Re: RDF:type vs. instanceOf, nextVal

Axel Hecht wrote:
> Hi,
> as indicated already on my blog, I'm going to pick up bugs 
> https://bugzilla.mozilla.org/show_bug.cgi?id=90566, 
> https://bugzilla.mozilla.org/show_bug.cgi?id=98392, and 
> https://bugzilla.mozilla.org/show_bug.cgi?id=110480. I would like to 
> know how many extensions or other parts of our code exlicitly rely on 
> these. I don't think that I can actually create something that is 
> reliably backwards compatible, though I could add some fakage into 
> nsInMemoryDataSource, which may cover quite a few szenarios. I'd really 
> prefer not to, though.

I have a pretty interesting idea to work around the instanceOf vs type 
problem, I could actually make the RDF service return rdf:type if 
queried for the instanceOf resource.

That would be backwards compatible my most accounts, though it would be 
quite a hack, in the sense that I return a resource that is semantically 
the same, but I would break folks doing assertions with rdf:type or 
rdf:instanceOf as source. The use cases that I saw on #swig (semweb 
interest group) didn't really show a difference to instanceOf, and 
they're really semwebby, so they likely don't matter to us.

I'm still not sure if that's a good thing to do.

Note that I don't intend to introduce this hack to nextValue, as that is 
not only changing the resource name, but also the type of the target, 
thus not being sematically indentical.

CC'ing staff for some project management idea.

Axel