W3C home > Mailing lists > Public > public-rdf-dawg@w3.org > January to March 2011

Re: Review: SPARQL 1.1 Federated Extensions

From: Axel Polleres <axel.polleres@deri.org>
Date: Tue, 1 Mar 2011 10:31:26 +0000
Cc: "Lee Feigenbaum" <lee@thefigtrees.net>, "SPARQL Working Group" <public-rdf-dawg@w3.org>
Message-Id: <307E5D2F-43E0-4721-B01E-B5BB95D3A076@deri.org>
To: "Carlos Buil Aranda" <cbuil@fi.upm.es>
Similar to Lee, the definition in 3.2 honestly doesn't make much sense to me.
What is ep(i) ?

you define eval(D(G), ep(i)) as D[i]

ep(i) is not defined, but I assume it shall return a graph?
but then in the next definition you call

 eval(D(ep(i)), P_1)

either there's some overloading on the function ep() ongoing here, or I don't understand what's actually going on here. Then you write "if i in dom(ep)" where dom(ep) is not defined.

For now, I am more leaning towards dropping 3.2 and getting 3.1 straight, following the comments we gave.

best,
Axel

On 1 Mar 2011, at 10:12, Carlos Buil Aranda wrote:

>   On 28/02/2011 20:44, Lee Feigenbaum wrote:
> > My original comment was asking how 4.1 and 4.3 are related. I'm still
> > not sure I understand; you seem to say that they are alternatives for
> > one another, but I don't see that.
> >
> > 4.1 deals with the translation from grammar to algebra and then
> > evaluation of algebra. I'm not sure what 4.3 is saying; for instance,
> > it uses ep(i) and then defines it, but I can't seem to make sense of
> > the use and definition of it.
> >
> > Also, whereas the evaluation algorithm in 4.1 (3.1 now) talks about
> > invoking the SPARQL protocol (which makes sense to me), the definition
> > in 4.3 (3.2 now) doesn't seem to say anything about how a pattern gets
> > evaluated against a remote endpoint. So as it stands now, if they
> > _are_ alternatives, I'd prefer to remove 3.2 and stick with 3.1, while
> > cleaning up 3.1 as per my comments in the original review.
> 3.1 and 3.2 are alternatives. What I say in 3.2 is that a pattern P is
> evaluated in the default graph. If that pattern P is of the form SERVICE
> i {P1} where i is a URI, then P1 is evaluated in the graph that is
> pointed by the function ep(i). What ep(i) does is to get the graph
> pointed by i. So in the end, the evaluation of SERVICE is the evaluation
> of its inner pattern in a graph pointed by i, like in the query
> document, but changing the default graph.
> 
> I do not know if explained myself? these are two proposals, in the end I
> will leave only one of them in final document.
> 
> Carlos
> 
Received on Tuesday, 1 March 2011 10:48:24 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 16:15:45 GMT