Items with 'lexical' scope have poor scoping

[nrc]

E.g.,

struct Foo;

impl Foo {
    fn bar(&self) {}

    fn foo(&self) {
        fn bar(x: &Foo) {}

        bar(self);
    }
}

fn main() {}

Is an error, but shouldn't be, because the bar method hides the locally declared bar function inside foo.

I think this is a bug, so it shouldn't be on any milestone, but nominating just in case.

[oli-obk]

It's not an error anymore: http://is.gd/CKvJn3

[nrc]

Ah curses, that isn't an error. Here's another go:

struct Foo<X> {
    x: Box<X>
}

impl<Bar> Foo<Bar> {
    fn foo(&self) {
        type Bar = i32;

        let _: Bar = 42;
    }
}

In this case the let statement should see the local type alias, but sees the type parameter.

[pnkfelix]

classifying as "just a bug", P-high, not 1.0.

Actually leaving as P-backcompat-lang, to represent the fact that this can lead to programs being accpeted that shoudl be rejected. So we will keep that tag as is.

But still not 1.0 blocker.

[nikomatsakis]

triage: P-medium

This has been sitting for a while and we don't expect immediate action, though hopefully a refactoring of resolve would lead to this being fixed.