- From: Ray Denenberg <rden@loc.gov>
- Date: Thu, 08 Mar 2001 15:40:14 -0500
- To: ZIG <www-zig@w3.org>
Dan Brickley wrote: > Could we use URIs for schemas/dtds instead of having a centralised > registry? That's two issue, as I don't see that using URIs to identify schemas and dts (an idea that I don't dissagree with), necessarily solves the registry problem. It does if you're talking about using http URIs. But (hopefully you would argree) we need a better long-term solution. Some URIs "address" (e.g. http) and others "identify". If we want to *identify* dtd/schemas, we would want to use URIs that "identify" wouldn't we? For example, we may someday want to convert all of our existing Z39.50 oids to URIs. One way that would be very straightforward (and therefore appealing) would be to cast each oid as "URN:OID:......" using the oid syntax of the new OID URN namespace proposal. But that doesn't solve the registry problem. I don't mean to trivialize this, I'm simply trying to make two points: (1) it's important to distinguish between URIs that "retrieve" and URIs that "identify". (2) converting oids to URIs doesn't solve all the problems of OIDs. --Ray -- Ray Denenberg Library of Congress rden@loc.gov 202-707-5795
Received on Thursday, 8 March 2001 15:40:08 UTC