Dealing with “var x” – Part 1

[Nigel-Ecma]

Review in conjunction with #1663.

---

Dealing with “var x” – Part 1

There is an ambiguity in patterns, the input “var x” could be recognised as a declaration_pattern or as a var_pattern. As is this PR only deals with the latter, on line 191 below, and says nothing about the former.

I think the intended semantics are:

• In declaration_pattern the type cannot be the identifier var; i.e. if it starts with var then var_pattern is considered.
• A var_pattern is invalid (not considered as a valid parse) if a type called var is in scope.

This is tackled with two suggestions:

1. This one, covering declaration_pattern
2. Reword the original statement on line 191 (Part 2)

The *type* of a *declaration_pattern* cannot be the *identifier* `var`.

> *Note:* In such a case the input may be recognisable as a *var_pattern* (§11.2.4).

Originally posted by @Nigel-Ecma in #873 (comment)

[KalleOlaviNiemitalo]

The type of a declaration_pattern cannot be the identifier var.

OK, that still allows the identifier @var, or v\u0061r for that matter. Might be worth a note, though.

class C {
    struct var {}
    
    void M(var j) {
        if (j is @var q) {
        }
    }
}