solve ICE of issue #85350 by avoiding the case with ty::Error

[ABouttefeux]

fix #85350

---

avoids looking for bounds in the case of supertrait on ty::Error

[rust-highfive]

r? @varkor

(rust-highfive has picked a reviewer for you, use r? to override)

[jackh726]

r? @jackh726

[ABouttefeux]

I just want to check something, maybe there is a better way to fix it. Anyway I will need to move the test for the CI to pass.

[jackh726]

Hmm. So this might be one way to solve this. What I was sort of expecting was something around when we resolve the function elision (because we follow function elision rules in FnMut(&Context)) we just add that anonymous region to the impl lifetimes. But that doesn't make sense now that I think about it, because impl lifetimes are early bound and function lifetimes are late bound

[jackh726]

Ah but I guess its treating that FnMut as a trait object. So we could add the lifetimes to that.

[jackh726]

Interestingly, adding a dyn to the original repro doesn't ICE anymore. So there's some mismatch somewhere

[ABouttefeux]

Ok thanks 😃

[ABouttefeux]

After a bit of digging I think E0229 is erroneously emitted

error[E0229]: associated type bindings are not allowed here
 --> issue-85350.rs:3:6
  |
3 | impl FnMut(&Context) for 'tcx {
  |      ^^^^^^^^^^^^^^^ associated type not allowed here

Where as

impl FnMut(&dyn Context) for 'tcx {
    fn print () -> Self::Output{ }
}

does not emit this error code ( and does not ICE)

---

rustc --explain E0229:

An associated type binding was done outside of the type parameter declaration
and where clause.

Erroneous code example:

pub trait Foo {
    type A;
    fn boo(&self) -> <Self as Foo>::A;
}

struct Bar;

impl Foo for isize {
    type A = usize;
    fn boo(&self) -> usize { 42 }
}

fn baz<I>(x: &<I as Foo<A=Bar>>::A) {}
// error: associated type bindings are not allowed here

To solve this error, please move the type bindings in the type parameter
declaration:

fn baz<I: Foo<A=Bar>>(x: &<I as Foo>::A) {} // ok!

Or in the where clause:

fn baz<I>(x: &<I as Foo>::A) where I: Foo<A=Bar> {}

[ABouttefeux]

Interestingly

impl FnMut(&'tcx Context) for 'tcx {
    fn print () -> Self::Output{ }
} 

does not ICE and emits E0229 so maybe it is something else

[jackh726]

FnMut(&'tcx Context) there should desugar to FnMut<(&'tcx Context, ), Output=()>; so, I think this makes sense:
the impl should be:

impl<'a> FnMut<(&'a Context,)> for 'tcx {
    type Output = ();
    fn print () -> Self::Output{ }
}

This really is a "it's really not clear what we should be doing because the syntax is quite wrong" situation. So it's hard to figure out what the user intended. So, a few options:

1. Try to bail out before lowering at all
2. Treat the FnMut(...) "normally" (i.e. as an elision context) somehow; this is difficult because the elided lifetimes are late bound, so I'm not sure where we would add that in terms of bound vars
3. Make elision here not allowed (i.e. fall back to 'static)
4. Treat the FnMut(...) as a trait object and add the elided lifetime to the trait object's trait ref

[ABouttefeux]

I am a bit stuck. I am do not find where the lowering precisely happens for FnMut implementation precisely or where the bound region are added for elided lifetime.

I do not really see how to do 3 or 4 yet.
and for 1 I am not sure at which point I should bail out.

[jackh726]

I am a bit stuck. I am do not find where the lowering precisely happens for FnMut implementation precisely or where the bound region are added for elided lifetime.

I do not really see how to do 3 or 4 yet.
and for 1 I am not sure at which point I should bail out.

@ABouttefeux let me think about this and try to write some notes this weekend

[ABouttefeux]

@jackh726 any updates ? Take the time you need 😄 . Lately I did not had a lot of time anyway.

[jackh726]

I haven't been able to get to this :/

[jackh726]

@ABouttefeux this is honestly kind of an edge case; I'm okay with just landing this as-is,

[jackh726]

That being said...the test location isn't correct. And I'm not quite sure the best place for it. Maybe just under src/test/ui/issues as issue-85350-anon-lifetimes-in-malformed-impl.rs or something like that.

@bors delegate+

just @bors r=jackh726 when test is moved if you're okay with this fix

[bors]

✌️ @ABouttefeux can now approve this pull request

[bors]

📌 Commit 42f7764 has been approved by `jackh726``

[jackh726]

@bors r- looks like bors picked that up...

[bors]

☔ The latest upstream changes (presumably #93938) made this pull request unmergeable. Please resolve the merge conflicts.

[jackh726]

I'm going to go ahead and close this as inactive.