Re: Summary of the QName to URI Mapping Problem

I think we've found the source of disagreement:

> > 1.  A QName identifies a (single) resource.
> I don't take that as an axiom. I don't see why one would.

From the XML Namespaces spec:

"An XML namespace is a collection of names, identified by a URI reference
[RFC2396], which are used in XML documents as element types and attribute
names."

So QNames are names (duh! :-) ).  A names has to name something.  In this
case, QNames name element types and attributes (well, really, attribute
types, but that's just terminology).  In RDF-speak, element and attribute
types are resources (as is everything else).  So QNames identify (specific
kinds of) resources.

Furthermore, the whole point of namespaces is to avoid "collisions" between
names.  A "collision" is when a single name identifies more than one thing.
Therefore, unless the XML Namespaces recommendation doesn't achieve its own
stated goal, a QName must identify a single resource.

I do think that the recommendation could be more explicit about this, but I
don't think I'm making anything up.  I certainly believe that QNames are
more than a purely syntactical construct.

> I am aware of specs that are inconsistent
> with this notion; for example, XML Schemas
> map the same qname to different element types depending
> on context.

I don't know XML Schemas, so I can't say much about this.  However, is it
possible that the QName identifies a single XML element type, and then the
combination of XML element type and "context" identifies some Schema element
type?

        -- P.

--
  Piotr Kaminski <piotr@ideanest.com>  http://www.ideanest.com/
  "It's the heart afraid of breaking that never learns to dance."

Received on Thursday, 16 August 2001 18:42:26 UTC