Re: GRDDL and OWL/XML

Greetings,

On 2008 May 16, at 17:52, Harry Halpin wrote:

> I think the central question is whether or not at this point the  
> GRDDL WG recommends that at least one GRDDL transform point to either:
>
> 1) A non-executable list of implementations
> 2) An executable transform, likely an XSLT one.
>
> One could of course have *two* GRDDL transforms, one that points to  
> 1) and one that points to 2), but we are not sure how 1) would  
> behave with current GRDDL transforms.

An apparently obvious solution here is to suggest that the  
transformation be retrieved using content-negotiation, so that  
dereferencing the GRDDL transform produces different documents  
depending on whether the request accept header included one or more of  
application/xslt+xml, application/x-javascript?, text/html, or even  
application/rdf+xml, with the last one potentially giving a machine- 
readable list of available transformers and their properties (of  
course, that machine-readable spec could be encoded within the text/ 
html version using GRDDL or RDFa, but this verges on the confusing...)

That way, if there's no XSLT transform available for a document, then  
the origin server returns 406 Not Acceptable, and if there's more than  
one transformer implementation, in different languages, it could  
potentially return 300 Multiple Choices (that's not realistically  
supported in browsers, but could be in a GRDDL library).

The example in the spec includes the XSLT transformation being  
retrieved with an accept:application/xml header, which is rather  
generic.

This seems so obvious that I'm sure I'm missing something, but the  
only mention of content-negotiation in the GRDDL spec is in the  
context of namespace documents, and there's no mention of MIME types  
at all, except rather in passing in the example.

Best wishes,

Norman


-- 
Norman Gray  :  http://nxg.me.uk
Physics and Astronomy, University of Leicester

Received on Saturday, 17 May 2008 12:39:06 UTC