- From: Uche Ogbuji <uche.ogbuji@fourthought.com>
- Date: Thu, 10 Jan 2002 08:55:56 -0700
- To: "Jeremy Carroll" <jjc@hplb.hpl.hp.com>
- cc: "Uche Ogbuji" <uche.ogbuji@fourthought.com>, "www-rdf-interest" <www-rdf-interest@w3.org>
> > I was not in the WG at the time the decision was made. But seeing no other > reply I will have a go. > > > If unqualified attributes are allowed then there is a specific problem with > cases where the element is not qualified in the RDF namespace. Unqualified attributes should only be allowed on elements that *are* in the RDF namespace. This is the XSLT approach (see literal result element as stylesheet, in particular) So, no more problem, yes? > In the other standards you mention, I believe that unqualified attributes > are used on elements that themselves belong to the standard's own reserved > namespace. > > Hence, I think there was a problem with M&S and four possible solutions: > > A: all attributes must be qualified > B: attributes of elements in the rdf namespace need not be qualified, > attributes of other elements should always be qualified This is the solution I'm advocating. > C: the reserved attribute names in an RDF/XML document always have their > special meaning, whatever the namespace of the element. Oh, but this would be beyond horrible. > B and C both seem to be in tension with the XML Namespace spec. I don't see how B does. > Unqualified > attributes do not share the namespace of their element. Why must they? XInclude attrs, for instance, are not in the XLink namespace from the Infoset POV, but XInclude processors recognize the attributes from their owner element. You are similarly free to pile on your own attrs, if you like, but they *must* be qialified: <xi:include href="foo.xml" parse="text" my:extattr="spam"/> I see no reason why this solution is not fit for RDF: RDF processors will have to be at least as sophisticated as XInclude processors, I would think. > Whereas B would > equate an ID attribute on an rdf:Description element with an rdf:ID on a > foo:bar element. Yes it would. And this does not violate anything in the namespace REC by itself. The RDF/XML syntax would need slightly new wording, but that should hardly be a problem. I have to run now, so I'll draft an example this afternoon. > C conversely would suggest that RDF/XML can define the > semantics of an attribute which modifies something not in its namespace. I'm not sure how C could ever come into sane contemplation. ;-) Thanks for the response. -- Uche Ogbuji Principal Consultant uche.ogbuji@fourthought.com +1 303 583 9900 x 101 Fourthought, Inc. http://Fourthought.com 4735 East Walnut St, Boulder, CO 80301-2537, USA XML strategy, XML tools (http://4Suite.org), knowledge management
Received on Thursday, 10 January 2002 10:58:16 UTC