Diagnostic on missing "?" suggests to ignore the error entirely, which can be problematic

[RalfJung]

Code

fn fallible() -> Result<i32, ()> {
    Ok(42)
}

pub fn caller() -> Result<(), ()> {
    fallible();
    Ok(())
}

Current output

warning: unused `Result` that must be used
 --> src/lib.rs:6:5
  |
6 |     fallible();
  |     ^^^^^^^^^^
  |
  = note: this `Result` may be an `Err` variant, which should be handled
  = note: `#[warn(unused_must_use)]` on by default
help: use `let _ = ...` to ignore the resulting value
  |
6 |     let _ = fallible();
  |     +++++++

Desired output

Rust should suggest to add the missing `?`, not to ignore the error entirely.

Rationale and extra context

See rust-lang/miri#3855 for an example where ignoring the error with let _ = can lead to critical bugs.

Other cases

No response

Rust Version

1.81.0

Anything else?

No response

[roife]

@rustbot claim

[roife]

In functions that return Result<>, handling it this way feels quite natural. However, doing so in functions that don't return Result<> seems a bit weird.

I'm wondering how we handle the latter case: should we keep using let _ =, or recommend users to handle with a match { Ok(_) => {} Err(_) => {} } clause?