Should `completeTyping()` include ancestors?

[ericprud]

1.5.2 Validation Definition says:

n@sl EXACTLY ::- satisfies(n, def(s.label), G, Sch, completeTyping(G, Sch), neigh(n))

n@sl ::- satisfiesDescendant(n, def(s.label), G, Sch, completeTyping(G, Sch), neigh(n))

where I understand completeTyping() to be making sure that the same total truth is maintained regardless of which pieces of it you interrogate with isValid queries. The same section then defines a correct typing:

A correct typing is a typing such that for every RDF node/shape pair (n,l), satisfies(…) holds or satisfiesDescendant(…) holds.

Should the correct typing be ONLY satisfies(…)? This would allow it to answer shapemaps where EXACTLY is true.

I'm guessing that the current text (post-review by Iovka and Labra) is fine because we're counting on the first two lines of this issue text to differentiate EXACTLY from, um, not EXACTLY. Just wanted your eyes to verify this.

[iovka]

I realize that, because now the typing contains labels and not shapes, isValid does not need to call satisfies.

If completeTyping contained only those who satisfy exactly, then the definition of isValid(G, Sch, ism) could be for every (n, sl, exact) in ism,

• if exact, completeTyping(G, Sch) contains (n, sl)
• otherwise, completeTyping(G, Sch) contains (n, sl2) for some sl2 non abstract descendant of sl

Now about the main question, does completeTyping need to have only the exactly satisfied shape expr labels, or also those that are satisfied by some descendant: I think we can do it this way. It is not yet completely clear for me whether the two are equivalent wrt the properties that we need for completeTyping, so I preferred to be conservative. But after some thoughts I'm relatively confident that completeTyping restricted to those that satisfy exactly could also be ok.