCncQWvxS0lf78TcbQrJl18CaGLcmOTsG07S6sJlKvsPQfiRATaydN +GbhExloe+wP2h5ZvJq7qTex8jKbmhvZmwdqEvehSa8Zz2Bs/GUImbVGFPtt2V/SCoL8 QsYsYuRYc0QA8ivNZIebGFwNVtStnhRjWmyJ5kyJcLmF09xeRA2dXaUbmK5SScWM9LRC rZ1cB194Ah1xVKyUqnYTq1+3nE/Yt40lbuMee8SLoKavaZzkaNMJflLkZtNaaG8qYyAS qXVw==
X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:mime-version:in-reply-to:references:from:date :message-id:subject:to:cc:content-type; bh=7bY9kfNt9ojJQjSnBpd/+B1vDwa++/j2x05KmIUzh6M=; b=SRJLuv73ESZ2oU0aUwKd7sMf+X4YTulWMWwNAxIjDc1iLpuSC8Uqg6GjbQz9Z2dzPC eP3fiuqhLHBq1AbK2hBzMrQ9UEktBJypmPbE4TV2uU8DeUDZVzhQxSfVQcUBFWVScoyP A+wh9VV6XSZl0e5jjwf8iNr5Wg6aMMFMe+EZIn5gdHd6JJ36BHIArUxenhmSazzx8Lki VWAa8BwS6VQ/KQ5NlcaXTjvwu8uOvOQ9hkTEzpjKhOrVb+2hG3PPOG3CvQ7HmkTTesNP B+rCitEBan29hRyaOHlW+CKpCdRLKDDFzM8LiYLqfUsCMTw9xmhBxVYxidTHFLzzcYEq LFFg==
X-Gm-Message-State: ALoCoQk6X9Q4LhJG+dbG3ZdQEpbHCf5jeeIqQtfxM7YxC/4ULvgmLYo5B5ZfABDP3+1YFPeIyalYnvNgiWtHRS8G5ak0DzaVCZn0mb43kLokAm9Vh6TA98b6hceu6c3W1rl6stL2VWApfgmzjmo6Jht6F3XyLFkCO8EoQ6VNdCsH4AQynUmUqJ2eKExRqhdAOm2XUbOojn4W
X-Received: by 10.180.19.138 with SMTP id f10mr3828596wie.11.1393610452325; Fri, 28 Feb 2014 10:00:52 -0800 (PST)
MIME-Version: 1.0
Received: by 10.194.8.231 with HTTP; Fri, 28 Feb 2014 10:00:32 -0800 (PST)
In-Reply-To: <AE8FD8AD-C52F-4E6C-9229-7990402C9414@zaphoyd.com>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <CAH9hSJbBmvNPBSSAk-khdWXgWw0GTt0FG3KsdzYeJcfiAPDk0A@mail.gmail.com> <CAHixhFq=wfmYH8-ij_WtsQLN=NUTJwRQ=k8jCPepQDM8V8ZZYA@mail.gmail.com> <CABihn6EN7V6XEwf6NWn78orxvr3XjGHxROJC4JjQ6RYYKEeCug@mail.gmail.com> <CACuKZqHNoR5GQmWyzbXAszZCOT2P4pjSmT3SF6ZG3X7hTY=1xw@mail.gmail.com> <CABihn6GC4VM2AHza-F7ML=FfHLZu7FNqx+BhbuVsfJLWk0P92w@mail.gmail.com> <CAH9hSJbSfQ2Abp6oLifi0dx4TZENzm2QRn8zMQfAv=vw+H12sw@mail.gmail.com> <alpine.DEB.2.00.1402261734120.31525@ps20323.dreamhostps.com> <CAH9hSJZs01ZuimkLWat8Au+HZ5MPp33x+V5CTbG9OUia=S-8Jw@mail.gmail.com> <alpine.DEB.2.00.1402261756240.31525@ps20323.dreamhostps.com> <CAH9hSJbEJpUeDjpNsroUZ1CuD+Jifq5qakCjmfviZjSgvHBmrA@mail.gmail.com> <AE8FD8AD-C52F-4E6C-9229-7990402C9414@zaphoyd.com>
From: Takeshi Yoshino <tyoshino@google.com>
Date: Sat, 1 Mar 2014 03:00:32 +0900
Message-ID: <CAH9hSJbeRsmFdzF5yDwr7aGMyhz0CMmMhQ8MdFonkHZ0Xu89aA@mail.gmail.com>
To: Peter Thorson <webmaster@zaphoyd.com>
Content-Type: multipart/alternative; boundary=bcaec53d550df77e0e04f37b3751
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/z9YvhMnJd6fn5Od8JQDH1o6L6Ss
Cc: "hybi@ietf.org" <hybi@ietf.org>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 28 Feb 2014 18:00:57 -0000

