Re: docsync: RDF URI Reference

At 13:37 04/11/2002 -0600, Dan Connolly wrote:
>On Mon, 2002-11-04 at 13:31, Brian McBride wrote:
> >
> > What do we call the things in an RDF graph that identify resources?
> >
> > These things are absolute URI's with optional fragment identifiers.  They
> > are not URI references, because URI references can contain a relative
> > URI.
>
>quanitifer bug.

Right.  The term URI Reference is insufficiently restrictive to precisely 
define what is and is not allowed in an RDF graph.


>That doesn't mean absolute URI references are not URI references.
>
> >  The concepts doc has been using, in part, the term RDF URI References
> > for these things.  Is that a term everyone feels comfortable with using?
>
>no.

K.  I'm inclined to go with it in the absence of a more acceptable alternative.


> > When talking about the RDF/XML syntax, whether in the syntax doc or the
> > primer, or anywhere else, we must be careful how we describe the value of
> > rdf:about and rdf:resource attributes.  The issue here is that we do not
> > process them the same way as the URI spec says.  We do not treat a 
> fragment
> > identifier as a same document reference.  It is important that we are 
> clear
> > that we are not claiming to conform to rfc 2396.
>
>what!?!??!
>
> >  We process RDF URI
> > references and URI references with a relative URI the same way as rfc 2396
> > and XML base specifies.  We process fragment identifiers differently.
>
>I don't think so.

Great.  It would be good if we were conformant.  The issue is fragment 
identifiers, right.
Ah.  Maybe its the xml:base rec we need to be sure to be correct about.

RFC 2396 defines a fragment identifier to be a same document reference.  At 
least one TAG member is prepared to argue that saying the URI for a same 
document reference is influenced by xml:base is wrong.  Now we can either 
say that a same document reference is translated to an absolute uri 
reference with optional fragment identifier using an algorithm involving 
xml base, in which case that tag member should rightly comment at last call 
and is likely to receive support, or we can say that we don't claim that 
the thing is a same document reference, its a fragment identifier and we 
will process it how we choose.

That is currently how I see things, but I'm open to better suggestions.

Brian

Received on Monday, 4 November 2002 17:51:11 UTC