RE: SYNTAX: RDF Syntax Telecon Friday

More substantive discussion.

Art said:
> I believe an automatic [golden] triple generator mechanism/program
> could be the WG's most significant deliverable wrt RDF interop and
> ultimately its long-term success.

It is unclear whether DaveB's message sees the golden triple or a spec as
the objective.

I think that the spec is much more important than the golden triple.
Any golden triple may be buggy; and is at some point, just another
implementation.

Earlier I privately posted about SNAIL.

> I currently have a subproject SNAIL fairly high on my todo list.
>
> The goal is to implement the transforming approach
> >  Transforming the Abbreviated Syntax into Very Basic RDF
> >  [6]
http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2001Sep/0224.html
>
> I am using xml source rules that get transformed by xslt into:
> both
> + a textual description of the same form as my message above
> and
> + an xslt program that will simplify an rdf file by one randomly chosen
step
> only
>
> SNAIL is a shell script that repeatedly applies the simplification
transform
> until no more rules apply; this is followed by a further transform to do
> aboutEach; finally there is another transform that outputs N-Triple.
>
> Currently SNAIL is almost entirely vapourware; some of the source rules
are
> written, none of the transforms are.
>
> SNAIL: the world's slowest RDF parser!
>

The objective here is to meet DaveB's second step:
> 2) Remove abbreviated syntax forms
(including propAttr, typedNode, nested descriptions, rdf:bagId, rdf:li)

but not as an XSLT program, but more as a specification. OK there is an XSLT
program, but that is simply as a confidence building factor. It conforms
closely with the spec, if it works and passes all the test cases then the
spec is more likely to be right. If it fails some test cases then maybe this
is indicative of bugs in the spec.

Ideally I would like to generate a form of the rules that could be
automatically tested for contradiction. (A single RDF/XML document
generating more than one graph). This would allow a computer generated proof
that the spec did in fact specify rather than under-specify the graph
corresponding to an XML document. An implementation cannot show that; since
it will give a result.

I am hoping that SNAIL will be non-deterministic in its choice of order of
rule application, this will make it easier to identify bugs in the spec.

Jeremy

Received on Tuesday, 2 October 2001 08:12:12 UTC