--bcaec53d550df77e0e04f37b3751
Content-Type: text/plain; charset=UTF-8

On Sat, Mar 1, 2014 at 1:16 AM, Peter Thorson <webmaster@zaphoyd.com> wrote:

> To clarify here. This is a decision made that affects the W3C browser API
> only, correct? It does not imply that injecting a close frame into the
> outgoing message to cleanly cancel and exit is violating RFC6455?
>

Right. Only Hixie expressed his thoughts and interpretation of the spec for
now. There should be more discussion for this.


>
> On Feb 26, 2014, at 12:08 , Takeshi Yoshino <tyoshino@google.com> wrote:
>
> On Thu, Feb 27, 2014 at 2:58 AM, Ian Hickson <ian@hixie.ch> wrote:
>
>> On Thu, 27 Feb 2014, Takeshi Yoshino wrote:
>> >
>> > Yeah, it's kinda twist.
>> >
>> > But also it's not strange that people wonder if close frame interjection
>> > like this is ok or not. Actually, even ping/pong sending are not
>> > explicitly allowed to be inserted between "Send a WebSocket Message"
>> > algorithm.
>>
>> If it's unclear, we should update the protocol spec to make it clear.
>>
>>
> I can't file a ticket.
>
> Sal, could you file a ticket for this for update in the future?
>
>
>>
>> > > If you send a bunch of messages, then start the closing handshake,
>> > > then the messages should all be sent. I don't think there's anything
>> > > in either spec that would allow the handshake to jump ahead.
>> >
>> > It would be nice if you add some note to the API spec to say "queuing"
>> > doesn't have any implication that the closing handshake can jump in in
>> > response to Zhong's question above.
>>
>> Can you file a bug explaining specifically what needs clarifying at the
>> API level? I'll get to it ASAP (I'm in the middle of an edit on focus
>> navigation, so I can't fix it right now). http://whatwg.org/newbug
>
>
> Thanks. Filed.
> https://www.w3.org/Bugs/Public/show_bug.cgi?id=24824
>
>
>

--bcaec53d550df77e0e04f37b3751
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote">On S=
at, Mar 1, 2014 at 1:16 AM, Peter Thorson <span dir=3D"ltr">&lt;<a href=3D"=
mailto:webmaster@zaphoyd.com" target=3D"_blank">webmaster@zaphoyd.com</a>&g=
t;</span> wrote:<br>


<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex"><div style=3D"word-wrap:break-word"><div>To =
clarify here. This is a decision made that affects the W3C browser API only=
, correct? It does not imply that injecting a close frame into the outgoing=
 message to cleanly cancel and exit is violating RFC6455?</div>


</div></blockquote><div><br></div><div>Right. Only Hixie expressed his thou=
ghts and interpretation of the spec for now. There should be more discussio=
n for this.</div><div>=C2=A0</div><blockquote class=3D"gmail_quote" style=
=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">


<div style=3D"word-wrap:break-word"><div><div><br><div><div>On Feb 26, 2014=
, at 12:08 , Takeshi Yoshino &lt;<a href=3D"mailto:tyoshino@google.com" tar=
get=3D"_blank">tyoshino@google.com</a>&gt; wrote:</div><br><blockquote type=
=3D"cite">


<div dir=3D"ltr"><div class=3D"gmail_extra"><div class=3D"gmail_quote">On T=
hu, Feb 27, 2014 at 2:58 AM, Ian Hickson <span dir=3D"ltr">&lt;<a href=3D"m=
ailto:ian@hixie.ch" target=3D"_blank">ian@hixie.ch</a>&gt;</span> wrote:<br=
><blockquote class=3D"gmail_quote" style=3D"margin:0px 0px 0px 0.8ex;border=
-left-width:1px;border-left-color:rgb(204,204,204);border-left-style:solid;=
padding-left:1ex">




