Use () instead of void

[ByteEater-pl]

If void and _ are both problematic, and no single character alternative reaches consensus, perhaps () would work? The case using () = needs a lookahead for = in order to distinguish from CallExpression, but it seems much less quirky than what the two main contenders would require so far.

[rbuckton]

Since using is not a reserved word, using () = x would be ambiguous with an Extractor.

[ByteEater-pl]

Indeed, that's going to be the syntax for extractors. Since using is already a contextual keyword, it could be special cased – e.g. either completely prohibited as an extractor or just in the only ambiguous context that comes to my mind, i.e. without const, let or var, a top-level assignment (not nested). If the latter is preferred, the assignment could be easily achieved by parenthesizing, thus: (using() = expression).

[ljharb]

Empty parens reads like a bug to me - this would be a very confusing syntax choice.

[ByteEater-pl]

@ljharb, your opinion, disagreeing with mine, is likely to have more clout, so let me add some arguments. Empty parentheses are already used to denote empty ArrowParameters. And in a number of other languages they denote the unit value or unit type (or even both, in Haskell), or some empty structure.

But if it won't fly, then maybe 0 (unless it's going to be allowed in target positions, like it already is for specifying property names in object literals, e.g. {2: 3})? It means pure virtual in C++ (admittedly a remote analogy), so there's also precedent for giving it special meaning outside number literals. (And, even more remotely, GOTO 0 and RESUME 0 in some dialects of BASIC.)

[rbuckton]

0 is not an option as it would conflict with pattern matching. This syntax must be valid for both destructuring, assignment, and pattern matching to be viable.

[rbuckton]

If we are unable to use _, then void will likely work just fine. The only major concern I've seen was related to the cover grammar, which I believe I have solved in #9.

[gamebox]

As I've said elsewhere, I think the simple fact that void is four letters will severely hamper its uptake as many developers will find it less convenient and not understand any issues with just binding the value to a much shorter identifier.

[rbuckton]

As I've said elsewhere, I think the simple fact that void is four letters will severely hamper its uptake as many developers will find it less convenient and not understand any issues with just binding the value to a much shorter identifier.

That is a different topic than using (). See #11 about the potential to use identifiers.

[rbuckton]

As () is not a viable option, I am closing this issue.