Re: shapes-ISSUE-97 (sh:derivedValues): Proposal to define constraints using derived values [SHACL Spec]

On 10/3/2015 2:46, Peter F. Patel-Schneider wrote:
> What issue with the spec is this proposal addressing?

I would say derived properties are a new feature, so there is nothing 
wrong with the current spec. Derived properties do cover additional use 
cases though and will likely attract additional users to SHACL, for a 
relatively little "costs".

>
> Does this proposal add any expressive power to SHACL?

Yes. I do not know how else to express such relationships with the 
current mechanisms.

>    Is it necessary for any identified use cases?

Few things apart from the SPARQL extension mechanism are necessary for 
any identified use cases. But we have come across *many* use cases for 
derived properties in the last few years (also on paid-for projects), 
and even had a specific customer request for exactly what I am 
proposing. You can regard derived properties as a generalization of 
sh:EqualsConstraint, only that values can be computed and do not already 
have to be there. So this overlaps a lot with the original S27 
"Relationships between values of multiple properties". Even more 
relevant is Simon's story "S43: Using Property Paths for Property Value 
Comparison". Simon had created ISSUE-41 for that, which got closed 
because it was IMHO too ambitious and would have allowed computed values 
almost everywhere, but I do agree with the sentiment of his requirement. 
Finally, even for structural descriptions of forms (e.g. input forms and 
query builders) it is quite common to have read-only fields for things 
like "full name" that are derived from first name and last name resp.

>
>
>
> Only providing a diff to look at is not helpful.  Diff output from github is
> not easy to read, particularly when the source has deep indentation.

For your convenience I have copied a version of the branch into the 
master branch, from where you can read the added chapter at

http://w3c.github.io/data-shapes/shacl/index-ISSUE-97.html#derivedValues

HTH
Holger

Received on Tuesday, 6 October 2015 01:38:33 UTC