<div>On Thu, 27 Feb 2014, Takeshi Yoshino wrote:<br>
&gt;<br>
&gt; Yeah, it&#39;s kinda twist.<br>
&gt;<br>
&gt; But also it&#39;s not strange that people wonder if close frame interj=
ection<br>
&gt; like this is ok or not. Actually, even ping/pong sending are not<br>
&gt; explicitly allowed to be inserted between &quot;Send a WebSocket Messa=
ge&quot;<br>
&gt; algorithm.<br>
<br>
</div>If it&#39;s unclear, we should update the protocol spec to make it cl=
ear.<br>
<div><br></div></blockquote><div><br></div><div>I can&#39;t file a ticket.<=
/div><div><br></div><div>Sal, could you file a ticket for this for update i=
n the future?</div><div>=C2=A0</div><blockquote class=3D"gmail_quote" style=
=3D"margin:0px 0px 0px 0.8ex;border-left-width:1px;border-left-color:rgb(20=
4,204,204);border-left-style:solid;padding-left:1ex">




<div>
<br>
&gt; &gt; If you send a bunch of messages, then start the closing handshake=
,<br>
&gt; &gt; then the messages should all be sent. I don&#39;t think there&#39=
;s anything<br>
&gt; &gt; in either spec that would allow the handshake to jump ahead.<br>
&gt;<br>
&gt; It would be nice if you add some note to the API spec to say &quot;que=
uing&quot;<br>
&gt; doesn&#39;t have any implication that the closing handshake can jump i=
n in<br>
&gt; response to Zhong&#39;s question above.<br>
<br>
</div>Can you file a bug explaining specifically what needs clarifying at t=
he<br>
API level? I&#39;ll get to it ASAP (I&#39;m in the middle of an edit on foc=
us<br>
navigation, so I can&#39;t fix it right now). <a href=3D"http://whatwg.org/=
newbug" target=3D"_blank">http://whatwg.org/newbug</a></blockquote><div><br=
></div><div>Thanks. Filed.</div><div><a href=3D"https://www.w3.org/Bugs/Pub=
lic/show_bug.cgi?id=3D24824" target=3D"_blank">https://www.w3.org/Bugs/Publ=
ic/show_bug.cgi?id=3D24824</a>=C2=A0</div>




</div></div></div>
</blockquote></div><br></div></div></div></blockquote></div><br></div></div=
>

--bcaec53d550df77e0e04f37b3751--


From nobody Fri Feb 28 12:46:05 2014
Return-Path: <zhong.j.yu@gmail.com>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 0EA831A0238 for <hybi@ietfa.amsl.com>; Fri, 28 Feb 2014 12:46:04 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2
X-Spam-Level: 
X-Spam-Status: No, score=-2 tagged_above=-999 required=5 tests=[BAYES_00=-1.9,  DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1, FREEMAIL_FROM=0.001, SPF_PASS=-0.001] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 28xKE9Oj6sW2 for <hybi@ietfa.amsl.com>; Fri, 28 Feb 2014 12:46:02 -0800 (PST)
Received: from mail-oa0-x235.google.com (mail-oa0-x235.google.com [IPv6:2607:f8b0:4003:c02::235]) by ietfa.amsl.com (Postfix) with ESMTP id 0E5F61A0316 for <hybi@ietf.org>; Fri, 28 Feb 2014 12:46:02 -0800 (PST)
Received: by mail-oa0-f53.google.com with SMTP id j17so78215oag.12 for <hybi@ietf.org>; Fri, 28 Feb 2014 12:46:00 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113;  h=mime-version:in-reply-to:references:date:message-id:subject:from:to :cc:content-type; bh=2RzHIffxC/MaetI6J2Y+VtUWz4fyB/XeRvL0n/MxMFs=; b=QmSrvCw3QDbWBNL3ZqSdTK1BkaXuZ4pYVY1oilUR6Xbrrmi9M8+MAC5KeTOHZS3/bc DY70oOfV7yZrcG2oux3auhr4V6H5OizB/qTzVIBsaudLIi3r8FDNjnp44msg+slhwje1 pECzuoj79zLLzEGbkJCtdYq3RBTTiHHBUdWY4wCBS9lctOntA4D2Oq6GP1ypkglpkZEi qeNLXQbXFbzmhYRk5cydeaDyeF3xRwEyi9ySGpeDSbf2vMPYepjTem7fxsX4kHH9O+oU i/B8FcsA/vhEwIX0fzbTziNtJg4tMnYzpYyd5zc4qgmycet0lzv7eJruiRoMRscTbxS9 wQ/g==
MIME-Version: 1.0
X-Received: by 10.60.52.241 with SMTP id w17mr4739672oeo.4.1393620360066; Fri, 28 Feb 2014 12:46:00 -0800 (PST)
Received: by 10.76.106.162 with HTTP; Fri, 28 Feb 2014 12:45:59 -0800 (PST)
In-Reply-To: <alpine.DEB.2.00.1402261734120.31525@ps20323.dreamhostps.com>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <CAH9hSJbBmvNPBSSAk-khdWXgWw0GTt0FG3KsdzYeJcfiAPDk0A@mail.gmail.com> <CAHixhFq=wfmYH8-ij_WtsQLN=NUTJwRQ=k8jCPepQDM8V8ZZYA@mail.gmail.com> <CABihn6EN7V6XEwf6NWn78orxvr3XjGHxROJC4JjQ6RYYKEeCug@mail.gmail.com> <CACuKZqHNoR5GQmWyzbXAszZCOT2P4pjSmT3SF6ZG3X7hTY=1xw@mail.gmail.com> <CABihn6GC4VM2AHza-F7ML=FfHLZu7FNqx+BhbuVsfJLWk0P92w@mail.gmail.com> <CAH9hSJbSfQ2Abp6oLifi0dx4TZENzm2QRn8zMQfAv=vw+H12sw@mail.gmail.com> <alpine.DEB.2.00.1402261734120.31525@ps20323.dreamhostps.com>
Date: Fri, 28 Feb 2014 14:45:59 -0600
Message-ID: <CACuKZqGQ6s9qFDuLL_95J+RpDSZ2cZWcESM_z=WgoUjbLHwrNw@mail.gmail.com>
From: Zhong Yu <zhong.j.yu@gmail.com>
To: Ian Hickson <ian@hixie.ch>
Content-Type: text/plain; charset=ISO-8859-1
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/b3m-2L8IXb9NiNXr2JGS8__bx3I
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 28 Feb 2014 20:46:04 -0000

