Re: RDF Schema and XML Schema Part 2: Datatypes

[Aaron Swartz <aswartz@swartzfam.com>]

Jonathan Borden <jborden@mediaone.net> wrote:

> The only problem is that this version of DAML+OIL uses the wrong namespace
> for XML Schema, the RDFism: http://www.w3.org/2000/10/XMLSchema# -- while
> this 'works' for RDF the trailing '#' means that it is a completely
> different namespace ... that is, namespace names cannot be sort of equal, or
> almost equal, they either are or are not equal.
> 
> The practical effect of this is that XML Schema software will not recognize
> the document: http://www.daml.org/2001/03/daml+oil-ex-dt.xsd because the
> root schema element is not in the XML Schema namespace.

(I've added this as a comment to Eric's article, but it bears repeating.)

This (and possibly the whole discussion about making better Qname->URI
functions) is a non-issue. The DAML group simply needs to remove the # from
the namespace for everything to continue to work. Here's why:

The XML Schema document will have the correct namespace and be processed
properly by XSchema processors. RDF parsers can't read it anyway, and so
don't care.

The RDF document will have a trailing #, so the correct URIs will be
generated (as in those that reference the correct object). XML Schema
processors won't read the document, so there is no issue.

There is another issue that the XML Schema URIs as defined by the MSL are
quite strange, but I don't know how to solve this, and a better Qname->URI
function won't fix it. (Furthermore the argument that the correct URIs are
http://www.w3.org/2000/10/XMLSchemaunsignedint (or whatever) are bogus,
because XML Schema doesn't use RDF's composition function.

So now that we've debunked this issue, are there any _real_ examples where a
better QName->URI function is needed?

-- 
[ Aaron Swartz | me@aaronsw.com | http://www.aaronsw.com ]