Re: Latest XLink Candidate Recommendation

Simon,

As always, I appreciate your thoughtful writeup.  We in the Linking group 
will definitely have to look at the issue, though my understanding agrees 
with yours (that RDF assumes absolutizing because an RDF schema may exist 
at the other end, and that XLink was getting this behavior "by reference").

One comment: If you want to use XLink linkbase arcs, you won't be able to 
"avoid using role and arcrole entirely," since an arcrole value is 
essential for making the right behavior happen.  In other words, this small 
part of XLink can be thought of as a tiny layered XLink application that 
dictates a particular semantic role.

         Eve

At 08:31 AM 7/8/00 -0400, Simon St.Laurent wrote:
>At 06:13 PM 7/7/00 +0200, Eric van der Vlist wrote:
> >If I bring it here, it's because I wonder if the answer shouldn't be
> >coherent with the one which is given to the namespaces URIs and that
> >this forum has already worked a lot on a similar subject.
> >
> >IMHO, it would be preferable to keep them consistent (the concepts would
> >be easier to explain to XML users and to implement in XML tools).
>
>I agree that keeping them consistent would be easiest, but I'm not sure the
>solution to the namespaces problem that's moving ahead actually provides an
>answer that's appropriate to XLink.
>
>Once again, we're using URIs to identify semantics.  In the XLink case,
>however, it seems like processors likely should have access to
>absolutization algorithms, though that process isn't specified.  It seems
>like this recent move is an attempt to align XLink with RDF, where
>absolutization appears to be the common understanding.  (Retrieval of some
>unidentified document may be an excuse to push that interpretation.)
>
>On the other hand, it seems worthwhile to consider relative URIs as
>something other than 'shortcuts' - as representing claims that the
>description provided is purely local, without global context.  Since
>they're being used for semantics, I think that approach may well make sense.
>
>This "URI reference identifies some resource that describes the intended
>property" seems like yet another leap off the cliff, providing
>underspecified functionality that leaves us all pointing to various random
>junk that may or may not be useful for processing.
>
>While it may be a useful bridge to RDF, I'd tend to suggest that people
>avoid using role and arcrole entirely until these issues (including what
>lives at that URI) are resolved, and stick to the weaker label, which
>carries no such luggage.
>
>Simon St.Laurent
>XML Elements of Style / XML: A Primer, 2nd Ed.
>http://www.simonstl.com - XML essays and books

--
Eve Maler                                    +1 781 442 3190
Sun Microsystems XML Technology Center    elm @ east.sun.com

Received on Monday, 10 July 2000 10:38:06 UTC