- From: Richard Light <richard@light.demon.co.uk>
- Date: Mon, 18 Jan 2010 12:02:25 +0000
- To: semantic-web@w3c.org
Hi, In writing a Turtle parser which serializes to RDF/XML, I'm hitting issues where RDF/XML can't represent some Turtle constructs. No surprises there, but I would welcome guidance on one specific issue. The null prefix construct (qname where the prefixname is empty), e.g.: @prefix : <http://example.org/> . ... :a1 :b1 :c1 . has the air of a default namespace about it. The temptation is to put: <rdf:RDF xmlns="http://example.org/" ... at the head of the output, output the triple as: <rdf:Description rdf:about="a1"><a2 rdf:resource="a3"/></rdf:Description> and consider the job done. Am I right in concluding that this is a snare and a delusion, and that the null prefix should simply be expanded each time it is encountered, and NOT treated as a namespace of any sort? My reason for drawing this conclusion is that declaring <rdf:RDF xmlns="http://example.org/" ... would cause this namespace to be incorrectly applied to all relative URIs in the document, not just those which have been declared as having a null prefix. If true, this means that the null prefix cannot be treated in the same way as non-null prefixes, which you can declare as namespaces in the RDF/XML representation. (This is rather significant, since predicates are represented as element names in RDF/XML, and so cannot be full URIs.) Richard -- Richard Light
Received on Monday, 18 January 2010 12:06:52 UTC