Re: comments on SHACL 3 March editors draft

I don't think that this solves the problem.

Spread around the document is wording like "first... then" or other ordering
constructs.  It may be that these ordering constructs change the result or it
may be that all they do is specify a particular sequence of execution.  The
problem with leaving them in without closely examining them is that it is very
hard to know which is which.  Instead of a statement saying that order only
matters when some little bit of the document has something where it does
matter, the document should be written in a way that order doesn't matter
unless it the ordering is necessary.

If the specification for SHACL was pure SPARQL then this would not matter so
much, but the specification for SHACL includes non-SPARQL stuff and some of
this non-SPARQL stuff looks very procedural.  Thus a lot of care needs to be
taken to ensure that no little thing deep down in the specification is some
place where order changes the results.

peter

PS:  I'm not sure which figure is in play here.  If it is the UML-like figure
then I don't see that it has any utility and it should go.


On 03/15/2016 10:36 AM, Dimitris Kontokostas wrote:
> Peter, can you check if this commit overcomes the ordering issue?
> https://github.com/w3c/data-shapes/commit/9edef4c82a7f6480d0c45e3e34656c7f93f1dfa5
> 
> Would you prefer to delete the figure completely? It that case I would delete
> that last two lines of the commit too
> 
> 
> On Mon, Mar 7, 2016 at 6:32 PM, Peter F. Patel-Schneider
> <pfpschneider@gmail.com <mailto:pfpschneider@gmail.com>> wrote:
> 
>     On 03/06/2016 08:46 PM, Holger Knublauch wrote:
>     [...]
> 
>     > On 7/03/2016 6:59, Peter F. Patel-Schneider wrote:
> 
>     [...]
> 
>     .2 Filters
> 
>     >> A SHACL processor might not begin by validating filters.  It might instead
>     >> look at all in-scope nodes and only later remove those that don't pass the
>     >> filters.  The document makes this illegal but it might be useful if the
>     >> filter shape is expensive to compute and few violations are expected.
>     >
>     > I believe this is already covered by stating that SHACL engines can produce
>     > additional validation results, see discussion above.
> 
>     This is not about validation results.  It is about order of execution.  The
>     document states:
> 
>     When a SHACL processor validates a focus node against a shape, it begins by
>     validating any filters associated with the shape via sh:filterShape.
> 
>     This is too procedural.
> 
>     peter
> 
> 
> 
> 
> 
> -- 
> Dimitris Kontokostas
> Department of Computer Science, University of Leipzig & DBpedia Association
> Projects: http://dbpedia.org, http://rdfunit.aksw.org,
> http://http://aligned-project.eu <http://aligned-project.eu/>
> Homepage:http://aksw.org/DimitrisKontokostas
> Research Group: AKSW/KILT http://aksw.org/Groups/KILT
>

Received on Tuesday, 15 March 2016 18:12:10 UTC