Re: shapes-ISSUE-89 (recursion without properties): How should recursion that does not involve a property be handled? [SHACL Spec]

Peter,

I agree that ex:s1 is not useful. My initial reaction is that it would
always be true since there is no way to generate violations. This
interpretation is consistent with the article [1] I wrote about the
interpretation of positive recursion using sh:valueShape (I have an
ACTION to write a proposal) . SHACL is in effect based the presumption
of innocence, i.e. "innocent until proven guilty".

The case of no properties may reduce to the case where properties are
present if we imagine that all nodes have a virtual "sh:self" property
that loops to itself (X sh:self X) and the above is like asserting
that the property sh:self has a sh:valueShape of ex:s1.

[1] http://arxiv.org/abs/1505.04972

-- Arthur

On Tue, Sep 15, 2015 at 5:32 PM, RDF Data Shapes Working Group Issue
Tracker <sysbot+tracker@w3.org> wrote:
> shapes-ISSUE-89 (recursion without properties): How should recursion that does not involve a property be handled? [SHACL Spec]
>
> http://www.w3.org/2014/data-shapes/track/issues/89
>
> Raised by: Peter Patel-Schneider
> On product: SHACL Spec
>
> Right now
>
> ex:s1 rdf:type sh:shape ;
>       sh:constraint [ a sh:AndConstraint ;
>                       sh:shapes ( ex:s1 ) ] .
>
> is valid SHACL.  However, it is not a very useful shape.
>
> In general, recursion between shapes where the recursion does not involve a property is not useful.
>
> Should such shapes be allowed in SHACL?
>
>
>

Received on Friday, 25 September 2015 20:26:06 UTC