Over the holiday break members of XML-DEV have decided to put a constructive end to the controversy over what should be "at the end" of a dereferenceable namespace URI.
The activity commenced after a long and involved discussion centering around the practice of placing XML Schemas at the end of HTTP URIs for namespaces, which has caused much disquiet in the XML community, not least because there are many other kinds of metadata aside from a schema that one might wish to provide.
Acknowledging the desire that most people want a namespace URI to point
to something, Tim Bray suggested that XML developers could create
an XML application to describe namespace-related resources.
I think that if you got a group of qualified people, mostly
computer programmers, together, you could cook, in a few
weeks, cook up a nice little markup vocabulary containing:
- a few basic chunks of XHTML
- the simple XLink element
- namespace in which are defined a bunch of reserved
role attribute values
He also commented:
If I were feeling particularly grandiose, I'd also describe
such a thing as a key building block for the Semantic Web.
But it would have to be done low, fast, and under the radar...
This challenge has so far yielded three proposals for such a language. The first came from Jonathan Borden:
I have written up a brief proposal for an XML Catalog specification based on
our discussions. Note, the proposal is in the XML Catalog format and is
located at:
http://www.openhealth.org/XMLCatalog/
The second proposal (a few hours later) came from Tim Bray. XNRL, XML Namespace Related-resource Language, has a strong emphasis on human-readable documentation:
XNRL is defined to be a variation of XHTML Basic. The main change is that certain div elements, which must be children of the body element, also serve as XLinks to related resources. These div elements are intended to contain human-readable descriptions of the related resources being linked to.
Sean Palmer, who had welcomed the two previous proposals, but queried their particular utilisation of XHTML, then proposed a third solution, XML Namespace Catalogue Language, which attempts to meld the earlier proposals.
Discussion and refinement is continuing, but it looks as though the XML community could be well on the way to making the use of resolvable URIs for namespaces a lot more comfortable.