ICE with nested macros

[alexcrichton]

This program:

use core::util::with;
use core::hashmap::linear::LinearMap;

fn main() {
  let mut map = LinearMap::new();
  map.insert(1, 1);

  macro_rules! add_foo(
    ($a1:expr, $b1:expr) => ({
      map.find(&$a1).map(|&m| *m + $b1);
    })
  );

  macro_rules! foo(
    ($a:expr, $b:expr) => ({
      add_foo!($a, $b);
      add_foo!($b, $a);
    })
  );

  let def = 1;
  foo!(def, def);
}

fail with:

$ rustc --test bug2.rs
bug2.rs:22:12: 22:15 error: internal compiler error: No variable registered for id 25
bug2.rs:22   foo!(def, def);
                       ^~~
bug2.rs:8:2: 12:4 note: in expansion of add_foo!
bug2.rs:16:6: 16:23 note: expansion site
bug2.rs:14:2: 19:4 note: in expansion of foo!
bug2.rs:22:2: 22:17 note: expansion site

[catamorphism]

Seems non-critical for 0.7. Nominating for milestone 5 (production-ready).

[graydon]

accepted for production ready

[emberian]

Updated test case:

use std::util::with;
use std::hashmap::HashMap;

fn main() {
  let mut map = HashMap::new();
  map.insert(1, 1);

  macro_rules! add_foo(
    ($a1:expr, $b1:expr) => ({
      map.find(&$a1).map(|&m| *m + $b1);
    })
  );

  macro_rules! foo(
    ($a:expr, $b:expr) => ({
      add_foo!($a, $b);
      add_foo!($b, $a);
    })
  );

  let def = 1;
  foo!(def, def);
}

$ rustc foo.rs
foo.rs:22:12: 22:15 error: internal compiler error: No variable registered for id 20
foo.rs:22   foo!(def, def);
                      ^~~
foo.rs:8:2: 12:4 note: in expansion of add_foo!
foo.rs:16:6: 16:23 note: expansion site
foo.rs:14:2: 19:4 note: in expansion of foo!
foo.rs:22:2: 22:17 note: expansion site

[kud1ing]

I think the tag "A-macos" is unfortunate, since this is not Mac OS specific. Or is it?

[emberian]

@kud1ing good catch, someone probably mistook it for macros. It's been updated to A-mac-osx though, which should prevent that.

[msullivan]

Still ICEs.

[alexcrichton]

Closed with what I believe was the renumbering of the AST. There's plenty of tests for this already present, so closing issue entirely.