On Wed, Feb 26, 2014 at 11:35 AM, Ian Hickson <ian@hixie.ch> wrote:
> On Wed, 26 Feb 2014, Takeshi Yoshino wrote:
>>
>> I think it's underspecified. In the WHATWG WebSocket API spec, close()
>> is required to "start the WebSocket closing handshake". RFC 6455 neither
>> says the algorithm may terminate ongoing "Send a WebSocket Message"
>> algorithm nor says it may not.
>
> If it doesn't allow it, then it's not allowed.
>
> If you send a bunch of messages, then start the closing handshake, then
> the messages should all be sent.

I think this is a good choice for most use cases. But eventually
someone will need the feature of abort/cancel an outgoing message that
takes too long to transmit.

If write() is a blocking method, it makes sense that close() after
write() should flush the data at best effort. But here write() is a
non-blocking, async method, it's not intuitively clear what should
happen if close() is called when a write action is still ongoing.

Zhong Yu


> I don't think there's anything in either
> spec that would allow the handshake to jump ahead.


>
> --
> Ian Hickson               U+1047E                )\._.,--....,'``.    fL
> http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
> Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'


From nobody Fri Feb 28 13:00:45 2014
Return-Path: <ian@hixie.ch>
X-Original-To: hybi@ietfa.amsl.com
Delivered-To: hybi@ietfa.amsl.com
Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id D39401A01C8 for <hybi@ietfa.amsl.com>; Fri, 28 Feb 2014 13:00:43 -0800 (PST)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2
X-Spam-Level: 
X-Spam-Status: No, score=-2 tagged_above=-999 required=5 tests=[BAYES_00=-1.9,  DKIM_SIGNED=0.1, DKIM_VALID=-0.1, DKIM_VALID_AU=-0.1] autolearn=ham
Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id wLsI6kjnAy-2 for <hybi@ietfa.amsl.com>; Fri, 28 Feb 2014 13:00:41 -0800 (PST)
Received: from homiemail-a46.g.dreamhost.com (caibbdcaacfb.dreamhost.com [208.113.200.251]) by ietfa.amsl.com (Postfix) with ESMTP id E184A1A0164 for <hybi@ietf.org>; Fri, 28 Feb 2014 13:00:41 -0800 (PST)
Received: from homiemail-a46.g.dreamhost.com (localhost [127.0.0.1]) by homiemail-a46.g.dreamhost.com (Postfix) with ESMTP id 0B3743E406C; Fri, 28 Feb 2014 13:00:40 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha1; c=relaxed; d=hixie.ch; h=date:from:to :cc:subject:in-reply-to:message-id:references:mime-version: content-type; s=hixie.ch; bh=p2+uzqtdjujUpB0VVhwSAipy+9c=; b=XpN WP/RljFjtt6hWpWzvO39Rm0O/YhTHTFY4KHGSRAlz+C/7I5Hc3Z0W33R7kDQUdu7 5wD+xTL9xtNgCN87NA2A68oz374EYV6R68L4X6kByLGTSG4rprJZK5xcNo0fBOk2 VZoAPer0MW6Z86yiIbeMez4NcJ//DfWz8SspArr4=
Received: from ps20323.dreamhostps.com (ps20323.dreamhost.com [208.113.236.128]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) (Authenticated sender: internal@index.hixie.ch) by homiemail-a46.g.dreamhost.com (Postfix) with ESMTPSA id F178B3E4057; Fri, 28 Feb 2014 13:00:39 -0800 (PST)
