- From: Graham Klyne <GK@acm.org>
- Date: Wed, 02 Jul 1997 17:01:50 +0100
- To: Josh <josh@netscape.com>
- Cc: http-wg@cuckoo.hpl.hp.com
At 06:18 AM 7/2/97 -0700, Josh wrote: >> >> There would be a *presumption* that any server/proxy that added a trail >> entry claiming to be HTTP/1.1 (or whatever) would obey some defined set of >> rules. > >This is really the bigger problem. We need to detect a finer grain >than 'http/1.1'. We need to check for either compliance >or implementation of optional parts of the spec or enhancements. Maybe I missed something. I thought the fundamental problem was with pre-HTTP/1.1 proxies which did not properly downgrade the HTTP version number. Of course, we must always expect broken implementations (which was why I thought an implementationm ID should appear in a server trail). Further, any approach which relies on down-the-line co-operation will be susceptible to maliciously broken implementations (e.g., a broken "MoonSoft" proxy which masquerades as a popular Netscape implementation, or whatever). But I think even this might be detectable *if* one is prepared to go to sufficient lengths. (Probably further lengths than those to which Larry would be prepared to consent, judging by a recent posting.) GK. --- ------------ Graham Klyne GK@ACM.ORG
Received on Friday, 4 July 1997 21:42:41 UTC