Re: URI references and aggregation

Martin,

I'm going to respond to your points here, partly in the hope of provoking a 
response that improves my understanding.  But you may notice from my other 
response that I had slightly mistaken the division between URNs and URLs, 
so my proposals are somewhat moot.

At 09:05 AM 1/11/00 +0900, Martin J. Duerst wrote:
>Just some very short points:
>
>- Unsing urn namespaces to solve problems of conneg syntax seems
>   not a good idea.

OK -- I see URNs are not enough.  But should they be excluded from a solution?

>- Restricting each class to a single urn namespace seems a bad
>   idea.

I was coming at this from the point of providing a common URI form 
associated with registered feature tags, rather than *restriction* of the 
syntax to specific URN forms.  Indeed, it was not my intent that there 
would be any changes to CONNEG syntax;  I was trying to run with Ted's idea 
of having some degree of uniformity in handling, while maintaining the 
fundamental distinction between features and aggregates.

>- Even if resolvability may not always be possible, and indeed
>   not always necessary, artificially making it difficult by
>   betting on the wide use of URNs and the associated resolver
>   infrastructure is also questionable.

Well, yes.  I was punting a bit with the idea of ad-hoc mechanisms, but I 
agree URNs are not needed for that.

#g

------------
Graham Klyne
([email protected])