RE: Resources and URIs

> -----Original Message-----
> From: ext Tim Bray [mailto:tbray@textuality.com]
> Sent: 07 May, 2003 21:56
> To: Stickler Patrick (NMP/Tampere)
> Cc: LMM@acm.org; sandro@w3.org; uri@w3.org
> Subject: Re: Resources and URIs
> 
> 
> Patrick.Stickler@nokia.com wrote:
> 
> > URIs used as XML Namespaces do *not* denote those namespaces.
> > 
> > XML Namespaces are punctuation. They are macros. They are
> > syntactic sugar for XML instances so that URIs can be
> > used as element and attribute names (by a little bit of
> > slight of hand).
> 
> You have now expressed this opinion on many occasions dating 
> back quite 
> a few months.  I remain unconvinced.  Yes, the main utility of 
> namespaces has been to sneak URIs into element/attribute names to 
> support dispatching and collision avoidance.  But if we choose to say 
> that the URIs thus used do actually identify resources, and provide 
> representations of them, this breaks nothing and is actually 
> useful in 
> the real world.  So why not?

It is not about whether the URIs used as XML Namespace names do or
do not denote resources, and whether representations of those
resources can be obtained, but whether the resources denoted by
a URI used as an XML Namespace name has any relation whatsoever
to either the namespace (the simple set of names), some vocabulary,
any of the resources denoted by any of the URIs grounded in that 
namespace, etc.

Let me give you a practical example that is 100% in conformance
with the XML Namespace and RDF specs:

We have defined a vocabulary which is denoted by the URI

   http://sw.nokia.com/FN-1

That vocabulary has numerous terms, some of which are grounded
in the XML Namespace name

   http://sw.nokia.com/FN-1/

when referred to in RDF/XML instances as element or attribute names.

Most of the terms belonging to that vocabulary, BTW, are not in
fact grounded in that XML Namespace name, yet are just as validly
terms of that vocabulary than the terms grounded in an XML Namespace
name that is lexically similar to the URI denoting the vocabulary.

The URI http://sw.nokia.com/FN-1/ is not defined (by us at least)
to denote anything. It is not the name of the vocabulary.

Certain individuals, however, seem to want to dictate that the
above would be considered incorrect. That one must use XML Namespace
names that end in '#' and denote terms of vocabularies using
fragment identifiers having the URI used for the XML Namespace
name as the base URI. Well, fine, if folks want to do it that way,
OK, but don't expect to be able to force other folks to have to
do it that way when the *SPECS* do not require it.

Positing that an XML Namespace name denotes a vocabulary and one
can get information about that vocabulary via that URI is *NOT*
reflected in either the specs nor in overwhelming common practice.

This position also misses entirely the relationship between
terms, vocabularies, models, schemas, and other resources all
of which have no need to depend on the XML Namespace names used.

If you want to know about the resource denoted by

   http://sw.nokia.com/FN-1/compatability 

you need not concern yourself at all with any XML Namespace name.
Such things are simply not relevant on the SW.

Rather, you ask the server sw.nokia.com to tell you about that 
resource. E.g.

   http://sw.nokia.com/URIQA?uri=http://sw.nokia.com/FN-1/compatability

and you learn that it is an rdf:Property and that it is also a term
that belongs to the vocabulary http://sw.nokia.com/FN-1/Examples which
is a sub-vocabulary of http://sw.nokia.com/FN-1 which are defined by
various schemas encoded in RDF/XML etc. etc.

Nowhere is any XML Namespace name that was used in any RDF/XML serialization
relevant to any knowledge about the resource denoted by any of those URIs.

Rather than using in my RDF/XML

   <x:compatability xmlns:x="http://sw.nokia.com/FN-1/">

I could just as well say

   <x:bility xmlns:x="http://sw.nokia.com/FN-1/combata">

and the semantics would be *identical* as either is just a means
to express

   <http://sw.nokia.com/FN-1/compatability>

which simply is not legal XML.

XML Namespace names are simply irrelevant to the SW so I wish folks 
would stop trying to posit some kind of special role for them and
special relationship between them and terms expressed in XML using them
etc. Any such role or relationship, insofar as the SW is concerned,
is an illusion.

Regards,

Patrick

Received on Friday, 9 May 2003 07:06:51 UTC