Re: GRDDL implementation in Raptor

On Tue, 5 Apr 2005 11:29:19 -0400, Dan Brickley <danbri@w3.org> wrote:

> [snip]
> 
> This is great :)
> 
> I'm wondering about the prospects for GRDDL'ing RDF/A.
> 
> Specifically, I want the FOAF spec (a) to validate, (b) to embed 
> within its XHTML some RDF statements, as well as having a 
> content-negotiable index.rdf RDF/XML version. 
> 
> Do you know which versions of XSLT are supported currently in your 
> GRDDL implementation? Is it whatever libxslt handles? 

yes, it's whatever libxslt does.  I'm not writing one of them too!
http://xmlsoft.org/XSLT/ seems to suggest it's XSLT1 with some extras.

> I don't know if unconstrained RDF/A can be GRDDL'd in XSLT 1.0 due to the 
> qnames in attributes... I remember Jeremy and MaxF both started on 
> RDF/A parsers in XSLT though.

No idea, I've not been following the detail of that, and happily I don't
need to code it as if anyone writes the XSLT transform it'll just work :)

Re conneg and so on.  In situations where raptor has to guess a
parser to use (say, as a SPARQL data source URI) and fetch an url it
sends an Accept: to prefer rdf/xml and not offer a choice to get the
html one and transform it.  I'm betting having both rdf/xml and
xhtml+grddl conneged is rare.  You could always provide different
urls without conneg (/index.html, /index.rdf).

Also, in a protocol situation is it necessary/required to only apply
grddl transform if the xhtml mime type is correct?  Or try to read it
into an XML document whatever the type?

Dave