- From: Larry Masinter <masinter@parc.xerox.com>
- Date: Tue, 13 May 1997 10:23:35 PDT
- To: "David W. Morris" <dwm@xpasc.com>
- Cc: Koen Holtman <koen@win.tue.nl>, http-wg@cuckoo.hpl.hp.com
It might be possible to have two URLs that reference the "same" resource, but that equivalence is complex and hardly anything that you want to have going on in your feature negotiation algorithm or protocol determination. The use of URLs at all in this area is somewhat suspect; yes, it's a distributed name space that doesn't require prior registration to use, but it's somewhat problematic over time. For example, let's suppose that NaviSoft had implemented an interesting protocol extension that was an integral part of the NaviPress web authoring tool; when AOL bought NaviSoft, the navisoft web server went away. The protocol extension has to have a lifetime that's longer than the lifetime of (at least some) domain names. So I'm wary of using URLs in PEP or in feature negotiation as a way of identifying content types and features. It's not that it is unworkable, but the issue needs to be addressed. Larry
Received on Tuesday, 13 May 1997 10:31:27 UTC