Lists Home |
Date Index |
>>>Rick Jelliffe said:
> Doh. I should have said "machine-exercisable" I guess.
> Meaning that there are formalisms that cannot be machine-checked
> and formalisms that can (to some extent). Look at the bogus
> formalism of the RDF specs' BNF: if they had been able to
> validate that life would have been much simpler. (I.e., given
> some valid RDF, you can point out which bits in the productions
> correspond to which bits in your data, but to actually parse
> the data using those productions is not at all straightforward,
> as the recent attempts to re-formulate the RDF-in-XML
> syntax have shown.)
What great timing.
I'm the editor of a W3C RDF Core WG working draft: RDF/XML Syntax
which is re-representing the existing syntax, using modern XML
technologies. We are now using a SAX-like sequence of events in the
style of the XPath 1.0 nodeset, derived from the XML Infoset.
Just today, I added a RELAX NG schema to my draft which I've got
machine validating the RDF test cases. It is at present a
Non-Normative part of the document which means it is for information
only and not the definitive declaration of the syntax. However, I
like it and I feel it should help a lot and is worth pursuing.
If you are curious have a look at my editors copy (very much work in
The XML version of the RELAX NG schema is
RDF Test Cases W3C Working Draft
-- The RDF/XML test cases that can be validated/rejected are
linked from here