Re: QName Problem Isn't One

Danny Ayers <> wrote:

> What you describe sounds like it will allow reference to concepts from XML
> Scema in practice, which is good enough in my book.
> Though I do reckon it depends on what you want to do with the bits of XML
> Schema in RDF. I don't think you can take bits of one language and paste
> them in another and expect it to automatically make sense.

I agree, the suggestion was to use the XML Schema URIs on typed nodes,
creating nodes with the type of an XML Schema datatype. This is perfectly
reasonable, as these URIs are defined as being able to do just that in the
XML Schema spec.

> <- The main point -- that the same URI is used for the same concept -- is
> <- resolved. The secondary point -- that a different namespace is
> <- used in each
> <- document -- is done correctly, since to use the same namespace would
> <- needlessly confuse processors.
> this sounds rather contradictory to me, as though you're saying that
> are both the same and different. Ok, some systems might interpret them the
> same way and others differently, but that's not quite the same thing.

No, I'm saying that they are both differently, and should be used
differently. Some insist that for RDF and XML Schema to work together, they
must be the same thing. I was simply making my difference of opinion clear.

> Since the XML Namespaces spec assigns no
> <- semantics to namespaces, there is no misuse or abuse of the XML Schema
> <- namespace. In the end, everything works beautifully.
> But aren't there semantics attached to the concept you're taking from XML
> Schema (defined in the context of an XML Schema) - what defines them in an
> RDF document?

The fact that XML Schema gives the datatypes URIs. If we were to say X
rdf:type Y and Y was a datatype, I think the semantics are rather clear.
(Using them as properties is something else entirely.)

I hope this helps.
[ Aaron Swartz | | ]

Received on Sunday, 6 May 2001 11:26:45 UTC