macro future-proofing regression: `$a:ty [ b ]` is now rejected

[pnkfelix]

(spawned off of #25658 (comment) )

macro future-proofing regression: $a:ty [ b ] is now rejected

This was indeed a direct consequence of the rules as stated in rust-lang/rfcs#1384

But since we may want to fix one-off cases like this, I am opening a separate issue and marking it as a regression.

[pnkfelix]

cc #30450

[pnkfelix]

cc @nikomatsakis @retep998

[LeoTestard]

As of 1.8.0, the following macro compiles on stable:

macro_rules! foo(
    ($a:ty [ $b:ident ]) => ()
);

I think this issue can be closed.