Date: Fri, 28 Feb 2014 21:00:39 +0000 (UTC)
From: Ian Hickson <ian@hixie.ch>
To: Zhong Yu <zhong.j.yu@gmail.com>
In-Reply-To: <CACuKZqGQ6s9qFDuLL_95J+RpDSZ2cZWcESM_z=WgoUjbLHwrNw@mail.gmail.com>
Message-ID: <alpine.DEB.2.00.1402282056570.31525@ps20323.dreamhostps.com>
References: <CAH9hSJbjQNKnZTJmBFtU8MgmnRTYjPopC4oP_78bWUGap-9CvA@mail.gmail.com> <CAH9hSJbBmvNPBSSAk-khdWXgWw0GTt0FG3KsdzYeJcfiAPDk0A@mail.gmail.com> <CAHixhFq=wfmYH8-ij_WtsQLN=NUTJwRQ=k8jCPepQDM8V8ZZYA@mail.gmail.com> <CABihn6EN7V6XEwf6NWn78orxvr3XjGHxROJC4JjQ6RYYKEeCug@mail.gmail.com> <CACuKZqHNoR5GQmWyzbXAszZCOT2P4pjSmT3SF6ZG3X7hTY=1xw@mail.gmail.com> <CABihn6GC4VM2AHza-F7ML=FfHLZu7FNqx+BhbuVsfJLWk0P92w@mail.gmail.com> <CAH9hSJbSfQ2Abp6oLifi0dx4TZENzm2QRn8zMQfAv=vw+H12sw@mail.gmail.com> <alpine.DEB.2.00.1402261734120.31525@ps20323.dreamhostps.com> <CACuKZqGQ6s9qFDuLL_95J+RpDSZ2cZWcESM_z=WgoUjbLHwrNw@mail.gmail.com>
User-Agent: Alpine 2.00 (DEB 1167 2008-08-23)
Content-Language: en-GB-hixie
Content-Style-Type: text/css
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Archived-At: http://mailarchive.ietf.org/arch/msg/hybi/fCRXOFII6ZirEdBWfF6yTo4ZsXU
Cc: "hybi@ietf.org" <hybi@ietf.org>, Peter Thorson <webmaster@zaphoyd.com>
Subject: Re: [hybi] Straw poll: Do you miss interjectable WebSocket level control frame? (was: Re: Discontinuation of mux ...)
X-BeenThere: hybi@ietf.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Server-Initiated HTTP <hybi.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/options/hybi>, <mailto:hybi-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/hybi/>
List-Post: <mailto:hybi@ietf.org>
List-Help: <mailto:hybi-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/hybi>, <mailto:hybi-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 28 Feb 2014 21:00:44 -0000

On Fri, 28 Feb 2014, Zhong Yu wrote:
> 
> I think this is a good choice for most use cases. But eventually someone 
> will need the feature of abort/cancel an outgoing message that takes too 
> long to transmit.

Once someone has a use case for this, we can study it and consider the 
right solution.


> If write() is a blocking method, it makes sense that close() after 
> write() should flush the data at best effort. But here write() is a 
> non-blocking, async method, it's not intuitively clear what should 
> happen if close() is called when a write action is still ongoing.

In the Web Socket API, WebSocket.send() and WebSocket.close() are both 
specced as atomic. The fact that buffering has to happen is an 
implementation detail (albeit one exposed in the bufferedAmount 
attribute). The protocol spec could provide more detailed hooks to control 
this, of course. But I don't think the specs are unclear.

-- 
Ian Hickson               U+1047E                )\._.,--....,'``.    fL
http://ln.hixie.ch/       U+263A                /,   _.. \   _\  ;`._ ,.
Things that are impossible just take longer.   `._.-(,_..'--(,_..'`-.;.'

