xml-dev - RE: Content negotation in namespace URI resolution was RE: TreatyofWulai

[Jonathan Borden <jborden@mediaone.net>]

Tim Bray wrote:
>
>
> At 11:52 AM 28/12/00 -0500, Jonathan Borden wrote:
>
> >        Content negotiation has been suggested. Personally I
> think it is a good
> >idea because the responsibility for how semantics are applied
> lies with the
> >user agent.
>
> I used to believe this, but I think that in practice it unfortunately
> doesn't work.  Reason is that content-negotiation dispatches on
> media type, and I can easily imagine a case where I have several
> different resources of the same media type that might apply.  E.g.
> I might have an authoring-time, an auditing-time, and an execution-time
> XSD.  XHTML has at least 3 DTDs.  Sigh. -Tim

	Sure. Content negotiation is very much like the <?xml-stylesheet .. ?> PI
which also dispatches on media type. It doesn't always provide the
stylesheet you want and in such cases there is no substitution for an
external process which applies a stylesheet or schema to an XML document.

	On the other hand, just because it wouldn't work in all cases, doesn't mean
the technique isn't generally useful. Perhaps my bias is toward RDF Schemata
where I expect the resolved schema to actually tell me something about my
document. Perhaps a use case for XSD might be if an XML document arrives on
my doorstep and I might wish for a schema to help me give it a proper home
in a database. Resolving a namespace URI into an XSD might allow me to store
the document more efficiently than if I were to use a generic mechanism
(e.g. if the namespace URI resolution fails).

	A more serious problem with content negotiation is that it depends on
protocol support i.e. it only works with HTTP URLs in common practice. I
wouldn't expect a namespace of the form data:... to resolve differently
given a different 'request' content type. But thesedays I'm happy to find a
solution that works even only sometimes :-)

Jonathan