Re: Tests I think are incorrect

 > From: Steve Harris <>
 > Date: 9 May 2006 11:03
 >
 > The following is a list of tests from my excuses file, was wondering if
 > anyone else agrees that they don't match the spec:
 >
 > Uses old syntax:

As part of preparing a publishable test suite, I presume that old syntax tests 
  don't go into it.

 >    http://www.w3.org/2001/sw/DataAccess/tests/#dawg-unsaid-001
 >    http://www.w3.org/2001/sw/DataAccess/tests/#dawg-unsaid-002
 >    http://www.w3.org/2001/sw/DataAccess/tests/#dawg-unsaid-003
 >    http://www.w3.org/2001/sw/DataAccess/tests/#dawg-source-simple-001
 >    http://www.w3.org/2001/sw/DataAccess/tests/#dawg-source-simple-002
 >    http://www.w3.org/2001/sw/DataAccess/tests/#dawg-source-simple-003
 >    http://www.w3.org/2001/sw/DataAccess/tests/#dawg-source-simple-004
 >    http://www.w3.org/2001/sw/DataAccess/tests/#dawg-source-simple-005
 >    http://www.w3.org/2001/sw/DataAccess/tests/#dawg-triple-
 > pattern-001-oldsyntax
 >    http://www.w3.org/2001/sw/DataAccess/tests/#dawg-triple-
 > pattern-002-oldsyntax
 >    http://www.w3.org/2001/sw/DataAccess/tests/#dawg-triple-
 > pattern-003-oldsyntax
 >    http://www.w3.org/2001/sw/DataAccess/tests/#dawg-triple-
 > pattern-004-oldsyntax
 >    http://www.w3.org/2001/sw/DataAccess/tests/#source-query-001
 >    http://www.w3.org/2001/sw/DataAccess/tests/#source-query-002
 >    http://www.w3.org/2001/sw/DataAccess/tests/#source-query-003
 >    http://www.w3.org/2001/sw/DataAccess/tests/#source-query-004
 >    http://www.w3.org/2001/sw/DataAccess/tests/#source-query-005
 >    http://www.w3.org/2001/sw/DataAccess/tests/#optional-and
 >    http://www.w3.org/2001/sw/DataAccess/tests/#optional-outer-and
 >    http://www.w3.org/2001/sw/DataAccess/tests/#optional-outer-and-
 > with-bound
 >
 > Spec gives different results:
 >    http://www.w3.org/2001/sw/DataAccess/tests/#extendedtype-ne-fail
 >    http://www.w3.org/2001/sw/DataAccess/tests/#extendedtype-literal-ne

These is bound up with what != means on unknown datatypes.  ARQ currently 
passes these tests because

"lex1"^^:unknown1 != "lex2"^^:unknown2 is false (i.e. is not positively known 
to be true)

but that is not what the spec says (!= is defined on RDF terms as 
not(RDF-term-equals))

I have an action to write some open world for = tests and so I will do some for !=

 >    http://www.w3.org/2001/sw/DataAccess/tests/#langmatches-3

It looks wrong:  I get:

---------------------
| p   | v           |
=====================
| :p5 | "abc"@fr    |
| :p4 | "abc"@en-gb |
| :p3 | "abc"@en    |
---------------------

(and the HTML has "abc" with no lang in all the answers).


 >    http://www.w3.org/2001/sw/DataAccess/tests/#langmatches-4

Looks wrong to me : I get:

---------------
| p   | v     |
===============
| :p1 | "abc" |
---------------

 >    http://www.w3.org/2001/sw/DataAccess/tests/#sorting-one-of-one-column

Query does not parse.
Then I get different answers because of duplicates

 >
 > Things I'm not quite sure about:
 >    http://www.w3.org/2001/sw/DataAccess/tests/#typepromotion-decimal-
 > decimal-pass
 >      Should promote to xsd:integer?

Yes, it should.

 >
 > I've omitted all the spec-* tests as loads of them are missing data or
 > broken in some way.
 >
 > If other people agree these are broken, perhaps we could mark them as
 > negative tests or remove them?

Or fix them?

	Andy

 >
 > - Steve

Received on Tuesday, 9 May 2006 16:38:10 UTC