More precisely point out what is immutable for E0596

[chenyukang]

Fixes #113842

[rustbot]

r? @compiler-errors

(rustbot has picked a reviewer for you, use r? to override)

[compiler-errors]

I'm not exactly sure what this is trying to demonstrate.

The issue here is a subtle detail -- &mut X will try to autoderef X to match it to the type it needs. From the surface, &mut "" looks like it should give you a type of &mut &'static str, which is valid, but it's not valid because we autoderef it to make it into a &mut str:

fn foo() {
    // works
    let x           = &mut "";
    // doesnt work
    let y: &mut str = &mut "";
}

If you're not pointing this behavior out, then I don't really know if it's more helpful or more misleading to say "this is an expr of type &str".

[chenyukang]

how we don't give this hint when X is a ExprKind::Lit(_) ?
since seems it does not provide more details.

[compiler-errors]

I still don't know if I like the presentation of this diagnostic. It still doesn't explain that the problem here is the autoderef behavior of &/&mut -- you're allowed to take a reference to &mut &Foo -- the problem is that there's a deref being inserted there that's not present in the code, and I think a diagnostic improvement here should explain this. Without that, you're really just adding more labels.

[compiler-errors]

@rustbot author

[chenyukang]

I think one of the confusion from the origin issue is that the span contains the &mut, this PR improved it.

Considering autoderef, most newcomers may overlook the subtle details, I'm not sure whether it's helpful to refer to it.

[compiler-errors]

Hi @chenyukang: I still don't believe this PR makes the underlying issue more easy to understand. It just points out that the pointee of the &mut ref is immutable. That alone does not explain what the root cause of the issue, which is that autoderef will attempt to coerce &mut &T to &mut T.

@rustbot author

[bors]

☔ The latest upstream changes (presumably #120576) made this pull request unmergeable. Please resolve the merge conflicts.

[Dylan-DPC]

@chenyukang any updates on this? thanks