- From: David Carlisle <david@dcarlisle.demon.co.uk>
- Date: Mon, 5 Jun 2000 23:06:59 +0100 (BST)
- To: michaelm@netsol.com
- CC: xml-uri@w3.org
> but then declaring that your not going to also use the equivalence > rules the same rules are used as when comparing html href attributes (using xslt or the dom or anything else) that is string comparison. There is no equivalence rule for URI references specified in the RFC, only an algorithm for converting a URI reference and a base to an absolute URI. But for HTML href you only do that when (if) you want to find out what resource the thing refers to (ie you don't do it when parsing) and for namespaces you similarly only do it when (if) you decide to refer to the resource, which if you do at all you do after namespace parsing. The situations are just the same. > That's one of the main reasons we designed URNs. If you argue that it would have been better if namespace names had been urn or FPI or java package names, i probably would agree with you, but since the vast majority of namespace names are http URI then that has to be allowed. Yes it is possible to use a namespace name with a urn and 10000 year persistence, but it is also possible to live with cheap and chearful http naming which isn't very persistent (on that time scale) and makes some people think that they should be able to cook namespaces with orange sauce. This may annoy you if you have spent time designing naming schemes to avoid these problems, but I suppose they needed something explainable and simple to describe that could be used from day one, and picking the name of a resource (any resource) is what they chose. It may not have been what you would have chosen, but taht doesn't mean it is actually broken. > Sure. I'd be happy with that. Any suggestions for what that Base > should be? http://www.w3.org/1998/xml/namespace/base/ (backdating it to the same 1998 date as the xml namespace in the spec, but I don't really care, just anything, with a quick addendum to the namespace spec saying "... it is not a goal..." However if an application does wish to retrieve a resource using the namespace name then it may wish to ensure that the resource associated with a namespace name is the same in all contexts. As namespace names may use relative URI, an application should [or must or could, whatver is decided] use the base URI http://www.w3.org/1998/xml/namespace/base/ to convert relative URI references to absolute URI before retrieving a resouce. > Oh lordy no. If anyone is really suggesting comparing things > retrieved check the archives of this list:-) > BUT, http://www.w3.org/1999/XSL/Transform and > http://WWW.w3.org/1999/XSL/Transform are equivalent URIs and IMHO, > the namespace document should inherit that equivalence rule, not > try and come up with its own... I thought you were arguing a minute ago that different URI <-> different resource? People arguing for the absolute interpretation have always stated that this _only_ affected relative URI and _only_ involved adding the base and removing extra . and .. segments. Any change that made http://WWW.w3.org/1999/XSL/Transform a valid representation of the XSL namespace would be _totally_ unacceptable and would break all XSL systems. > Not necessarily. You could in RDF say that you are making assertions > about Aha, Hm OK makes sense, OK so I'll accept that for rdf relative uri are a problem, but supplying a base for all such uses would solve that I believe. David
Received on Monday, 5 June 2000 18:35:49 UTC