- From: Roy T. Fielding <fielding@gbiv.com>
- Date: Wed, 8 Jun 2011 13:35:26 -0700
- To: Poul-Henning Kamp <phk@phk.freebsd.dk>
- Cc: Julian Reschke <julian.reschke@gmx.de>, Mark Nottingham <mnot@mnot.net>, httpbis Group <ietf-http-wg@w3.org>
On Jun 8, 2011, at 5:23 AM, Poul-Henning Kamp wrote: > In message <4DEF6761.1000605@gmx.de>, Julian Reschke writes: >> On 2011-06-08 13:57, Poul-Henning Kamp wrote: > >>> 203 should just be dropped. >>> ... >> >> Well, we just came to the conclusion a few days ago that we can't drop it. > > Then keep it, and give it a sensible definition, without inventing > a new class of unnecessary HTTP mangling beasts. > > Suggestion: > > 203 Non-Authoritative Information > > For reasons not specified, the response is not authoritative, > but expected to be sufficiently indicative of what the > authoritative response would have been, that it merits > attention. Users should be alerted to this fact if the > client decides to present the response. One example use > could be proxies returning expired copies when the origin > server is unreachable. No, these beasts have been with us since 1994. They are part of the design of HTTP. They are chosen by the client that use them. For all of those reasons, your suggestions for 203 are simply unfounded. Cache-control is not a substitute because it only applies to caches. Warning might have been a substitute but recipients ignore it. A status code other than 203 was originally chosen, IIRC, because caches would treat them differently than 200 responses even if they did not understand the semantics of 203. I have no idea how extensive that is in practice today, and I have no desire to change the semantics of HTTP right now just to find out what breaks later. ....Roy
Received on Wednesday, 8 June 2011 20:35:50 UTC