- From: Tim Bray <tbray@attglobal.net>
- Date: Sat, 24 Jun 2000 22:52:33 -0700
- To: xml-uri@w3.org
Namespaces were originally designed to work as globally unique names. They do that pretty well. URI-reference syntax has all the right technical properties. Dan and TimBL (at least, probably more) think that this use of URIref syntax violates the spirit of the URI framework. I think I understand that framework pretty well (and have learned more about it in the last weeks), and I think I understand their arguments, but I still don't agree. I do agree with Dan on the best way forward right now (some variant of deprecate/undefined). And I *don't* think that we should have used java class names or URNs or some other sideways dodge. Because, I have a dream. I dream of writing software that when it runs into a namespace name that it doesn't recognize, can use the name as a URI, dereference it, and (maybe, no guarantees in the Web architecture) get something back that will help something useful happen. Sometimes names are all you're gonna get, and sometimes there's nothing much more useful that can be said. But the current practice leaves the door open to doing more. I think the W3C should make a serious priority of getting to work on figuring out how to make this a reality. I think that the future goal of useful dereferencing of namespace names should *not* get in the way of the current goal of using them as globally-unique names. For that reason, relative URI references in this role should be at least deprecated. Weirdly enough, sometime in the last 1000 messages I swung over from the "deprecate" to the "forbid" camp, becaus the moral problems involved in blessing their use in this application exceed those of forbidding them. But I recognize that this position is probably extremist and can't command consensus. -Tim PS: [Note to TimBL for (i = 0; i < 1000; i++) { my $j; foreach $j (0 .. 1000) { for (int k = 0; k < 1000; k++) System.out.println("I am not against relative URIs!"); } } printf(".... except in namespace names\n".); ]
Received on Sunday, 25 June 2000 01:52:36 UTC