RE: Documents, Cars, Hills, and Valleys

Manos Batsis wrote,
> Miles Sabin wrote,
> > In other situations there's no document for the URI to represent. 
> > This is often the case with namespace URIs, in which case they 
> > represent an abstract namespace or nothing at all. If they do 
> > represent an abstract namespace, then why would putting a 
> > retrievable document at the end of them defeat their original non-
> > retrievable reference?
>
> Seems that the XML Names spec gets all the blame for this week. 
> IMHO, non-retrievable namespace URIs should only be URNs. BTW, I 
> strongly believe that a namespace declaration should be used to
> bind a Schema or other (meta)information to a document (instead of 
> the current mechanism). I never understood why W3C decided against 
> this much more solid design.

I think you misunderstood my point. I'm now reasonably convinced that 
there is absolutely no problem with the namespace REC precisely
_because_ it's possible to think of namespace URIs as ambiguous
between abstract namespaces and (possibly absent, possibly multiple)
retrievable documents.

Context does the work of disambiguation here. In namespace processing
it's only the non-retrievable abstract namespace which does any work,
so any ambiguity in the URI is automatically resolved in favour of
the abstact namespace. OTOH when attempting to GET associated meta-
data, the abstract namespace is automatically ruled out because it's
non-retrievable, whereas an RDDL or schema document could be
candidate referents (and we might use Accept: to further 
disambiguate).

IOW, I think that acknowledging the ambiguity in URIs, far from being 
a problem, migth actually help us resolve some long standing problems.

Cheers,


Miles

Received on Wednesday, 10 April 2002 11:02:07 UTC