- From: Brian Behlendorf <brian@organic.com>
- Date: Wed, 8 Nov 1995 00:41:25 -0800 (PST)
- To: Kee Hinckley <nazgul@utopia.com>
- Cc: "M. Hedlund" <hedlund@best.com>, www-talk@w3.org
Sigh. Folks, we really need to get this system working. It's well past the deadline, we got an "incomplete" for the semester but the prof has given us another chance, so let's not mess it up this time, K? There are TWO chief weapons, er, excuse me, TWO chief things that browser vendors need to do to support content negotiation. 1) Browsers need to be honest about what they send. If you support text/x-netscape-html, say so! If you support application/director, say so! If you can shuffle director off to a helper app, but can't inline it, don't send it in the Accept:s of inlined object requests. You don't have to put every mime type in the accept header, just the most common. As Koen noted, if a browser is noticing that it's having to default to reactive content negotiation for a large number of hits it can adjust its Accept:s accordingly - it might also guess based on filename extension as an optimization, too. Because if none is acceptible, it can fall back on Thing #2: 2) A response mechanism for 406 None Acceptible. In <URL:http://www.ics.uci.edu/pub/ietf/http/draft-ietf-http-v10-spec-01.txt>, which has been obsoleted by the BCP HTTP/1.0 but from which most of the new parts of HTTP/1.1 will come, section 8.28 clearly outlines a system which would be adequate for our needs. Can we talk about its adequacy for the purposes of copntent negotiation, for which is was clearly intended, or are we going to keep passing the "406's format's not defined" buck until someone gets shot? If the definitions in that system appear alright, why not use it? What's not there that needs to be? If I get a few nodding heads we can commit to implementing it in Apache. I'm working on a "Content negotiation for dummies" paper to be circulated soon - I'd like to be a little more optimistic about things than the paper currently is. As for writing conditional HTML to be bugwards-compatible with all browsers - how are people using broken software supposed to know their software is broken? Makes me pine for a future where all objects render themselves. Brian --=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-- brian@organic.com brian@hyperreal.com http://www.[hyperreal,organic].com/
Received on Wednesday, 8 November 1995 04:14:35 UTC