W3C home > Mailing lists > Public > ietf-http-wg@w3.org > April to June 2008

Re: [google-gears-eng] Re: Deploying new expectation-extensions

From: Charles Fry <fry@google.com>
Date: Sat, 5 Apr 2008 10:41:37 -0400
Message-ID: <b549193f0804050741y1d5dce81s5ff046523b6349f5@mail.gmail.com>
To: "Adrien de Croy" <adrien@qbik.com>
Cc: "Julian Reschke" <julian.reschke@gmx.de>, "Brian McBarron" <bpm@google.com>, google-gears-eng@googlegroups.com, "Mark Nottingham" <mnot@yahoo-inc.com>, "HTTP Working Group" <ietf-http-wg@w3.org>

As you call out, the real problem here is HTTP/1.0 and lower proxies.
For that matter, it seems that the primary focus of section 8.2.3 is
to deal with intermediate HTTP/1.0 and lower proxies (otherwise the
origin server could simply do all the work, and Expects wouldn't even
be needed).

So why would this be necessary for 100 Continues but not other 1xx responses?

Charles

On Fri, Apr 4, 2008 at 9:24 PM, Adrien de Croy <adrien@qbik.com> wrote:
>
>  the spec requires that any intermediary pass on 1xx messages, and that any
> number of 1xx messages may be sent
>
>  So any intermediary thinking that a 1xx response is the final one, is
> non-compliant.
>
>  Having said that, there are some that behave like that (possibly HTTP/1.0
> proxies?)
>
>  Adrien
>
>
>
>
>  Charles Fry wrote:
>
> >
> > >  Can't the origin server just send the 103s without being asked for it?
> That
> > > would allow the client to discover support for the feature.
> > >
> > >
> >
> > Hmm. Now this is starting to come full-circle. As I understand it the
> > whole reason that Expect: 100-continue is used in conjunction with 100
> > Continue responses is to ensure, as the request is finding its way to
> > the origin server, that the response will be able to find its way
> > back, being properly interpreted as an intermediate response. Without
> > this there is the risk that a non-100-continue-aware proxy would
> > interpret the 100 response as a final response.
> >
> > Is this not a requirement of any client-elicited 1xx response? I.e.
> > can we really just send 103s when they aren't asked for, with full
> > confidence that they won't break anything as they travel back to the
> > client?
> >
> > Charles
> >
> >
> >
>
>  --
>
>  Adrien de Croy - WinGate Proxy Server - http://www.wingate.com
>
>
Received on Saturday, 5 April 2008 14:42:22 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 06:50:46 GMT