W3C home > Mailing lists > Public > ietf-http-wg@w3.org > January to March 2013

Re: SYN_REPLY

From: Martin Thomson <martin.thomson@gmail.com>
Date: Thu, 21 Feb 2013 16:39:55 -0800
Message-ID: <CABkgnnXq-vWXg_H-eJhm=Jn-0q13ok6ZMbx5yjU93Qhe1ibKvQ@mail.gmail.com>
To: Amos Jeffries <squid3@treenet.co.nz>
Cc: HTTP Working Group <ietf-http-wg@w3.org>
On 21 February 2013 16:31, Amos Jeffries <squid3@treenet.co.nz> wrote:
> When we start rationalizing server-push semantics so multiple replies can be
> sent for one request. Most of the replies will need to be assigned some
> lower priority than the requested reply - hopefully all within the one
> stream. We will have to define a new response frame entirely to replace
> reply. Why not just do it now and have initial implementations compatible
> with those later ones?

I don't get where you are going with this.  It's already possible to
send multiple replies for a single request.  Each request gets its own
stream.

The server actually sends a SYN_STREAM for each extra "reply" (pushed
resource).  These streams are unidirectional.  The client doesn't get
to do anything to these streams, except reject them.  With push
promises, there is a push promise, followed by a later SYN_STREAM.

Currently, the server chooses priority for pushed streams, but our
Google friends (sorry I can't remember who specifically raised this,
probably Will) are interested in reprioritising streams, which would
allow a client to have a say about request priority.
Received on Friday, 22 February 2013 00:40:22 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 22 February 2013 00:40:24 GMT