Definitions and linking of `binding` and `solution`

[afs]

The definition of binding is id="#dfn_bindings" and the singular does not link. This is needed for reference from other documents.
It should be <dfn>binding</dfn> and reSpec will also allow the plural within the document.

"solutions" has the same linking issue.

The definition of "solution" is incomplete and not compatible with SPARQL.

It is defined as a set of pairs but does not say that the the variables of the pairs are unique.The current definition allows (?x, 1), (?x , 2).

SPARQL section may help:
http://w3.org/TR/sparql12-query/#defn_sparqlSolutionMapping:

A solution mapping, μ, is a partial function μ : V → T, where V is the set of all variables and T is the set of all RDF terms.

SHACL says:

"Variables are not required to be bound in a solution." so (?x , _nothing_) is legal.

[ajnelson-nist]

I'm curious: how does SPARQL define variables in solutions corresponding to OPTIONAL clauses that end up with no binding?

[afs]

In SPARQL, a variable is in a solution mapping with a value or it is not mapped (a SPARQL solution is a partial function, not an arbitrary set of pairs). A variable in a non-matching OPTIONAL is not recorded, which allows it to become bound later in the query.

[ajnelson-nist]

Is there any particular desire to distinguish SHACL and SPARQL on this point? My hunch would be they should be consistent, esp. in binding nothing vs. not binding at all.

[afs]

I believe not, but the text in SHACL Core defines solutions as a set of pairs, not a mapping. Solutions are used in shacl-sparql.

There is a different mechanism for node-expression evaluation. It overloads many SPARQL functions for node-expressions.