Nightly CI test failures due to unused assignment linting changes

[ranger-ross]

Seems a recent change in nightly-2025-10-13 broke this test

- [WARNING] `foo` (lib) generated 1 warning
+ [WARNING] `foo` (lib) generated 1 warning (run `cargo fix --lib -p foo` to apply 1 suggestion)

Example failures in CI:

• https://github.com/rust-lang/cargo/actions/runs/18465801706/job/52607374538

Looking at the logic for this line it appears we add the suggestion to run cargo fix if the warning is "fixable" and in a local crate.

The following snippet is how we determine if diagnostic is fixable

cargo/src/cargo/core/compiler/mod.rs

Lines 2169 to 2179 in bee8cb7

	let machine_applicable: bool = msg
	.children
	.iter()
	.map(|child| {
	child
	.spans
	.iter()
	.filter_map(|span| span.suggestion_applicability)
	.any(|app| app == Applicability::MachineApplicable)
	})
	.any(|b| b);

I think rustc changed the value of suggestion_applicability in its json message but I am unsure which PR in rust-lang/rust changed this behavior

[ranger-ross]

From #16092 (comment)

It failed due to this rust-lang/rust#142390. We addressed it in advance by #16082 but not for nightly-only cargo tests.

[weihanglo]

thread 'closure_immutable_outer_variable' (125364) panicked at crates/rustfix/tests/parse_and_replace.rs:141:9:

---- expected: ./tests/everything/closure-immutable-outer-variable.fixed.rs
++++ actual:   Current Fix
   1    1 | // Point at the captured immutable outer variable
   2    2 | 
   3    3 | fn foo(mut f: Box<dyn FnMut()>) {
   4    4 |     f();
   5    5 | }
   6    6 | 
   7    7 | fn main() {
   8      -     let mut y = true;
        8 +     let mut _y = true;
   9    9 |     foo(Box::new(move || y = false) as Box<_>); //~ ERROR cannot assign to captured outer variable
  10   10 | }

This failure seems like a false positive to me. #16098 (comment) and the failed job

We temporarily disable it to unblock the CI pipeline. Will investigate a bit more and talk to compiler folks to sort it out

[weihanglo]

After tracking the issue and the pull request in rustc, I think the suggestion is actually intentional ad liveness is tracked at MIR level, and the only way to supress unused assignment error is actually underscore-prefixing the original variable.

$ rustc +nightly main.rs
warning: value captured by `y` is never read
 --> main.rs:9:26
  |
9 |     foo(Box::new(move || y = false) as Box<_>); //~ ERROR cannot assign to captur...
  |                          ^
  |
  = help: did you mean to capture by reference instead?
  = note: `#[warn(unused_assignments)]` (part of `#[warn(unused)]`) on by default

warning: value assigned to `y` is never read
 --> main.rs:9:26
  |
9 |     foo(Box::new(move || y = false) as Box<_>); //~ ERROR cannot assign to captur...
  |                          ^^^^^^^^^
  |
  = help: maybe it is overwritten before being read?

warning: unused variable: `y`
 --> main.rs:8:9
  |
8 |     let mut y = true;
  |         ^^^^^ help: if this is intentional, prefix it with an underscore: `_y`
  |
  = note: `#[warn(unused_variables)]` (part of `#[warn(unused)]`) on by default

warning: 3 warnings emitted

So, we should probably re-enable the test and add an underscore.