rdf:about, rdf:ID and anonymous resources

From : "McBride, Brian" <bwm@hplb.hpl.hp.com>
> What I was trying to say to Tim was there was nothing to 
stop
> him doing what he suggested, but that he might also 
consider
> the alternative of adding the fragment id's automatically.

Once we are on the subject,
I still have some problem with that ID thing.

What exactly is the point of writing, in a file foo.rdf
 <rdf:Description ID="bar">
As I understand it, it "defines" the resource
 foo.rdf#rdf
So, why not simply write
 <rdf:Description about="#bar">
Again, as I understand it, this is because the fragment 
id "bar" does not exist in the file foo.rdf,
so rdf:ID allows to define AND describe it.

Well, the fragment id "bar" does not exist in foo.rdf. So 
what ? Most of the time, rdf:ID is used (as far as I know) 
for classes or properties: resources that are abstract by 
essence, that can not be retrieved anyway ! Naming them 
foo.rdf#bar or whatever is nothing but a convention.

Defining a fragment id allows the URI of the resource to 
return the description of the resource ; but the 
description is a PROPERTY of the resource, not the resource 
itself. The metaporperty rdfs:isDefinedBy is intended for 
that, and XPointer allows to point to any rdf;Description 
tag in a more standard way.

 T(rdfs:isDefinedBy, my_resource, foo.rdf#xpointer(...))

So, did I miss something about rdf:ID ?

 Pierre-Antoine
______________________________________________________
Boîte aux lettres - Caramail - http://www.caramail.com