xml-dev - RE: Bring back src? (was Re: simple question on namespaces.)

[Jonathan Borden <jborden@mediaone.net>]

Tim Bray wrote:
>
> I still think that dereferencing a namespace URI in an ideal
> world should yield something containing lots of human-readable
> explanatory text (ideally in a minimum of two widely-used
> human languages), plus pointers to lots of different
> machine-readable related resources, but I've been saying this
> for 2 years now and don't seem to be winning the argument. -Tim
>

	A namespace catalog would fill in the deficiencies in a content negotiation
system. In an ideal world, a user agent would be able to request such a
package, perhaps with a content type = (e.g.)
text/xml+namespace-description-package. This would allow a very general
catalog of namespace information including vocabulary, links to schemata
etc.

	Yet still there are requirements for more direct links to certain types of
schemata, particularly in RDF applications, e.g. where an instance is linked
to its rdf:type by a URI reference and where the fragment identifier
identifies the Class definition of the instance:

<RDF xmlns:foo="http://www.foo.org/theFooSchema#">
	<foo:bar ID="instance123">
		<foo:baz>abc</foo:baz>
	</foo:bar>
</RDF>

and theFooSchema.rdfs:

<Class ID="bar">
</Class>

	It is expected that "foo:bar" is expanded into the URI
http://www.foo.org/theFooSchema#bar In this situation the RDF
application/user agent could signal that it is interested in the RDF Schema
using content negotiation.

	I think that now that we have multiple types of Schemata, and hence several
use cases with differing needs, this is the time to reinitiate the packaging
discussion.

Jonathan Borden
The Open Healthcare Group
http://www.openhealth.org