James M Snell wrote:
It's definitely better than bastardizing the scheme element to provide a dereferenceable URL. I would say that this definitely falls within the realm of the validator's "questionable use" warning... at least until there's spec text that describes the semantics.
Atom has no semantics, and that's a conscious decision of the WG, going back as far as rejecting semweb technology. We also organise out markup fairly differently to the RDF guys - they explicitly hang everything off a URI; we don't and because of that this kind of issue will come up every time some wamts to take an Atom construct and make a resource out of it (root problem - we have links /and/ ids). The validator then, and imvho, has no business nosing around denotational semantics beyond saying "no notation with denotation!" when it sees an atom:id in a new place.
I think what's needed here is a) a definition of atom:id wrt atom:category, b) a uniform means of categorising entries.
Bill
