- From: dehora <dehora@eircom.net>
- Date: Fri, 7 Sep 2001 04:22:54 +0100
- To: "'RDFCore'" <w3c-rdfcore-wg@w3.org>
> Jan Grant: > > > On Mon, 3 Sep 2001, dehora wrote: > > > 4: prefixes give futureproofing. For example with an 'rdf:' prefix, > > infoset would have been a snap to add this time around > > (parseType="rdf:Infoset"). It makes architectural sense to > allow a way > > for future wgs to feed back known useful value > interpretations into the > > standard. > > Since namespaces don't leak into rdf:parseType attributes, do you want > "foo:Literal" > to be the same as "rdf:Literal" if "foo" is declared as "the same" > namespace as "rdf"? [...goes away and rereads namespaces and schema...] Short answer: yes. Might that imply additions to RDF Schema? Possibly. A mandated string seems short sighted, having the namespace binding allows one to make RDF assertions about an attribute value...that fits in with describing RDF in RDF, or even, allowing more complex parseTypes to been created (not sure that's a good thing actually). And really, if we're MUSting namespacing RDF attributes...may as well be hung for a sheep as a lamb. This isn't a controversial technique anymore, but one issue come to mind. Lack of tool support for qualified attributes; problems might arise with RDF-XML fragments and transforms. My thinking on this is that most XML plumbing will be upgraded to cater for XML Schema (which requires attribute namespace qualification) since most XML plumbing will want to play the (XML Schema dependent) web services game sooner rather than later. And we can have words indicating that the prefix 'rdf:' is a convention, in the same sense that 'xsd:' is a convention. Bill de hÓra
Received on Thursday, 6 September 2001 23:24:06 UTC