Re: RDF in parts of XHTML documents

Le mar 09/03/2004 à 23:15, Dan Connolly a écrit :
> We've talked about pushing the tx_ptr down into the profile
> or namespace document; I think Dom implemented that

I implemented only the namespace document case; see
http://www.w3.org/2004/01/rdxh/grddl-xml-demo and specifically the "P3Q"
example.

> I want to do that for the HTML profile case too; e.g. add
> one pointer from the XFN profile document to grokXFN.xsl,
> so we don't need a pointer to grokXFN.xsl from each
> XFN document.

So do I; I believe it's fairly easy to add in GRDDL (and the associated
software) and would have big benefits for already deployed profiles.

> Now the question is: just like we pushed the grokXFN.xsl
> pointer into the profile document, can we somehow push the
> profile pointer down into the license link? [...]

> But... what trigger does a client use to extract data from such
> links? hmm...
> 
> def glean_links(doc):
>   statements = []
>   for elt in doc.link + doc.a:
>     if ':' in elt.rel: # KLUDGE to detect fully qualified rel names?

I may be a purist, but I really think that's too much of a kludge to be
acceptable... That looks to me as going against one of the TAG
principles:
"Specifications that use QNames to represent URI/local-name pairs SHOULD
NOT allow both forms in attribute values or element content where they
would be indistinguishable from URIs."
http://www.w3.org/TR/2003/WD-webarch-20031209/#qname-uri-syntax

Although rel is not specifically a QName in the HTML data model, it's a
simple URI which seems to have the same kind of problems as those
triggered by mixing QNames and URIs. (besides, the HTML WG proposal do
use qnames in rel IIRC).

> Is is it reasonable to look at any fully-qualified link relationship
> as a pointer to a data view transformation? That seems like a stretch.

Quite. Meanwhile, I don't have a good solution to this specific use
case.

Dom
-- 
Dominique Hazaël-Massieux - http://www.w3.org/People/Dom/
W3C/ERCIM
mailto:dom@w3.org