Support for classification

[afs]

Issue from a comment on #702.

The new text would be better if "deactivated" is defined to skip shapes/constraints rather than "All RDF terms conform to a deactivated shape". Classification tasks are affected.

There really are three cases, "known to positively conform", "known to not conform" and "not considered/unknown". sh:targetWhere is positive selection.

Note that sh:targetWhere can be interpreted as a "definition" providing necessary and sufficient conditions for a shape. It can therefore potentially be used for "classification" tasks, e.g., to collect all nodes that fulfill a given set of conditions.

Text references:

• sh:deactivated
• Note on sh:targetWhere (anonymous note anchor; link may break)

If classification tasks are a possibility, sh:deactivated should be defined to cause skipping, i.e. "not considered", and not defined to be conforming.

This also is better for a deactivity shape which references a shared part (sh:nodethat might be a shared shape accessible by a different route which would still apply).

There may be other places in the documents where similar revisions are needed.

[HolgerKnublauch]

@afs how is this issue different from #702 ?

[afs]

#702 is specific to sh:deactivated, working on the details of wording. Part of that is "classification" and may get lost in the detail.

This is the broader case of "classification" across all of SHACL.

Maybe there isn't anything else to do in which case this can be closed when that is confirmed.