shapes-ISSUE-227 (SPARQL-based constraint components using EXISTS): Behavior of SPARQL-based constraint components that use EXISTS [SHACL - SPARQL]

shapes-ISSUE-227 (SPARQL-based constraint components using EXISTS): Behavior of SPARQL-based constraint components that use EXISTS [SHACL - SPARQL]

http://www.w3.org/2014/data-shapes/track/issues/227

Raised by: Irene Polikoff
On product: SHACL - SPARQL

A comment was received from Peter Patel-Schneider regarding the use of EXISTS in the SPARQL-based constraint components:

What is the behaviour of SPARQL-based constraint components that use EXISTS?

Is it the broken definition of EXISTS from the SPARQL document?  
Is it some particular fixed version of EXISTS?  
Is it undefined?  
Or is EXISTS not to be used in SPARQL-based constraint components?

When asked why the SHACL WG needed to address this question i.e., why it can't be left to SPARQL implementations, response was: 

As it stands right now, no implementation of SPARQL that I know of can be used
in the implementation of SPARQL-SHACL without significant modification because
implementations of SPARQL diverge from the definition of SPARQL.

Personally, I do not quite follow this logic. For example, it would seem to me to imply that no TR for SPARQL could exist because all implementations diverge from it. Nevertheless, I feel this comment needs to be discussed and the fact that it was discussed recorded.

If the WG decides we need to provide some information in the SHACL document to address this comment, I propose to say that SHACL spec does not fully define the behavior of SPARQL-based constraints that use EXIST and leaves this to implementations.

Received on Thursday, 9 February 2017 22:27:02 UTC