[redland-dev] SPARQL DESCRIBE unimplemented/broken?

[Dutton, Jeff]

> > Is the SPARQL DESCRIBE verb unimplemented or broken?
> 
> It's not implemented.  What do you think it should do by 
> default when there is no application logic present?

When using the library directly (from a Perl script) I got a null
assertion error when doing a DESCRIBE (but not other verbs).  That
should have been enough info.  My problem was that I couldn't find
anywhere that said it wasn't implemented, and being new to SPARQL, I
couldn't convince myself that it wasn't me.

I did tip-toe into the Rasqal code, but saw defines for DESCRIBE and
assumed that it was supposed to be implemented.

> I was wondering about making it default to returning triples 
> query like:
>   CONSTRUCT { <uri> ?p ?o }
>   WHERE  { <uri> ?p ?o }
> 
> assuming that you had done a DESCRIBE <uri>

Without a bnode-closure, I guess this is what I'm going to do to
simulate DESCRIBE for unfamiliar subjects.  If DESCRIBE did the same
thing, then (+) at least I wouldn't have thought I was crazy :-) but (-)
SPARQL newbies (like myself) will probably think that is the spec'd
behavior of DESCRIBE.

> Other possibilities include returning the bnode-closure, but 
> that's writing a lot more code.

Is this on the roadmap?  I didn't see anything about DESCRIBE on the
To-Do.

Thanks for the response.  I've only been using Redland a week, but I was
able to get it working in no time, and I've found the source very
straightforward.  Great stuff!

Jeff