W3C home > Mailing lists > Public > ietf-http-wg@w3.org > April to June 2010

Re: Expect: 100-continue and proxies

From: Henrik Nordström <henrik@henriknordstrom.net>
Date: Sat, 15 May 2010 10:14:00 +0200
To: ietf-http-wg@w3.org
Message-Id: <1273911240.9968.43.camel@localhost.localdomain>
fre 2010-05-14 klockan 13:41 -0700 skrev Roy T. Fielding:
> On May 14, 2010, at 1:12 PM, Henrik Nordström wrote:
> 
> > fre 2010-05-14 klockan 12:31 -0700 skrev Roy T. Fielding:
> > 
> >> A 100 message may be hop-by-hop if the intermediary has request
> >> storage capability -- requirements placed only on origin servers
> >> imply that all other servers are unconstrained, not that all
> >> other servers are MUST NOT.
> > 
> > Ok. One follow up question on that. Is it OK to respond with 417
> > Expecttion Failed after 100 Continue have been sent?
> 
> Sure, it is a valid status code.

Right. Just me putting it in the "clients won't expect this" category
based on the poor experience regarding 417 handling in general.

Not sure if I mentioned this before, but specifications probably should
clarify that if a proxy adds Expect: 100-continue to the request then it
must also be prepared to deal with the 417 response this may result in.
A proxy adding Expect and then forwarding the resulting 417 response to
the client would break things badly. Now this is already covered by the
specs in the sense that a proxy is both a client & server, but it's
still very easy to overlook.

Regards
Henrik
Received on Saturday, 15 May 2010 08:14:55 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 06:51:18 GMT