Re: rdf semantics test suite setup

On Sep 26, 2013, at 4:09 AM, Sandro Hawke <sandro@w3.org> wrote:

> The 2004 test suite is described and enumerated here: http://www.w3.org/TR/rdf-testcases/#testcases
> 
> I've copied the tests themselves (the contents of the last test zipfile generated) to mercurial, at https://dvcs.w3.org/hg/rdf/file/default/rdf-mt/tests/2004-test-suite

I created a script to generate a 2013-formatted manifest from the Positive and NegativeEntailmentTests in the 2004-test-suite and copy the referenced files into https://dvcs.w3.org/hg/rdf/file/default/rdf-mt/tests, including transforming those in RDF/XML format to Turtle.

An example entry is the following:

<#datatypes-intensional-xsd-integer-string-incompatible> a mf:PositiveEntailmentTest;
  mf:name "datatypes-intensional-xsd-integer-string-incompatible";
  rdfs:comment """
    The claim that xsd:integer is a subClassOF xsd:string is
    incompatible with using the intensional semantics for
    datatypes.
  """;
  rdfs:approval rdft:Approved;
  mf:action <datatypes-intensional/test002.nt>;
  mf:result false .

Note that it uses mf:action and mf:result in place of test:premiseDocument and test:conclusionDocument. We could just as easily use different properties to define these, but it didn't seem to me to be worth chaning the basing test format for. Note that either mf:action or mf:result may be false, which is equivalent to test:False-Document in the 2004 manifest.

I also did not copy test:entailmentRules or test:datatypeSupport, as they seem to be purely infomational. Again, we could mint new predicates in the mf: namespace and transform those as well.

For the test sources formerly in RDF/XML, I attempted to preserve the original comments in the Turtle output too.

Also, note that some files, for example <https://dvcs.w3.org/hg/rdf/file/default/rdf-mt/tests/rdfs-domain-and-range/premises005.ttl>, use IRIs from the former test suite; I believe they could be IRIs in the new location, but there doesn't seem to be a compelling reason to change them.

I have not yet attempted to run the tests myself, at this point.

We also need to create a Wiki entry describing how to run the tests, similar to what we've done for Turtle, and probably update the manifest vocabulary for mf:PositiveEntailmentTest and mf:NegativeEntailmentTest.

Gregg

> I hope that's what you wanted.       (cf ACTION-298)
> 
>        -- Sandro
> 
>

Received on Sunday, 29 September 2013 01:09:26 UTC