W3C home > Mailing lists > Public > ietf-http-wg@w3.org > October to December 2015

Re: http/2 and non-safe http methods

From: Willy Tarreau <w@1wt.eu>
Date: Wed, 4 Nov 2015 07:36:08 +0100
To: Patrick McManus <mcmanus@ducksong.com>
Cc: James M Snell <jasnell@gmail.com>, "ietf-http-wg@w3.org" <ietf-http-wg@w3.org>
Message-ID: <20151104063608.GD21740@1wt.eu>
On Wed, Nov 04, 2015 at 09:14:39AM +0900, Patrick McManus wrote:
> Hey James,
> 
> In this respect h2 isn't a lot different than h1 with parallel connections
> (which doesn't have any specified rules either). At least in my experience,
> the application needs to figure out above http whether the two requests
> need to be serialized (and hold back, if so).

I'd say that H2 is even more robust regarding this. In H1 the problem was
caused by the fact that the connection could be closed between two requests
and that you had no idea whether the unresponded request was processed or
not. In H2, you have one new stream for your request, so you don't have
the problem of this stream being closed by a keep-alive timeout or whatever
before the request is processed.

Willy
Received on Wednesday, 4 November 2015 06:36:37 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 11:11:40 UTC