Re: Properties v classes in validation

On 9/1/2015 12:51, Arnaud Le Hors wrote:
> Holger Knublauch <holger@topquadrant.com> wrote on 08/31/2015 05:26:24 PM:
> > ...
> > > And this has to be set as a Graph-level/global constraint, right?
> >
> > Yes, if you prefer to call it as such, this is graph-level.
>
> This isn't just a matter of what I prefer. That's what the spec call 
> it as far as I understand:
>
> *7.1 Graph-level ("global") Constraints*
> https://w3c.github.io/data-shapes/shacl/#graph-constraints
>
> Isn't that what we are talking about?

Yes it is. Maybe - since this topic is quite common - should we split 
this part with Example 36 into another small section 7.2 on "Global 
Property Constraints", and call the other "Global Graph-level 
Constraints"? People with RDFS background will look for the equivalent 
of rdfs:range...

>
> > Technically it's just a Shape like any other one, only that it has a
> > sh:scope instead of the (more common) class scoping. The sh:scope is
> > set to iterate over all nodes that appear as object in an rdf:type
> > triple. Alternatively it could also be turned into a
> > sh:AllObjectsScope statement. (We may need to expose this to some
> > practical experience to see which design patterns emerge).
> >
> > Holger
>
> We ought to be consistent in how we call things or confusion will 
> prevail. I have to tell you that reading through the spec I find a lot 
> of inconsistencies throughout and I think the spec would benefit from 
> adhering to more consistency. Not a showstopper for FPWD but something 
> to worry about for the long term.

Certainly. And the spec has been sitting relatively unchanged for a few 
months now, always welcoming feedback. I appreciate any kind of 
actionable details.

Thanks,
Holger