Re: [#issue-output-formats] Four options for discussion

On Wed, 2006-09-20 at 08:23 -0400, Harry Halpin wrote:
> Since we've promised to discuss this issue of output formats today[1], 
> maybe we can clarify some of  our options. Here's how I see some of them, 
> although I'm sure this is incomplete at best:
> 
> 1) GRDDL Processors should output only RDF in the RDF/XML concrete syntax.

I think you mean GRDDL transformations, right?

> 2) GRDDL Processors should output the RDF/XML concrete syntax and may to 
> other W3C-recommended concrete syntaxes for RDF.
> 
> 3) GRDDL Processors should output to a W3C-recommended concrete syntax 
> for RDF.
> 
> 4) GRDDL Processors can output to any concrete syntax for RDF.

The way I would probably phrase it in the spec is to say
that the grddl:transformation property relates a document
to an algorithm whose input is XML
and whose output is RDF abstract syntax.

Hmm... XML infoset? or DOM? or XPath data model? I wonder
if it matters. But I don't think it should be raw
XML syntax; i.e. a GRDDL algorithm shouldn't have
different output for <abc/> and <abc  /> .

> Also, we should clarify the level of our mandate - i.e. whether we should
> use stronger or weaker language that "should" - and so clarify whether 
> a certain output syntax is recommended or required.

Note that "should" is for constraining agents in protocols;
sometimes the right verb is just "is".
i.e. there's no reason to say "2 + 2 must be 4" when we
can just say "2 + 2 is 4".

> 
> So, for option 1), any GRDDL Processor that produces something besides 
> RDF/XML (i.e. RDFa or n3) would be going against the recommendation. 
> However, assuming RDFa or n3 becomes at some point in the future W3C Recs 
> (which is likely for RDFa), then option 2) would also recommend these, 
> although we suggest that RDF/XML be preferred if all other things are 
> equal.  Currently option 3) is the same as option 1) since, well, RDF/XML is the only 
> W3C-approved syntax, but again this could change in the future - so 3) is 
> just a less contrained version of 2). 4) is the most unconstrained, as it 
> allows any person to design their own syntax for RDF (possibly without 
> publishing their "spec"), and then have their GRDDL output to that syntax 
> - and still be within the recommendation.
> 
> [1] http://www.w3.org/2004/01/rdxh/spec#issue-output-formats


-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/
D3C2 887B 0F92 6005 C541  0875 0F91 96DE 6E52 C29E

Received on Wednesday, 20 September 2006 12:49:22 UTC