Re: Problem with property path test case pp15

> This needs fixing - literals should be included.
>
>     Andy
>
> On 08/04/11 08:23, Olivier Corby wrote:
>> I think there is a problem with pp15
>>
>> SELECT *
>> WHERE { ?X :p{0} "o" .
>> ?Y :p{0} :o .
>> :s :p{0} ?Z . }
>>
>> The query is applied to an empty graph and it returns:
>>
>> "o"
>> :o
>> :s
>>
>>
>> However, the current SPARQL 1.1 Query document specifies:
>>
>> Definition: ZeroLengthPath
>> A zero length path matches all subjects and objects in the graph, and
>> also any IRIs explictly given as endpoints of the path pattern.
>>
>>
>> But literal "o" is not an IRI and it is not in the graph, so the query
>> should fail.
>>
>>
>> In addition, what is the rationale for matching "IRIs explicitly given
>> as endpoints of the path pattern" ?
>
> Originally from:
>
> ?x rdf:type/rdfs:subClassOf* <T> .
> ---------
> project ?x
>   ?x rdf:type ?v . ?v rdfs:subClassOf* <T> .
>
> so if there is no
>
> <a> rdfs:subClassOf* <T> .
>
> this works when done in either order:
>
>   ?x rdf:type ?v . ?v rdfs:subClassOf* <T> .
> or
>    ?v rdfs:subClassOf* <T> . ?x rdf:type ?v .


  In the example above, ?x rdf:type<T>  must be in the graph
  and hence<T>  is in the graph.
  But my question was about URI that are*not present*  in the graph
  (like in pp15) : what is the design rationale for this exception
  to SPARQL graph pattern matching ?




>
>
> Olivier