- From: Andy Seaborne <andy.seaborne@epimorphics.com>
- Date: Wed, 20 Jul 2011 20:02:13 +0100
- To: public-rdf-dawg@w3.org
On 20/07/11 03:53, Carlos Buil Aranda wrote: > any more opinions on that? Another option is to leave the possibility open - slightly different from Lee's making it an error. Or an optional feature of an optional feature. It's a constraint - just like any other constraint in the language. For most of them, we define via the algebra what the constraint means and we get exact answers for any non-fed query. If we think of ?g ranging over all possible SPARQL endpoints (a theoretical device - bottom up evaluation - and say that an engine executes with a subset of possibilities, and of course that subset can be constrained by a join relationship. Andy > > thanks, > > Carlos > > 2011/7/19 Gregory Williams <greg@evilfunhouse.com > <mailto:greg@evilfunhouse.com>> > > On Jul 19, 2011, at 5:02 PM, Carlos Buil Aranda wrote: > > > just summarizing a bit, the main concern is the use of variables > in SERVICE. The problem relies in the specification of the semantics > and the boundedness condition which suggests an order for executing > the query. The problem is first, the way of defining the execution > of SERVICE VAR, which using the current join semantics is wrong > because it is not possible to evaluate Join(G, Service(VAR, G, > Transform(P), SilentOp)) because VAR still hasn't a value. Following > this, the boundedness condition, which has to be completed for all > the SPARQL 1.1 operators. Am I right? > > Thanks Carlos. Yes, I think this is a good summary of the biggest > issues. > > > So, the possible solutions are: > > - For the SERVICE VAR semantics > > - add a new operation that could allow the evaluation > (operation which wouldn't be bottom-up) > > - define its whole semantics in a new way > > - For the boundedness restriction > > - specify all cases: it may take a bit long > > - remove it? I do not think this is a good idea, how do we > specify then that a variable is bound (which is needed for the > evaluation semantics of SERVICE VAR)? > > It might be needed for the evaluation semantics, but I don't think > the current text connects the two. There is no mention of strongly > bounded variables or service safeness in the "Definition: Evaluation > of a Service Pattern" section. > > thanks, > .greg > > >
Received on Wednesday, 20 July 2011 19:02:47 UTC