W3C home > Mailing lists > Public > ietf-http-wg@w3.org > July to September 2014

Re: h2 frame layout

From: Greg Wilkins <gregw@intalio.com>
Date: Mon, 1 Sep 2014 12:44:09 +1000
Message-ID: <CAH_y2NEOeXXJJvxEQKWoY4zeRoYhkBtebnWbNkWRT=QwASP-Vw@mail.gmail.com>
To: Matthew Kerwin <matthew@kerwin.net.au>
Cc: "Roy T. Fielding" <fielding@gbiv.com>, HTTP Working Group <ietf-http-wg@w3.org>
On 1 September 2014 11:57, Matthew Kerwin <matthew@kerwin.net.au> wrote:

> If the framing machine were frame-type-agnostic,


+1.  That is my main point.


> all frame types could carry the END_STREAM flag


that is one way of achieving it.    Roy's proposal B is another.

I'm a bit agnostic as to which is best.   The flag is probably simplest and
more conventional, but would mean that we need to review silly combinations
like RST_STREAM with no END_STREAM.    But maybe in a frame-type agnostic
world, we don't need RST_STREAM and GO_AWAY frames?   Isn't RST_STREAM just
a GO_AWAY with a non zero stream ID?

Maybe we can replace the END_STREAM flag and both GO_AWAY and RST_STREAM
frames with a single END_STREAM frame:

   -   END_STREAM(SID==0, CODE==0),  normal graceful go away
   -   END_STREAM(SID==0, CODE!=0),  error connection shutdown
   -   END_STREAM(SID!=0, CODE==0),  normal stream shutdown
   -   END_STREAM(SID!=0, CODE!=0),  reset stream style error

cheers









-- 
Greg Wilkins <gregw@intalio.com>
http://eclipse.org/jetty HTTP, SPDY, Websocket server and client that scales
http://www.webtide.com  advice and support for jetty and cometd.
Received on Monday, 1 September 2014 02:44:38 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 30 March 2016 09:57:10 UTC