Lifetime-polymorphic anonymous function causing compiler panic

[asajeffrey]

The program at https://play.rust-lang.org/?gist=1a60fe4cd514848f6788 generates the following:

<anon>:44:5: 44:26 error: internal compiler error: Encountered errors `[FulfillmentError(Obligation(predicate=Binder(TraitPredicate(<[closure@<anon>:40:19: 40:65] as core::ops::Fn<(<Str as TypeWithLifetime<'_>>::Type,)>>)),depth=1),OutputTypeParameterMismatch(Binder(<[closure@<anon>:40:19: 40:65] as core::ops::Fn<(&'static str,)>>), Binder(<[closure@<anon>:40:19: 40:65] as core::ops::Fn<(<Str as TypeWithLifetime<'_>>::Type,)>>), Sorts(ExpectedFound { expected: <Str as TypeWithLifetime<'_>>::Type, found: &'static str }))), FulfillmentError(Obligation(predicate=Binder(TraitPredicate(<[closure@<anon>:40:19: 40:65] as core::ops::FnOnce<(<Str as TypeWithLifetime<'_>>::Type,)>>)),depth=1),OutputTypeParameterMismatch(Binder(<[closure@<anon>:40:19: 40:65] as core::ops::FnOnce<(&'static str,)>>), Binder(<[closure@<anon>:40:19: 40:65] as core::ops::FnOnce<(<Str as TypeWithLifetime<'_>>::Type,)>>), Sorts(ExpectedFound { expected: <Str as TypeWithLifetime<'_>>::Type, found: &'static str })))]` fulfilling during trans
<anon>:44     consumer.accept("hi");
              ^~~~~~~~~~~~~~~~~~~~~
note: the compiler unexpectedly panicked. this is a bug.
note: we would appreciate a bug report: https://github.com/rust-lang/rust/blob/master/CONTRIBUTING.md#bug-reports
note: run with `RUST_BACKTRACE=1` for a backtrace
thread 'rustc' panicked at 'Box<Any>', ../src/libsyntax/diagnostic.rs:176

Reported on #rust-lang irc, @aturon reckons this is new, and should be filed with a mention of @nikomatsakis.

[jonas-schievink]

Reduced:

trait TypeWithLifetime<'a> {
    type Type;
}

type At<'a,T> where T: TypeWithLifetime<'a> = T::Type;

fn main() {
    let _ = |x:At<()>| false;
}

[asajeffrey]

The reduced example should fail to typecheck, I'd hope that my one actually would typecheck.

[jonas-schievink]

Oh, you're right. I'm wondering if at least my example is caused by #21903

[nikomatsakis]

Some data. First off, I modified the test by removing the At alias and got this, which fails differently:

http://is.gd/xIgh8f

error: internal compiler error: unexpected panic
note: the compiler unexpectedly panicked. this is a bug.
note: we would appreciate a bug report: https://github.com/rust-lang/rust/blob/master/CONTRIBUTING.md#bug-reports
note: run with `RUST_BACKTRACE=1` for a backtrace
thread 'rustc' panicked at 'assertion failed: !projection_ty.has_escaping_regions()', ../src/librustc/middle/traits/fulfill.rs:149

At one point, I had modified the list closure to read |x:&str| instead http://is.gd/jSRlJG (which I think is equivalent) and then the error is somewhat different:

<anon>:44:14: 44:26 error: type mismatch: the type `[closure@<anon>:40:19: 40:62]` implements the trait `for<'r> core::ops::Fn<(&'r str,)>`, but the trait `for<'r> core::ops::Fn<(<Str as TypeWithLifetime<'r>>::Type,)>` is required (expected associated type, found &-ptr) [E0281]
<anon>:44     consumer.accept("hi");
                       ^~~~~~~~~~~~
<anon>:44:14: 44:26 help: see the detailed explanation for E0281
<anon>:44:14: 44:26 error: type mismatch: the type `[closure@<anon>:40:19: 40:62]` implements the trait `for<'r> core::ops::FnOnce<(&'r str,)>`, but the trait `for<'r> core::ops::FnOnce<(<Str as TypeWithLifetime<'r>>::Type,)>` is required (expected associated type, found &-ptr) [E0281]
<anon>:44     consumer.accept("hi");
                       ^~~~~~~~~~~~
<anon>:44:14: 44:26 help: see the detailed explanation for E0281
error: aborting due to 2 previous errors

what all of this is pointing at is some errors around the late-bound lifetimes. The ICE in question may well arise due to getting the accounting wrong in trans, unclear.

[istankovic]

With rustc 1.26.0-nightly (adf2135ad 2018-03-17) the reduced example by @jonas-schievink fails to typecheck i.e. it no longer ICEs, however http://is.gd/xIgh8f is still problematic.

[pnkfelix]

The only remaining issue here is http://is.gd/xIgh8f, which I am treating as duplicate of #62529, the issue where I will try to track future instances of this field of ICE. Closing as fixed for the initial bug report, and duplicate for the related issue http://is.gd/xIgh8f