Re: ISSUE-64: Spec updated to include sh:minInclusive etc.

On 7/31/2015 23:01, Dimitris Kontokostas wrote:
>
>
> On Fri, Jul 31, 2015 at 1:29 AM, Holger Knublauch 
> <holger@topquadrant.com <mailto:holger@topquadrant.com>> wrote:
>
>     On 7/30/2015 23:46, Dimitris Kontokostas wrote:
>>
>>
>>     On Fri, Jul 24, 2015 at 6:51 AM, Holger Knublauch
>>     <holger@topquadrant.com <mailto:holger@topquadrant.com>> wrote:
>>
>>         As resolved in the call today, SHACL now supports
>>
>>         - sh:maxExclusive
>>         - sh:maxInclusive
>>         - sh:minExclusive
>>         - sh:minInclusive
>>         (hopefully not including too many copy and paste errors)
>>
>>         - sh:maxLength
>>         - sh:minLength
>>         - sh:pattern
>>
>>         I have implemented the latter three so that they work with
>>         any datatype and even IRIs. 
>>
>>
>>         Errors are reported on blank node values, because they cannot
>>         be turned into strings. Errors are also reported if the <, >
>>         comparisons fail.
>>
>>
>>     I have some minor concerns on this as it might not be consistent
>>     with other cases where a comparison fails and a user would expect
>>     an error report as well.
>
>     Do you have a specific example for the latter? SPARQL usually
>     "silently" fails on function evaluation errors, i.e. it returns no
>     matches in such SELECT queries. Would your preference be that we
>     do the same and not return an error when a > comparison failed,
>     i.e. returned unbound?
>
>
> I do not have a strong preference on this, I suggest that if we go 
> with your approach we should make a note that this is not the expected 
> SPARQL behavior

Ok that's fine. I hope this paragraph clarifies it:

https://github.com/w3c/data-shapes/commit/005604e8aa0b6c6843bd1525dbe260a4ed30d7ab

Holger

Received on Sunday, 2 August 2015 23:01:16 UTC