Function call with attributes on arguments fail to parse

[repnop]

I just ran into a problem with r-a not parsing the function call when I have something like this

fn foo(a: Foo, #[cfg(debug_assertions)] b: Baz) { ... }

// ...

foo(a, #[cfg(debug_assertions)] b);

(r-a says it expects an expression, not attribute)

#1661 fixed the attributes when put on parameters, but not when calling functions with attributes on the arguments.

I did a test with adding AttrsOwner to Expr in the grammar file and that fixed it, but I'm not sure if that's the way to go or not. I could write up a quick PR if its something simple enough that though :)

[matklad]

I think this needs some research. The main question is where the arguments are paced: on the expresison, or on the function argument?

Ie, if foo(#[dummy] x) is allowed, is foo((#[dummy] x,)) allowed as well?

The former case should be easy-ish to fix, the latter might be more involved.

[repnop]

The second case does seem to be allowed: playground link example, running in debug/release passes different sized tuples to the function

[matklad]

Aha, here's the list where we should allow attributes on the expressions: https://doc.rust-lang.org/reference/expressions.html#expression-attributes

…

On Wed, 11 Dec 2019 at 14:06, Wesley Norris ***@***.***> wrote: The second case does seem to be allowed: playground link example <https://play.rust-lang.org/?version=stable&mode=release&edition=2018&gist=7b46482e93686158ef7f00bf42b1231a>, running in debug/release passes different sized tuples to the function — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#2521>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AANB3MZ4D7KI5ZHPMXQA65TQYDQT7ANCNFSM4JZGPWRA> .

[EddieIvan01]

Any progress?

[lnicola]

This is working now: