Re: rdf in atom

[David Powell <djpowell@xxxxxxxxxxxx>]

> http://xml.coverpages.org/draft-nottingham-atomtriples-00.txt

I like the idea of the option of using link constructs to specify the
subject of the triples, but as Henry said, using the entry id as the
default subject of triples conflicts at least with both Henry and my
Atom vocabularies.  We both use a, typically anonymous, resource to
represent an entry instance.

Adding indirection between the Entry resource named by the entry id,
and the representation delivered in the feed isn't always the right
thing to do, sometimes a simpler, more direct, model is more
appropriate.  Both models are valid, so I think that this draft
should support both.

I think both models should be accommodated by being less explicit
about what the subject URI should be for entry triples.  Then people
could use the draft with whatever Atom model is most appropriate.


Also, I think I agree with Bill regarding QNames in content.  And I'm
not clear how the QName should be processed.  Should the match be
between the value and the prefix + ":" + localname of the element, or
does the prefix need to be resolved into an ns-uri, and matched
against the (ns-uri, localname) of the element.


BTW:  Atom/RDF lives here: <http://djpowell.net/atomrdf/>, it is
pretty complete, but the task of making it releasable with tests etc
is happening at a glacial pace.  It has a GRDDL compatible XSLT,
which I would be happy to extend to support an Atom-in-RDF proposal
that proves to be successful.


-- 
Dave