- From: <jos.deroo@agfa.com>
- Date: Wed, 6 Oct 2004 01:51:53 +0200
- To: andy.seaborne@hp.com
- Cc: connolly@w3.org, RDF Data Access Working Group <public-rdf-dawg@w3.org>, public-rdf-dawg-request@w3.org
Andy, I was believing there was a solution when we automatically kept extra triples like <u1.n3> a q:Source. and rewrite WHERE SOURCE ?src (?x ?y ?z) as q:where {?S a q:Source. ?S.log:semantics log:includes {?X ?Y ?Z}}. but then we also get back solutions from those extra triples in the case of your first simple test case WHERE (?x ?y ?z) So we actually fall back to a multiquery rewrite such as in http://eulersharp.sourceforge.net/2004/04test/stc2a.n3 At this moment I'm not seeing any light in a recording of the container composition in triples in the query graph... -- Jos De Roo, AGFA http://www.agfa.com/w3c/jdroo/ "Seaborne, Andy" <andy.seaborne@hp.com> Sent by: public-rdf-dawg-request@w3.org 05/10/2004 19:42 Please respond to andy.seaborne To: Jos De_Roo/AMDUS/MOR/Agfa-NV/BE/BAYER@AGFA cc: connolly@w3.org, RDF Data Access Working Group <public-rdf-dawg@w3.org>, public-rdf-dawg-request@w3.org Subject: Variable in SOURCE (Was Re: A few simple tests) jos.deroo@agfa.com wrote: > DanC wrote: > >>On Tue, 2004-09-28 at 06:23, Seaborne, Andy wrote: >> >>>Attached is a revised set of tests that I have changed to be in be in >>>the current syntax. >> >>Has anybody else tried these? > > > Yes and we can only run query1 and query3a in AndyS original message > http://lists.w3.org/Archives/Public/public-rdf-dawg/2004JulSep/0541.html > > Among other things, our code can't handle unbound ?src in > ?src.log:semantics log:includes {...}. > (is how we actually try to do the SOURCE ?src (...)(...)) > > Jos, Thanks for pointing that out. I see now that equating "SOURCE ?src" with "?src.log:semantics log:includes" doesn't cover all cases as SOURCE operates in the query context and log:semantics operates on the whole web. For the case of "SOURCE <uri>" the restriction to a named web resource makes that OK; for a variable, it isn't. There is also the matter of the named containers also forming the RDF merge of the whole graph. What ways of approaching it in N3QL are there? It seems to me that the data needs to be merged into the main graph as well as being in the formula graphs. It's the relationship between the two that isn't quite log:semantics because it's scoped to the query context. That relationship isn't available to the query at all as it is not recorded in the outer merged RDF graph. Would something some way of recording the container composition in triples in the query graph be workable, using a predicate that is different from log:semantics? Andy
Received on Tuesday, 5 October 2004 23:52:36 UTC