Hi Karen, all,
This looks like a lot related to optimization of constraint checks. I'm not sure we have identified anything like this. As far as I can remember we have never discussed possible requirements on the order in which different constraints could be checked on a dataset. Have we?
We have discussed requirements to express order of statements in descriptions, and such requirement could help define a 'user perspective'. But as we have not kept this in our requirements, I guess we really have not much to say.
If someone in the group has a position, it would be great that we consider it. But given the tight timing (and the holiday period), it will be more likely a personal opinion than 'official' group feedback.
Conversely, these non-committal statements of mine may well form the base for an official 'no, our group doesn't have any user requirement on this'. What do you think?
Cheers,
Antoine
PS: Karen thanks a lot for this liaison work. It's a very good sign to see that the W3C group asks for feedback from us!
On 8/1/15 5:37 PM, Karen Coyle wrote:
> Next Thursday's meeting of the W3C group is going to attempt to resolve Issue #76 [1] regarding the treatment of AND and OR in SHACL. Here's a brief statement of the question:
>
> "The Data Shapes WG is currently discussing whether the operands of AND and OR constraints shall be evaluated in order (from left to right) or in an undefined order. They would appreciate input on what would match the users' expectations better. (In most programming languages the ordering matters and can be used to take control of performance, while in query languages such as SPARQL the order does not matter)."
>
> Since this is more of a user question than a SHACL engine question, we've been asked to advise the group on our preference.
>
> kc
> [1] http://www.w3.org/2014/data-shapes/track/issues/76
|