Warning about non-local impl-definition in derive-macro output

[avl]

Code

I tried this code:

const _: () = {
    const _: () = {
        impl Callable for Dummy {}
    };
    pub trait Callable {}
    struct Dummy;
};

I expected to see this happen: The code should compile without errors or warnings.

Instead, this happened: There is a warning: "warning: non-local impl definition, impl blocks should be written at the same level as their item".

This might seem very convoluted. It is a minimization of code generated by the savefile-derive crate when it is generating implementations of AbiExportable for traits returning boxed Fn-trait objects. It generates helper-types for these, and then effectively recurses, generating impls for these helpers types. That's why we get two levels of the const _: () = {...}-trick. This trick is used by crates like savefile and serde, but I suppose serde doesn't end up recursing and is probably not affected by this issue.

Version it worked on

It works in stable rust 1.81, and also in rustc 1.83.0-nightly (52fd99839 2024-10-10).

Version with regression

It does not work in rustc 1.83.0-nightly (1bc403daa 2024-10-11).

@rustbot modify labels: +regression-from-stable-to-nightly -regression-untriaged

[jieyouxu]

cc @Urgau as you know about non_local_definitions.

EDIT: I'm dumb and was looking at different issue pages.

[avl]

Hmm, I wonder if this is a duplicate of #131474 ?

The problem is still present in the latest nightly, but maybe the fix for the above issue is not yet in nightly?

[avl]

Do you have a reproducer that we can look at to determine if the lint should or should not fire in this case? In any case, cc @Urgau as you know about non_local_definitions. I suspect this might be "working as intended" but it's hard to say without being able to look at an actual example.

Isn't the example in the bug-description enough? What more would you like to see?

[avl]

Here is an example in the rust playground: https://play.rust-lang.org/?version=nightly&mode=debug&edition=2021&gist=852a5e96a872b19ee42b26a0105275cf

[jieyouxu]

Ah wait I confused myself, I think this is #131474 my bad.