- From: David Carlisle <david@dcarlisle.demon.co.uk>
- Date: Thu, 1 Jun 2000 17:07:45 +0100 (BST)
- To: michaelm@netsol.com
- CC: xml-uri@w3.org
> This definition seems to say to me: You can't assume that the namespace > URI _always_ resolves to some resource but that it isn't prohibited either. and also, if it does resolve to a resource, you can't assume that that resource has anything much to do with the namespace (other than the fact that the same organisation is in some way responsible for both) However in certain restricted contexts, an application may specfy that certain namespaces do have some relevant resource at the namespace name. For example xsl engines might specify that the namespace uri identifies a resource which is the script implementation of the extension functions provided by that namespace. This is why I am not too concerned that using relative namespace names is not rdf friendly, as using uuid or similar schemes also causes problems for rdf's notion of sticking element names on to the end of namespace names and hoping to get a valid uri. It is up to rdf to say what are the possible namespace names about which it can reason, if it can't reason about all of them. Basically any namespace that is declared by xmlns:x="x" is probably of only transient interest anyway, and not the kind of thing one would want to set up rdf schema to describe. > It seems to be that this document at least made the attempt to discuss > our present problem(s). The question the document seems to leave > unanswered is, what does it consider to be definitive: > syntactic equivalence or functional equivalence? How can you say it leaves it unanswered? It goes to some lengths to stress that the character for character comparison of the xmlns attribute value is the only thing considered and that the functional behaviour of the URI is of no importance at all to the use of the URI reference as a namespace name. David
Received on Thursday, 1 June 2000 12:03:36 UTC