Re: on evaluation

Thanks Peter,

I tried to come up with a better wording as well but ended up deleting the
paragraph completely
https://github.com/w3c/data-shapes/commit/139f7de173b0ee6e054c4f250f33e18a6dac78fd#diff-69303a57193e6c2d7327c8de0fc977caL1069

We already have a definition of how filters & validation work in section 4
which, IIRC came from you:
A node validates against a shape iff either it does not validate against
some filter of the shape or none of the constraints in the shape produce a
validation result or a failure for the node.

This definition also covers the filter ordering as well as how node failure
is handled.
I am not sure if there is a need to handle other kinds of failure that may
arise from the different order of evaluation, even though these cases are
probably not common.

Best,
Dimitris

On Thu, Sep 29, 2016 at 1:09 AM, Peter F. Patel-Schneider <
pfpschneider@gmail.com> wrote:

> So, yes, I have two problems with the current wording:
>
> 1/ It is yet another case of loose terminology.
> 2/ It is the wrong way to do things.
>
> There are two reasons that it is wrong:
> a) It prevents optimization of the SPARQL processing.
> b) It even prevents translation of entire SHACL shapes to single SPARQL
> queries, *requiring* something like sh:hasShape in *every* SHACL
> implementation.
>
> I think that the order in which I discovered these issues was 2a, then 1,
> then
> 2b.
>
> The net result is that I believe that the entire sentence needs to be
> completely scrapped, and thus I don't see that there is any utility in
> wordsmithing the sentence to fix problem 1.
>
> Peter F. Patel-Schneider
> Nuance Communications
>
>
> On 09/28/2016 02:41 PM, Karen Coyle wrote:
> > Oh, I should have included the link to the diagram:
> >
> > https://www.w3.org/TR/2016/WD-shacl-20160814/images/SHACL-
> Validation-Process.png
> >
> > That's a quick read. - kc
> >
> > On 9/28/16 2:38 PM, Karen Coyle wrote:
> >>
> >>
> >> On 9/28/16 1:56 PM, Peter F. Patel-Schneider wrote:
> >>> My view is that there should be no requirement that focus nodes are
> >>> validated
> >>> against the filter shapes of a shape before they are validated against
> >>> the
> >>> constraints of the shape.  I have previously mentioned several reasons
> >>> that
> >>> led me to this view.
> >>
> >> OK. This is a different objection to what I had previously understood. I
> >> thought your objection was to saying that they are "validated".
> >>
> >> In the introduction to section 2.0 of SHACL there is a diagram that
> >> shows what I have taken to be an execution flow:
> >>
> >> data graph -> targets are used to select focus nodes -> Filters are used
> >> to eliminate some focus nodes -> Constraints are used to produce
> >> validation results.
> >>
> >> Is it this flow that you are objecting to?
> >>
> >> If so, can you either point to or reiterate your reasons?
> >>
> >> Thanks,
> >> kc
> >>
> >>
> >>>
> >>> Peter F. Patel-Schneider
> >>> Nuance Communications
> >>>
> >>>
> >>> On 09/28/2016 06:40 AM, Karen Coyle wrote:
> >>>> Peter, how would you describe the action that takes place then?
> >>>> Because there
> >>>> is an implied action and work flow. Implemented? executed? enforced?
> >>>> resolved?
> >>>> (I'm running through the thesaurus entries.)
> >>>>
> >>>> kc
> >>>>
> >>>> On 9/27/16 3:43 PM, Peter F. Patel-Schneider wrote:
> >>>>> For starters, filter shapes are not "applied to the data graph", so,
> >>>>> no.
> >>>>>
> >>>>> peter
> >>>>>
> >>>>>
> >>>>> On 09/27/2016 03:38 PM, Holger Knublauch wrote:
> >>>>>> Peter, would this suggestion work better for you? I assume someone
> >>>>>> could argue
> >>>>>> that "applied" is not defined.
> >>>>>>
> >>>>>> Holger
> >>>>>>
> >>>>>>
> >>>>>>
> >>>>>> On 28/09/2016 1:44, Karen Coyle wrote:
> >>>>>>> Filter shapes must be ...
> >>>>>>>  - applied to the data graph
> >>>>>>> ... before validating....
> >>>>>>>
> >>>>>>>
> >>>>>>> On 9/26/16 4:38 PM, Holger Knublauch wrote:
> >>>>>>>> Does anyone else find this sentence unclear? If yes, could someone
> >>>>>>>> suggest alternative wording?
> >>>>>>>>
> >>>>>>>> "Filter shapes MUST be validated before validating the associated
> >>>>>>>> shapes
> >>>>>>>> or constraints."
> >>>>>>>>
> >>>>>>>> Thanks,
> >>>>>>>> Holger
> >>>>>>>>
> >>>>>>>>
> >>>>>>>> On 27/09/2016 1:52, Peter F. Patel-Schneider wrote:
> >>>>>>>>> How is a shape "validated"?
> >>>>>>>>>
> >>>>>>>>> This problem even affects the second half of the sentence I
> >>>>>>>>> initially
> >>>>>>>>> quoted.
> >>>>>>>>>
> >>>>>>>>> peter
> >>>>>>>>>
> >>>>>>>>> On 09/26/2016 12:18 AM, Holger Knublauch wrote:
> >>>>>>>>>> Replaced with "validated":
> >>>>>>>>>>
> >>>>>>>>>> https://github.com/w3c/data-shapes/commit/
> d4fbdebd7044cd79f35985a75a54994ea3facde9
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>> Holger
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>>> On 26/09/2016 15:59, Peter F. Patel-Schneider wrote:
> >>>>>>>>>>> "Filter shapes MUST be evaluated before validating the
> associated
> >>>>>>>>>>> shapes or
> >>>>>>>>>>> constraints."
> >>>>>>>>>>>
> >>>>>>>>>>> Evaluation is not defined.
> >>>>>>>>>>>
> >>>>>>>>>>>
> >>>>>>>>>>> peter
> >>>>>>>>>>>
> >>>>>>>>>>
> >>>>>>>>
> >>>>>>>>
> >>>>>>>>
> >>>>>>>
> >>>>>>
> >>>>>>
> >>>>>
> >>>>>
> >>>>
> >>>
> >>
> >
>
>


-- 
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