- From: Jeremy Carroll <jjc@hplb.hpl.hp.com>
- Date: Tue, 25 Jun 2002 15:17:10 +0100
- To: "RDF Core" <w3c-rdfcore-wg@w3.org>
I have implemented the new Syntax in ARP. I had a few issues, mainly with the test cases. One substantive issue with the text is as follows: ==== In M&S rule [6.21] we have: [[[ [6.21] IDsymbol ::= (any legal XML name symbol) ]]] where the hyperlink on XML name is to http://www.w3.org/TR/REC-xml#NT-Nmtoken In the new WD we have: http://www.w3.org/TR/rdf-syntax-grammar/#rdf-id [[[ An attribute ·string-value· matching any legal [XML] token Nmtoken ]]] Looking at it I think neither is correct and suggest that instead it would be better to match against: http://www.w3.org/TR/1999/REC-xml-names-19990114/#NT-NCName the namespaces NCName construct. I will describe the differences: NT-Nmtoken ========== Any non-empty sequence of NameChar characters XML name ======= Any NT-NMtoken that does not start with: Digit | '.' | '-' | CombiningChar | Extender NCName ====== Any XML name that does not contain ':'. ........................ Discussion ---------- A) If we allow an rdf:ID that matches more than NCName then it cannot be used in a qname e.g. <rdf:Property rdf:ID="300x" /> names the property with a name that cannot be used in an XML serialization. B) Allowing rdf:ID's that start in CombiningChar and Extender is likely to be at odds with our general position vis-a-vis charmod. (I have not checked that in detail). C) Both M&S and the current WD permit <a:foo rdf:ID="ex:ex"/> which looks distinctly dubious to me. Possible confusion as to whether or not the ID is a qname etc. Jeremy
Received on Tuesday, 25 June 2002 10:17:28 UTC