Re: ISSUE-51: Generalized sh:severity

On 8/25/15 10:29 PM, Simon Steyskal wrote:
> Hi!
>
>> I am concerned that a lack of result could be an
>> indication of a bug, and therefore the return of no result would mask
>> that.
>
> I'm not sure what kind of "bug" you mean.
> 1) a bug in terms of an incorrect shape definition (wrt. the official
> SHACL syntax)
> 2) a bug in terms of an insufficient shape definition (e.g. sh:maxCount
> restriction is missing causing the validation not to fail although it
> should)

Any/all of the above. The key statement is "causing validation not to 
fail although it should."

>
> Either way, 1) should be caught by the respective validation engine that
> implements SHACL and regarding 2) it wouldn't make any difference if the
> engine returns true instead of nothing (imho). In both cases I only know
> that no SHACL constraint failed, thus all individuals conform to their
> respective shapes.

I guess I'm just more skeptical than you are. - kc

>
> cheers,
> simon
>
> ---
> DDipl.-Ing. Simon Steyskal
> Institute for Information Business, WU Vienna
>
> www: http://www.steyskal.info/  twitter: @simonsteys
>
> Am 2015-08-26 00:57, schrieb Karen Coyle:
>> (moving my comments to this thread)
>>
>> I'm a bit uneasy about the fact that the validation vocabulary returns
>> only "negative" results (e.g. violations). Presumably that means that
>> no result = true/aok. Would it be possible to return a positive value
>> for "true"? I am concerned that a lack of result could be an
>> indication of a bug, and therefore the return of no result would mask
>> that.
>>
>> kc
>>
>> On 8/6/15 4:24 PM, Holger Knublauch wrote:
>>> In the call today, I was asked to clarify how the severity of constraint
>>> violations can be specified. Several WG members also voiced their
>>> support for being able to specify the severity for each occurrence of a
>>> template, which was not supported until today.
>>>
>>> Based on this preference, I have made a small generalization to the
>>> handling of sh:severity and will describe how it works below. I have
>>> made this change directly on our master copy as it seems a fairly
>>> straight-forward and hopefully uncontroversial change. I am holding off
>>> with the other changes until we had another meeting about this.
>>>
>>> To get started, please read the new paragraph
>>>
>>>      http://w3c.github.io/data-shapes/shacl/#severity
>>>
>>> especially Example 31 (Declaring the Severity using sh:severity)
>>>
>>> This shows that there is now a way to specify the severity for each
>>> property occurrence (hopefully addressing Eric's point today). Each of
>>> these is a template call, instantiating sh:PropertyConstraint. I have
>>> moved the property sh:severity into the sh:Constraint class, which is a
>>> superclass of sh:PropertyConstraint. If left unspecified, it will use
>>> the severity declared at the template itself (i.e. with
>>> sh:AbstractCountPropertyConstraint as its subject). If even this is left
>>> unspecified, then it will apply sh:Error as a default.
>>>
>>> For native constraints (in SPARQL) the situation is unchanged, e.g.
>>>
>>> ex:MyShape
>>>      a sh:Shape ;
>>>      sh:constraint [
>>>          sh:sparql "..." ;
>>>          sh:severity sh:Warning ;
>>>      ] .
>>>
>>> will always produce a warning.
>>>
>>> I have also updated the Turtle file and changed the prose in each
>>> textual definition to say "violation" instead of "sh:Error". I did not
>>> yet update the shacl-ref file.
>>>
>>> I would appreciate a second pair of eyes to verify that I didn't miss
>>> anything in this refactoring.
>>>
>>> Regards,
>>> Holger
>>>
>>>
>>>
>

-- 
Karen Coyle
kcoyle@kcoyle.net http://kcoyle.net
m: 1-510-435-8234
skype: kcoylenet/+1-510-984-3600

Received on Wednesday, 26 August 2015 13:55:49 UTC