`Box<dyn FnOnce(T)>` doesn't seem to be really contravariant with T

[fakedrake]

I tried this code:

type F<'arg> = Box<dyn FnOnce(&'arg ())>;
fn contravar<'small, 'large: 'small>(f: F<'small>) -> F<'large> {
    f
}

I expected it to compile but I got the following:

error: lifetime may not live long enough
 --> src/lib.rs:3:5
  |
2 | fn contravar<'small, 'large: 'small>(f: F<'small>) -> F<'large> {
  |              ------  ------ lifetime `'large` defined here
  |              |
  |              lifetime `'small` defined here
3 |     f
  |     ^ function was supposed to return data with lifetime `'large` but it is returning data with lifetime `'small`
  |
  = help: consider adding the following bound: `'small: 'large`

playground

$ rustc --version --verbose
rustc 1.65.0-nightly (29e4a9ee0 2022-08-10)
binary: rustc
commit-hash: 29e4a9ee0253cd39e552a77f51f11f9a5f1c41e6
commit-date: 2022-08-10
host: x86_64-unknown-linux-gnu
release: 1.65.0-nightly
LLVM version: 14.0.6
`rustc --version --verbose`:

I asked this on stackoverflow too hopefully I am not misinterpreting something about variance.

[fakedrake]

I looked into the issue a bit carefully and the problem seems to be that the type checker seems to treat the Fn/FnMut/FnOnce traits just like any other trait w.r.t. to variance, making them invariant to their argument lifetimes. Unfortunately by the point when the point when variance is checked I can't find a way to get the TyCtxt in order to check that the trait is one of Fn, FnMut, FnOnce so that it can check for contravariance instead of invariance.

I don't even know if that is good enough because what about traits like:

trait MyFn<'a>: Fn(&'a ()) {}

But it would be a start. As things stands, it seems impossible to have a Vec of closures that behaves as expected w.r.t. to variance.

[zirconium-n]

Might be related #23938 (comment)