search

w3c / data-shapes

RDF Data Shapes WG repo
87 stars 33 forks source link

Issue with "strange path" tests #124

Open martinmaillard opened 4 years ago

martinmaillard commented 4 years ago

Hi,

I'm working on implementing the SHACL Test Suite for rdf-validate-shacl and am having issues with the two "strange path" tests.

The tests define the following expected resultPath: sh:resultPath ( ex:p ex:q ) ;

As far as I know, the spec doesn't specify how to handle these weird paths, so I don't understand why the extraneous inversePath is stripped out.

My ValidationReport graph contains sh:resultPath [ rdf:first ex:p ; rdf:rest [ rdf:first ex:q ; rdf:rest rdf:nil ] ; sh:inversePath ex:p ] ; (copied directly from the shape's sh:path). Shouldn't this be valid?

HolgerKnublauch commented 4 years ago

Hmm, I am puzzled right now. This was a test that was added rather late in the process, at a time where the WG was basically out of time. From my current understanding the paths are ill-formed because they are both lists and inverse paths yet all these path types must be disjoint. If my interpretation is correct then the test result should be undefined, and the tests not be in the test suite.

There was some discussion about this in https://www.w3.org/2017/05/03-shapes-minutes.html that lead to some editorial clarification. Does anyone else reading this ticket have comments?

Meanwhile I suggest ignoring this test.

martinmaillard commented 4 years ago

Ok, thanks for the feedback :)