error monomorphizing rather complex use of ifaces

[nikomatsakis]

If you look at the iter branch in the nikomatsakis/rust repo, you will find that it fails in stage0. I debugged this a bit by building a vanilla compiler in another repoistory and managed to trace it as far as a failure in make_mono_id() because the call to map2() that is processing bounds and substs receives a bounds vec of length 1 and a substs vec of length 2. This seems to occur with the call to vec::flatmap() that occurs in librustsyntax processing the attributes. I tried to narrow down the test to a smaller test file but was unable to find anything that reproduced the error other than the full rustc.

@marijnh if you wouldn't mind taking a look and let me know if something jumps out at you, I would appreciate it.

[catamorphism]

The symptoms look a lot like #2185 -- you have @marijnh 's fix for that, right?

[marijnh]

Ugh. It turns out that the last snapshot, although registered after my fix, is actually a snapshot of a tree that's quite a bit older. I've verified that when I do a local snapshot, this problem goes away. So this is a duplicate of #2185, as Tim pointed out. I'll start a new snapshot.