Re: Namespace document issues

On Wed, 2006-09-06 at 18:51 +0200, Danny Ayers wrote:
> On 9/6/06, Dan Connolly <connolly@w3.org> wrote:
> > On Wed, 2006-09-06 at 10:50 +0200, Danny Ayers wrote:
> > [...]
> > > There's also  a potential conflict situation. I suspect we need a rule
> > > that says that if a transformation has been explicitly stated in the
> > > instance document, it SHOULD be applied, and that the transformation
> > > provided in the namespace doc SHOULD not (unless that too is referred
> > > to explicitly in the instance doc).
> >
> > Umm... why not? I don't see a conflict. If there are multiple
> > applicable transformations, there are multiple applicable
> > transformations.
> 
> There are transformations that may be applicable in certain contexts,
> but not in all contexts, the publisher may want one and not another.
> 
> A specific case is A) Atom/AtomOWL and B) Atom/RSS 1.0 -  if you want
> to be able to roundtrip in an RDF system, A) should work, B) might not
> (thanks to multiple values for a given property on an item).
> 
> Another scenario might be where the potential GRDDL Results Graph for
> a given GRDDL Source Document is OWL Full according to one
> transformation, OWL DL according to another.
> 
> What I had in mind with the "SHOULD"s above was for the namespace doc
> to provide a default transformation, that could be overridden/switched
> off if the publisher preferred.

I see what you mean, now.

That smells non-monotonic, to me. I'm inclined to not go there at all.
I think this would be sufficiently incompatible with the deployed
GRDDL stuff that we should change the GRDDL namespace name if
we were to introduce this feature.

I'll stand by for input on whether to add this to the issues list.
It's clearly a separable issue; what's not clear is whether it's
worth putting a WG decision about it (even to postpone/decline it)
on the critical path.

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

Received on Wednesday, 6 September 2006 17:25:12 UTC