Cannot use ty followed by >> on macro

[thelink2012]

Using ty followed by >> doesn't work in a macro but it does works when followed by > > instead. This seems very like one of the old C++03 template issues.

Using >>: http://is.gd/KTICUL

macro_rules! test {
    (
        fn fun() -> Option<Box<$t:ty>>;
    ) => {
        fn fun(x: $t) -> Option<Box<$t>>
        { Some(Box::new(x)) }
    }
}

test! {
    fn fun() -> Option<Box<i32>>;
}

fn main() {
    fun(0).unwrap();
}

Gives

<anon>:4:32: 4:37 error: `$t:ty` is followed by `>>`, which is not allowed for `ty` fragments
<anon>:4         fn fun() -> Option<Box<$t:ty>>;
                                        ^~~~~

Using > > instead of >> fixes it: http://is.gd/gSD3MR

Meta

All the tests were conducted on the playpen using the nightly and beta option (no difference) so I'm not sure about the version but the issue also happens on my local installation rustc 1.0.0-beta.3 (5241bf9c3 2015-04-25) (built 2015-04-25) (yep a bit outdated but cba to update eh).

[steveklabnik]

Triage: no change

[keeperofdakeys]

This might be fixed now
ping @jseyfried

[jseyfried]

This is still a bug -- I believe the issue is that the follow set checker does not allow >> after ty fragments.
cc @LeoTestard

[LeoTestard]

@jseyfried: I guess you meant ty instead of tt? Indeed, it seems to me that it's a bug since >> can already occur umambiguously after a type in the grammar so it should be added to the FOLLOW set.

I'm not sure what the procedure is for adding tokens to FOLLOW sets. Should we go through the RFC process or just assume that it's a bug in the current design and just make a PR for it?

[jseyfried]

@LeoTestard

I guess you meant ty instead of tt?

Yeah, edited.

I'm not sure what the procedure is for adding tokens to FOLLOW sets.

I think we should assume it is a bug and make a PR to fix. Adding a comment on the RFC PR or tracking issue would probably be a good idea, but I don't think we need to go through the RFC process.