Re: XMLLiteral belongs in RDF namespace, not RDFs

>>>Dan Connolly said:
> On Mon, 2002-11-11 at 03:29, Brian McBride wrote:
> [...]
> > That settles it then.  It stays where it is for now.
> 
> This is clearly new information since our 9Sep datatypes
> decision and our May decision on literals.

This is way-old information for me.

This is issue http://www.w3.org/2000/03/rdf-tracking/#rdf-namespace-change

  [[Summary: Some changes have been made to the RDF language (deletion
  of aboutEach*) and definition of terms (rdfs:domain,
  rdfs:range). This would normally call for a change of namespace
  URI's. If they are not changed, a strong case must be made.]]

which says "normally call for a change of namespace URI's"

This is more a W3C policy issue I see here, not a technical one.

We last discussed it near
  http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Apr/0466.html


> Please reopen them.

That's for Brian, who I think I also discussed this with too, many
months ago.  I recall we asked some senior W3C team member about this
but I can't remember the details.


> > The main reason: the more we *add* things of significance to the RDF
> > namespace (not syntax construction machinery such as rdf:datatype,
> > rdf:nodeID) and the RDF namespace URI RDF Schema document, the more
> > likely the URI of the namespace will have to change since the
> > namespaces's meaning
> 
> ???
> 
> I can't get my head around that phrase... "namespace's meaning".

I wish there was one that related
   (the set of names)
   the namespace URI
   the bits you get back from such a URI

It is such an undefined mess that any discussion of using XML
namespaces just ends up ratholing.  Basically

  If the (set of terms) associated with an XML namespace URI changes,
  should the namespace URI change?

In the RDF namespace, the set of terms has changed.  We removed
aboutEach and aboutEach and added some new terms.  The only new terms
that appear in the graph are rdf:first, rdf:next and rdf:List.  (and
as I noted, I asked that they didn't appear in the RDF namespace).
So the sets are close, but not the same.

> You have to be able to build an RDF parser without ever
> seeing the RDFS spec nor the RDFS namespace name, true?
> 
> And an RDF parser has to generate triples including
> the terms first/rest/nil/List and now XMLLiteral, no?

There is not a huge technical problem with any of this, but the
social one of modifyin the (implied?) promise given in the RDF M&S
REC that the RDF namespace is (set of terms) and then changing that
set.

Dave

Received on Wednesday, 13 November 2002 05:17:23 UTC