On Aug 18, 2005, at 8/18/05 8:11 PM, Renato Iannella wrote:
> I assume the requirement is interoperability and uniqueness for
> identifiers.
> And *a* solution is to ensure they are cast as URIs.
Certainly, yes... but we're at the point of having to assess whether
it's a workable solution. Personally, I'd like it very much if it
were, but we weren't certain that all addressing schemes had agreed-
upon URI representations... so we came up with the expedient of
identifying the scheme explicitly (and of course "uri" could be one
such identification).
We used the pattern of <valueListURN> / <value>*, just as you'd
suggested previously, except that in this case we couldn't guarantee
that the scheme would necessarily be in the form of a finite list or
would itself have a conventional URN designation, so literal reuse of
those particular elements didn't strike us as appropriate.
(And by the way, I'm not sure you're correct in saying, in a
different thread, that <valueListURN> doesn't contain a URN... I'm
pretty sure we specified that it should be, or at least we intended
to, else I don't think we would have named it so specifically.)
So anyway, if we could rely on all network addressing schemes having
well-known URI representations, even though we can't control those
schemes... or if we agreed to accept excluding the use of any that
don't... then I'd certainly support defining this as a URI.
- Art
