[XPath] Allowing multiple let clauses in LetExpr and for clauses in ForExpr

[qt4cg]

[XPath] Allowing multiple let clauses in LetExpr and for clauses in ForExpr

[michaelhkay]

It's increasingly difficult to justify leaving any XQuery FLWOR constructs out of XPath. The only place I'd draw the line is where the semantics can only be described in terms of "tuple of variable bindings", because (a) that requires internal machinery in an XPath engine that's not otherwise needed, and (b) the only things it enables are things that aren't remotely useful, like being able to order the sequence of tuples independently of any of its input sequences.

[rhdunn]

This interacts with the for member syntax when that is part of the SimpleForClause. Specifically, it means that SimpleForClause cannot be expressed as:

SimpleForClause ::= "for" "member"?  SimpleForBinding (("," | "for")  SimpleForBinding)*

as it would break the semantics of for member expressions due to the for keyword switching to sequence iteration, but not being able to easily identify that from the parse tree. The separator would also need to be ("," | "for" "member"?) to support mixing array and sequence for expressions.

As such, this would need to be done on the ForExpr as originally proposed (which would continue to work if/when extended to maps):

ForExpr ::= (SimpleForClause)+ "return" ExprSingle

[rhdunn]

Fixed by PR #344.