- From: Axel Polleres <axel.polleres@deri.org>
- Date: Fri, 21 Jan 2011 08:55:40 +0000
- To: Andy Seaborne <andy.seaborne@epimorphics.com>
- Cc: "Carlos Buil Aranda" <cbuil@fi.upm.es>, "Sandro Hawke" <sandro@w3.org>, "SPARQL Working Group" <public-rdf-dawg@w3.org>
On 20 Jan 2011, at 21:55, Andy Seaborne wrote:
>
>
> On 20/01/11 17:05, Axel Polleres wrote:
> > Just to make clear, what I suggested is to not necessarily have a full SPARQL endpoint up, but just some pre-canned responses to
> > precanned HTTP requests for the test cases. (that would be even simpler than guaranteeing to maintain and run a fully-fledged
> > SPARQL endpoint that can be used for evaluating the test cases on one of W3C's servers)
> >
> > Opinions?
>
> Probably a good way to do it although I've been running a public SPARQL
> endpoint with a very small amount of data. It runs an EC2 small
> instance so it's hardly a lot of hardware.
>
> One thing to watch - there is no guarantee exactly what remote queries
> are made unless the test is carefully written to have only one possible
> (reasonable) execution strategy.
yes, you are right, thanks for pointing that out. I hadn't yet thoroughly thought about that. Will need to think through some test case first... .
Axel
>
> There's no guarantees how BINDINGS might be used.
>
> Andy
>
> >
> > Axel
> >
> >
> > On 20 Jan 2011, at 17:01, Carlos Buil Aranda wrote:
> >
> >> For me that works, I already have some test endpoints with dummy data
> >> for these purposes. It would be easy to place that data in SPARQL endpoints.
> >>
> >> Carlos
> >>
> >> On 20/01/2011 10:48, Axel Polleres wrote:
> >>> Hi Carlos, Sandro,
> >>>
> >>> I am a bit lost with that...
> >>> I remember we had some discussion internally about that, e.g. providing a dedicated "endpoint" responding to specific queries in a certain way...
> >>> That could be done by providing some hard-coded "dummy" endpoint that responds to a couple of specific HTTP encoded SPARQL queries which we use in the test cases with the respective SPARQL-result.
> >>>
> >>> If we had that, I believe, we wouldn't need to specify any extra vocabulary for the test cases for fed query, the only thing is to get some URIs like e.g.
> >>>
> >>> http://www.w3.org/2009/sparql/docs/tests/endpoint1?query=SELECT+*+WHERE+{%3FS+%3FP+%3FO+}
> >>>
> >>> to serve some specified results. Would that do?
> >>>
> >>> Axel
> >>
> >>
> >
> >
>
Received on Friday, 21 January 2011 08:56:14 UTC