W3C home > Mailing lists > Public > public-rdf-shapes@w3.org > September 2016

Re: on recursion in SHACL

From: Dimitris Kontokostas <kontokostas@informatik.uni-leipzig.de>
Date: Mon, 26 Sep 2016 12:17:56 +0300
Message-ID: <CA+u4+a3aZf2a3hBn-Gc1-aSR1AEj-9tTTK0NtanFNQbqMpLjWg@mail.gmail.com>
To: "Peter F. Patel-Schneider" <pfpschneider@gmail.com>
Cc: "public-rdf-sha." <public-rdf-shapes@w3.org>
Thanks for your feedback Peter,

I moved
<https://github.com/w3c/data-shapes/commit/f80725b162937683cc2f8aaecf497f2e0a3339c8>
the recursion definition to section 3.1.1 and adjusted the definitions a
bit.
I also made a request to revisit issue-22 since property paths already
provide an easy way to define recursion in SHACL
https://lists.w3.org/Archives/Public/public-data-shapes-wg/2016Sep/0054.html

Your comments for recursion on section 9.4 were addressed by Holger in a
separate email

Let me know if this resolves your issue
Dimitris

On Fri, Sep 23, 2016 at 8:00 PM, Peter F. Patel-Schneider <
pfpschneider@gmail.com> wrote:

> Following up one of the recent responses to my comments on Shapes
> Constraint
> Language (SHACL) lead me to look at how recursion works in Shapes
> Constraint
> Language (SHACL), W3C Editor's Draft 22 September 2016.
>
>
> Here is the initial discussion of recursion, from Section 4.8.1.
>
> "A shape may refer to itself directly or indirectly via sh:shape,
> sh:filterShape, etc. Such a shape is said to be recursive. The meaning of
> non-recursive shapes is always well-founded. In contrast, the meaning of a
> recursive shape may not be well-founded. The handling of recursive shapes
> in
> SHACL is left to implementations. Some implementations MAY reject shapes
> graphs containing recursive shape definitions. Some implementations MAY
> report a failure if a recursion has been detected at validation time."
>
> This is the wrong place for the initial discussion of recursion.  First,
> sh:filterShape was discussed much earlier, in Section 2.2.  Second, the
> discussion of recursion deserves not to be buried within the discussion of
> sh:shape.
>
> The definition of recursive shapes is much too sloppy.  What is covered by
> the "etc."?  Is
>   s:s1 rdf:type sh:Shape ;
>     rdfs:comment s:s1 .
> a recursive shape?
>
> What is the process for rejecting a shape graph containing recursive shape
> definitions?  The term "reject" occurs only in this one place.
>
> What does well-founded mean here?
>
> The meaning of "a recursion has been detected at validation time" has
> several problems.  Validation time is not defined.  What counts as
> detecting
> a recursion is not defined.
>
>
>
> Here is another aspect of recursion in SHACL, from Section 9.4.
>
> "Recursive use of functions is undefined: If a SPARQL-based function
> contains calls to other functions so that the same function with the same
> combination of parameters would be visited twice then the result of the
> function call is undefined. An implementation may either return no result
> (unbound) or terminate the surrounding SPARQL query with an error."
>
> It is not that all recursive use of functions is undefined.  What is
> undefined here by the more detailed description is a call to the same
> function and with the same parameters within another call.  It appears that
> this is an attempt to prevent infinite recursion.  Such calls, however,
> need
> not lead to infinite recursion if uncaught, even in a limited language like
> SPARQL.  Nor is it that all cases of infinite recursion involve calls of
> this sort.  As detecting such calls is neither necessary nor sufficient to
> prevent infinite recursion it is puzzling as to why a complex and
> potentially expensive mechanism is being described, and maybe even
> mandated.
>
>
> Peter F. Patel-Schneider
> Nuance Communications
>
>
>


-- 
Dimitris Kontokostas
Department of Computer Science, University of Leipzig & DBpedia Association
Projects: http://dbpedia.org, http://rdfunit.aksw.org,
http://aligned-project.eu
Homepage: http://aksw.org/DimitrisKontokostas
Research Group: AKSW/KILT http://aksw.org/Groups/KILT
Received on Monday, 26 September 2016 09:18:58 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:02:44 UTC