`unused_imports` false positive

[Kixunil]

Code

#![no_std]

#[cfg(feature = "alloc")]
#[macro_use]
extern crate alloc;

#[cfg(test)]
mod test {
    #[test]
    #[cfg(feature = "alloc")]
    fn test() {
        format!("foo");
    }
}

Current output

warning: unused `#[macro_use]` import
 --> src/lib.rs:4:1
  |
4 | #[macro_use]
  | ^^^^^^^^^^^^
  |
  = note: `#[warn(unused_imports)]` on by default

Desired output

No warning.

Rationale and extra context

#[macro_use] is actually used. When I removed it I get compilation failure. And while it can be argued, this is not the best style and use alloc::format; is better, the lint should say it if it's the case. (But really, I think this belongs to clippy, not rustc.)

Other cases

No response

Rust Version

rustc 1.75.0 (82e1608df 2023-12-21)
binary: rustc
commit-hash: 82e1608dfa6e0b5569232559e3d385fea5a93112
commit-date: 2023-12-21
host: x86_64-unknown-linux-gnu
release: 1.75.0
LLVM version: 17.0.6

rustc 1.77.0-nightly (5bd5d214e 2024-01-25)
binary: rustc
commit-hash: 5bd5d214effd494f4bafb29b3a7a2f6c2070ca5c
commit-date: 2024-01-25
host: x86_64-unknown-linux-gnu
release: 1.77.0-nightly
LLVM version: 17.0.6

### Anything else?

_No response_

[ChrisDenton]

It's only used in #[cfg(test)] though? So it will be unused without that cfg.

[Noratrieb]

Yeah, cfgs are not (and cannot really be) factored into these lints.

[Kixunil]

@ChrisDenton oh, good point, that's both true and still confusing.

@Nilstrieb IIRC there was some work to be able to at least suggest things in lints based on inactive cfg, maybe the same could work here?

[Noratrieb]

There was (and I happened to do that work :3) but this is pretty unrelated and would require properly checking cfg-ed out code, which is exactly what we don't want to do and can't do.

[Kixunil]

Maybe a different approach would work: when running cargo test run with test config first, make a list of used items in tests and apply that list next when compiling non-test to emit "Note: this is used in test but not in non-test" message. Some hint like that would be very useful even if it's not perfect and there are probably edge cases.

[petrochenkov]

All unused lints work after cfgs are expanded, this one is no different, so I'm going to close the issue.

The recent extension of unused_imports had some large effect on users in general, that is discussed in #121708 which is still open.