- From: Seaborne, Andy <andy.seaborne@hp.com>
- Date: Tue, 20 Jul 2004 19:41:39 +0100
- To: Steve Harris <S.W.Harris@ecs.soton.ac.uk>
- Cc: 'RDF Data Access Working Group' <public-rdf-dawg@w3.org>
> SELECT ?s, ?src > WHERE (?s ?p ?o) SOURCE ?src > > +-------+-------+ > | ?s | ?src | > +-------+-------+ > | :foo | a.rdf | > | :bar | b.rdf | > +-------+-------+ I would have expected +-------+-------+ | ?s | ?src | +-------+-------+ | :foo | a.rdf | | :foo | a.rdf | | :bar | b.rdf | +-------+-------+ without a DISTINCT being added to the query. Andy Steve Harris wrote: > On Mon, Jul 19, 2004 at 05:21:29PM -0500, Dan Connolly wrote: > >>Perhaps the objective is fuzzy enough so that the conflict isn't >>clear, but all the concrete designs I have seen >>(e.g. the BRQL SOURCE mechanism) involve the input to the QUERY >>being more than just an RDF graph. > > > I understand it to be (opionally) a set of RDF grpahs. I would like it if > the intention is that servers that do not need/use source information can > ignore it, eg. treat all triples as having a single source. > > >>The spec doesn't yet explicitly say what the input to a query >>is, but it seems to be more than just a graph: >> >>"The SOURCE clause added to a triple pattern causes the variable or >>literal to match or bind the RDF graph which the statement came from, if >>the server knows." >> -- http://www.w3.org/2004/07/08-BRQL/ >> $Revision: 1.14 $ of $Date: 2004/07/19 13:11:51 $ >> >>Does anybody have any test cases for SOURCE? I'm sure those would >>make this crystal clear. > > > Some trivial ones, BRQLised: > > Data > > a.rdf: > :foo :a "a" > :foo :b "b" > > b.rdf: > :bar :a "a" > > > SELECT ?s > WHERE (?s ?p ?o) SOURCE <a.rdf> > > +-------+ > | ?s | > +-------+ > | :foo | > +-------+ > > > SELECT ?s, ?src > WHERE (?s ?p ?o) SOURCE ?src > > +-------+-------+ > | ?s | ?src | > +-------+-------+ > | :foo | a.rdf | > | :bar | b.rdf | > +-------+-------+ > > > CONSTRUCT * WHERE (?s ?p ?o) SOURCE <a.rdf> > > should return an RDF graph equivalent to the orignal asserted a.rdf I > think. I dont have a real test for this, but it seems very useful so I > included it anyway. > > +---------------+ > | * | > +---------------+ > | :foo :a "a" | > | :foo :b "b" | > +---------------+ > > - Steve
Received on Tuesday, 20 July 2004 14:42:02 UTC