- From: Peter F. Patel-Schneider <pfpschneider@gmail.com>
- Date: Fri, 22 Jan 2016 06:58:07 -0800
- To: Arthur Ryman <arthur.ryman@gmail.com>, "public-data-shapes-wg@w3.org" <public-data-shapes-wg@w3.org>
On 01/21/2016 06:57 PM, Arthur Ryman wrote: > As per the discussion today. See [1]. > > [1] http://w3c.github.io/data-shapes/shacl/#definition-of-implicit-scopeClass > > -- Arthur The relevant text is, I believe: A SHACL processor MUST recognize a resource X in the shapes graph as a shape if and only if the shapes graph contains a triple X rdf:type S where S is either sh:Shape or S is connected to sh:Shape by a property path in the shapes graph consisting of one or more triples whose predicate is rdfs:subClassOf. Similarly, a SHACL processor MUST recognize a resource X in the shapes graph as a class if and only if the shapes graph contains a triple X rdf:type C where C is either rdfs:Class or C is connected to rdfs:Class by a property path in the shapes graph consisting of one or more triples whose predicate is rdfs:subClassOf. If a SHACL processor recognizes a resource X in the shapes graph as both a shape and a class as defined here then all instances of the class X in the data graph MUST be included in the scope of the shape X. For the purposes of this definition, a resource R in the data graph is said to be an instance of the resource X if and only if the data graph contains a triple R rdf:type X or the data graph contains a triple R rdf:type Y and Y is connected to X by a property path in the data graph consisting of one or more triples whose predicate is rdfs:subClassOf. Quite a mouthful, but this appears to do the trick. This may, however, upset some of the "implicit typing" simplifications elsewhere in the document, such as 11. Validation of shapes graphs and the sh:defaultValueType property Example 8 Example 27 Example 28 The first mention of what is a shape is right at the beginning of Section 2. The relevant text above is quite a bit later, which should be remedied. The comment Issue 23: Use of shapes graph versus data graph for metadata We should be consistent here. We put shape information in the shapes graph. Class information should also be in the shapes graph, especially for designers who couple shapes and classes. Both shapes and classes are kinds of metadata so they should be in the same graph. should probably be removed, but I do agree that the early examples should not use a combined data and shapes graph. peter
Received on Friday, 22 January 2016 14:58:37 UTC