Re: first setup of reorganized tests checked in

Jeen Broekstra wrote on 01/08/2007 06:53:12 AM:

> In http://www.w3.org/2001/sw/DataAccess/tests/data-r2/
> 
> you will find:
> 
>  - manifest-evaluation.ttl:  a 'super-manifest' containing
>    references to all manifests with query evaluation test cases.
>  - manifest-syntax.ttl: a 'super-manifest' containing references
>    to all manifests with query syntax test cases.
> 
>  - syntax/ : a directory containing a manifest and a set of files
>    documenting both positive and negative syntax tests (these are
>    copies from /data/SyntaxFull/).
>  - triple-match/ : a directory containing a manifest and a set of
>    files documenting a few simple query evaluation tests (these are
>    copies from /data/simple/).
> 
> The manifests have been updated to use URIs for each test (instead of
> blank nodes). They currently do not yet contain EricP's annotations but
> I guess that can be easily amended.
> 
> Regarding the syntax tests: a few of these I have spotted to be out of
> data with respect to the current spec, specifically the following two:
> 
> http://www.w3.org/2001/sw/DataAccess/tests/data-
> r2/syntax/manifest#syntax-bnodes-03
> http://www.w3.org/2001/sw/DataAccess/tests/data-
> r2/syntax/manifest#syntax-bnodes-04
> 
> (which both deal with blank nodes in a predicate position). Instead of
> removing these tests I have relabeled them as negative syntax tests and
> have removed the 'dawg:Approved' annotation.
> 
> Before we go into (re)approving any of these tests (both the syntax and
> the evaluation tests) I would like a few others to eyeball them and
> filter out possible obsoletes. I have ran this set through our SPARQL
> engine and have not encountered any glaring errors, but...

I've updated my "test harness" (cobbled together collection of Ant tasks 
and various other bits) to handle the new manifest setup for the syntax 
tests and to handle both positive and negative syntax tests. With these 
updates, my engine passes all* of the syntax tests (positive and 
negative). 

It also passes all four of the triple-match tests.

Looks like a good start to me.

Lee

* Actually, my engine right now does function resolution while parsing, 
and so fails a few tests which include functions in ORDER BY conditions or 
FILTERs (but doesn't fail with parse errors). The queries themselves parse 
just fine -- my code is just being a bit overeager. So I counted them as 
passes.

> 
> Comments welcome.
> 
> Cheers,
> 
> Jeen
> 
> PS I also made a modest addition to the
> DataAccess/test/test-manifest.n3: I've added a QueryEvaluationTest 
class.
> -- 
> Dr. Jeen Broekstra                                          Den Dolech 2
> Information Systems Group                                        HG 7.76
> Department of Mathematics and Computer Science              P.O. Box 513
> Technische Universiteit Eindhoven                      5600 MB Eindhoven
> tel. +31 (0)40 247 36 86                                 The Netherlands
>

Received on Monday, 15 January 2007 00:41:46 UTC