- From: Holger Knublauch <holger@topquadrant.com>
- Date: Mon, 23 Mar 2015 11:20:05 +1000
- To: public-data-shapes-wg@w3.org
On 3/22/2015 2:07, Karen Coyle wrote: > > On 3/19/15 7:44 PM, Holger Knublauch wrote: >> By the way I have also restructured the Overview of the Core Features so >> that it is only about *local* constraints. I have moved *global* >> constraints to the second half. > > My gut feeling is that the global constraints will be used most often > with the least complex data, and therefore should be in the core. Some "global" constraints that are only about a single property can be attached to the root class, rdfs:Resource/owl:Thing, making them apply to instances of all classes. We currently have not identified any high-level built-ins for other global constraints, which means that from the perspective of the Specification we could only talk about SPARQL-based global constraints or corresponding templates. However, since the Template mechanism is only introduced in the second part of the spec, we cannot rely on it in the first part. Furthermore, global constraints are not part of ShEx, so I doubt this feature will be embraced to go into part 1 by them. > > I don't see language that states how to define the desired focus node > in the SHACL language, and that should be very early on in the > document. Is this because we do not have agreement on that? If so, I > would like there be a placeholder, probably in section 3, Shapes. This > would define local and global shapes. Correct, I had moved the treatment of sh:nodeShape and rdf:type into section 11 (Supported Operations) because it required some understanding of the operations before they made any sense. I would need to copy some of the Operations into Part 1 to make it self-contained again. Yet these Operations need to mention Templates, so this cannot be done. These are all yet more problems caused by the stricter separation of these parts. In order to introduce how sh:nodeShape and rdf:type work, I repeat my offer to add a Quick Start section (2), and I'll probably work on a proposal of this next. > > BTW, the term "local shapes" does not appear anywhere in the document > yet. My SHACL spec only has "local constraints". What would "local shapes" be? Thanks Holger
Received on Monday, 23 March 2015 01:21:16 UTC