Re: shapes-ISSUE-191 (Parameter value types): Should the value types of parameters be constraints [SHACL Spec]

On 19/10/2016 2:41, Karen Coyle wrote:
> Here are my comments:
>
> 1)
> 1761     The values of <code>sh:datatype</code> must be IRIs 
> representing datatypes, such as <code>xsd:string</code>
>
> If datatype were defined in the terminology section as RDF datatypes, 
> then this could be simply said as:
>
> "The values of sh:datatype are <link to terminology>datatypes</link...>"
>
> That would remove "representing" which is problematic, and would 
> define datatype, which is essential since it is used in the document.

I have added a definition of "datatype" to the terminology section.

>
> 2)
>
> 2945    <code>sh:qualifiedValueShape</code> must be accompanied by a 
> <code>sh:qualifiedMinCount</code> or a 
> <code>sh:qualifiedMaxCount</code>, or both.
>
> To eliminate "accompanied by", it could be stated as:
>
> "For each sh:qualifiedValueShape there must be either one 
> sh:qualifiedMinCount or one sh:qualifiedMaxCount, or one 
> sh:qualifiedMinCount and one sh:qualifiedMaxCount."
>
> alternate wording (hard to make this elegant)
>
> "For each sh:qualifiedValueShape there must be either one 
> sh:qualifiedMinCount or one sh:qualifiedMaxCount, or one of each."

Ok, I chose your last suggestion.

>
> 3)
> The statements nearly all use plurals where I could generally use the 
> singular, such as:
>
> 2597     The values of <code>sh:lessThanOrEquals</code> must be 
> <a>IRIs</a>
>
> This is probably a difference in mental models, but I think of a 
> property as singular in a triple with a single object. Perhaps 
> thinking of it more as a graph it can be a property with multiple 
> objects. I haven't found anything in other RDF standards that would 
> show a usage pattern. In any case, it probably matters more that the 
> document be consistent.

I checked that they all use plural, so unless I am missing something 
it's already consistent. I prefer plural over singular because there can 
be multiple values, and it sounds strange to speak about "the value of 
<code>...</code>" if no specific subject is given.

https://github.com/w3c/data-shapes/commit/eb2b55f88d9ef4c085a55a3b33b4e8f9c330d63e

Thanks
Holger



>
> kc
>
> On 10/17/16 9:07 PM, RDF Data Shapes Working Group Issue Tracker wrote:
>> shapes-ISSUE-191 (Parameter value types): Should the value types of 
>> parameters be constraints [SHACL Spec]
>>
>> http://www.w3.org/2014/data-shapes/track/issues/191
>>
>> Raised by: Holger Knublauch
>> On product: SHACL Spec
>>
>> In the currently published draft of the spec, each parameter of the 
>> core vocabulary is annotated with a column "Value type" that carries 
>> no meaning. Peter also stated that some of these value types are 
>> rather unhelpful:
>>
>> https://lists.w3.org/Archives/Public/public-rdf-shapes/2016Oct/0059.html
>>
>> I think we should do a proper job here and make the value types more 
>> useful, making SHACL more predictable. The Value types column should 
>> be deleted and instead the TEXTUAL DEFINITION of each component 
>> should enumerate constraints on these values. Shapes graphs that 
>> violate these constraints are invalid.
>>
>> I have made these changes to the draft and would like the WG to 
>> review them:
>>
>> https://github.com/w3c/data-shapes/commit/292f12936181ca2d3fd5c096a7880f2de6054f02 
>>
>>
>> My proposal is to approve these changes.
>>
>>
>>
>>
>

Received on Tuesday, 18 October 2016 23:35:17 UTC