Coercing higher ranked associated functions into a function pointer requires wrapping in a closure

[parrottq]

Playground

struct A<'a>(&'a ());

impl<'a> A<'a> {
    fn associated(a: &'a (), b: &()) {}
}

fn main() {
    let works_not: for<'a> fn(&'a (), &()) = A::associated; // Error
    let works_not: for<'a, 'b> fn(&'a (), &'b ()) = A::associated; // Error
    let works: for<'a> fn(&'a (), &()) = |a, b| A::associated(a, b); // Ok
    let works: for<'a, 'b> fn(&'a (), &'b ()) = |a, b| A::associated(a, b); // Ok

    // Free standing functions work fine
    fn fun<'a>(a: &'a (), b: &()) {}
    let works: for<'a> fn(&'a (), &()) = fun; // Ok
}

I expected the associated function to be coercible into a function pointer directly or for a diagnostic to tell me to wrap the function in a closure to fix the error.

Instead, I got a confusing compiler error which lead me to think that the for<'_> was wrong, which doesn't seem to be the case.

error[E0308]: mismatched types
   --> src/main.rs:8:46
    |
195 |     let no_works: for<'a> fn(&'a (), &()) = A::associated; // Error
    |                   -----------------------   ^^^^^^^^^^^^^ one type is more general than the other
    |                   |
    |                   expected due to this
    |
    = note: expected fn pointer `for<'a, 'b> fn(&'a (), &'b ())`
                  found fn item `for<'a> fn(&(), &'a ()) {A::<'_>::associated}`

Meta

rustc --version --verbose:

rustc 1.85.1 (4eb161250 2025-03-15)
binary: rustc
commit-hash: 4eb161250e340c8f48f66e2b929ef4a5bed7c181
commit-date: 2025-03-15
host: x86_64-unknown-linux-gnu
release: 1.85.1
LLVM version: 19.1.7

[compiler-errors]

This is not a bug, but an observable reality of the difference between early- and late-bound lifetimes. In the impl impl<'a> A<'a>, 'a is "early-bound", and thus it cannot be treated like a late-bound/higher-ranked lifetime.

[parrottq]

Thanks for the explanation!

I felt that there might be a reason that this wasn't coercible but it wasn't clear to me why or how to find out more information. This is probably a diagnostic issue then.

I'll close the issue for now.