Re: RDDL: new natures

�Elliotte Harold wrote:

Jonathan Borden wrote:

When I say that the rddl:nature of http://example.org/foo.xsd is "XML Schema", this is intended to assert that it is reasonable to assume that http://example.org/foo.xsd ought comply with the "XML Schema" specification i.e. validate as an "XML Schema".

I believe this to be sufficiently asserted by xlink:role="http://www.w3.org/2001/XMLSchema"

What I *don't* want to say is that <http://example.org/foo.xsd> is a member of the XML Schema namespace.�

Good. xlink:role="http://www.w3.org/2001/XMLSchema" does not say that.

In fact, I'm not sure anything would. URLs and documents are not generally considered to be members of a namespace. The document at http://example.org/foo.xsd could say that the root element is a member of the namespace with a xmlns="http://www.w3.org/2001/XMLSchema" attribute; but that's a very different thing.

Using <http://www.w3.org/2001/XMLSchema> as the URI for the nature of "XML Schema" creates this ambiguity for ***software agents***.�

In practice XML software agents are indeed smart enough to distinguish between xlink:role="http://www.w3.org/2001/XMLSchema" and xmlns="http://www.w3.org/2001/XMLSchema" and even xlink:href="http://www.w3.org/2001/XMLSchema". I don't think there's any ambiguity here we need to worry about.

In reconsidering this, and also in light of Leigh's comments, I am coming to think that both of you are correct.

My original thinking (which has clouded my own understanding of this issue) was that the use of RDDL Nature (xlink:role) was equivalent to asserting an <rdf:type> between the related resource and the nature URI. This was in accordance with Ron Daniel's W3C Note on Harvesting RDF statements from XLink�http://www.w3.org/TR/xlink2rdf/ , informative reference 9.2 in the RDDL spec.

On the other hand Norm Walsh provides an alternate view in�http://www.w3.org/2001/tag/doc/nsDocuments/ specifically the "RDDL Model for Docbook"

If we adopt this model (NDW/TAG) then there is no real problem using a namespace URI as the rddl:nature. (rddl:nature is now a plain 'ole property).

So what I currently propose is that we continue to allow/recommend RDDL Natures like http://www.w3.org/2001/XMLSchema *and* that we change the documentation (and rddl2ref.xsl) to reflect the model�http://www.w3.org/2001/tag/doc/nsDocuments/