Re: shapes as classes

Yes I would also like to hear examples for disconnected graphs of 
related resources. Your link has something about Birth Certificates vs. 
Person instances and points out there is no triple between them. But if 
there is no triple between them, why not just check them separately, 
i.e. have one set of constraints attached to BirthCertificate and 
another set at Person, and the engine just validates all instances.

Holger


On 12/19/14, 4:44 AM, Eric Prud'hommeaux wrote:
>
>
> On Dec 18, 2014 6:10 PM, "Arthur Ryman" <ryman@ca.ibm.com 
> <mailto:ryman@ca.ibm.com>> wrote:
> >
> > Holger Knublauch <holger@topquadrant.com 
> <mailto:holger@topquadrant.com>> wrote on 12/15/2014 06:06:22 PM:
> >
> > > > In OSLC Resource Shapes, a shape program consists of a set of 
> related
> > RDF
> > > > :ResourceShape resources.
> > >
> > > To clarify: with "related" you mean via oslc:valueShape, or how else?
> > > With valueShape you could only cover connected subgraphs, not 
> arbitrary
> > > graphs.
> >
> > There are two OSLC mechanisms for defining a set of related shapes:
> > 1. A service, e.g. a URL where you can POST requests to create new
> > resources, may link to one or more shape resources using
> > oslc:resourceShape
> > 2. A resource shape may contain links to one or more other resource 
> shapes
> > associated with the values of properties using oslc:valueShape
> >
> > Yes, this means that OSLC resource shapes can only describe OO-like
> > connected graphs. That is a limitation which I pointed out in [1]. I 
> hope
> > this WG will define a spec that can handle disconnected graphs.
>
> What are some examples of when this is needed?
>
> > [1] http://www.w3.org/Submission/shapes/#disconnected-graphs
> >
> >
>