> It sounds like you're slipping from context sets into profiles.
That was my initial thought, but there's no way to say 'this conforms to X
profile' (outside of an extraData element).
But if a relation modifier (for example) simply requires extra data to be
returned for it to be useful, why does the client also have to request
that extra data separately?
If the server doesn't understand the relation modifier, then it will
error, unlike not understood extensions, so there's no problem there.
My example is somewhat contrived in that the extraData isn't actually
required for relevance ranking to be useful, and perhaps there's no case
when part of a query will require extra data, so I'm happy enough to just
leave it as it is, but wanted to at least raise the question.
> > I think that we should allow context sets to define semantics for
> > extraData in the response.
> My gut reaction: NO WAY!
> > If you send the relation modifier rel.CORI then you should be
> > prepared to accept the <rel:rank> <rel:score> and <rel:rawScore>
> > elements in extraTermData.
> Hmm. Interesting.
,'/:. Dr Robert Sanderson ([log in to unmask])
,'--/::(@)::. Special Collections and Archives, extension 3142
,'---/::::::::::. Nebmedes: http://nebmedes.o-r-g.org:8000/
I L L U M I N A T I