Float comparison lint doesn't seem to trigger on float array comparison

[Lokathor]

#[test]
fn m128_max() {
  let a: m128 = cast([5.0_f32, 6.0, 7.0, 8.5]);
  let b: m128 = cast([-8.0_f32, 4.0, 12.0, 0.5]);
  let out: [f32; 4] = cast(a.max(b));
  assert_eq!(out, [5.0_f32, 6.0, 12.0, 8.5]);
  assert_eq!(out[0], [5.0_f32, 6.0, 12.0, 8.5][0]);
}

the second assert triggers clippy, the first doesn't.

[briankabiro]

Hi, I would like to take a look at this.

[flip1995]

Great! If you have any questions, don't hesitate to ask.

[briankabiro]

Cool, I am going to go through it and make sure to reach out. :)

[briankabiro]

Hi, @flip1995. I took the time to go through the codebase and think that the code change should be somewhere in the misc.rs

I am not sure why it wouldn't trigger when comparing array values. Could you offer some pointers?

[flip1995]

You're on the right track. In the line you linked is a check for the left and right side of the Eq/Ne Binop if both sides are floats:

rust-clippy/clippy_lints/src/misc.rs

Lines 521 to 523 in e7d153a

	fn is_float(cx: &LateContext<'_, '_>, expr: &Expr) -> bool {
	matches!(walk_ptrs_ty(cx.tables.expr_ty(expr)).sty, ty::Float(_))
	}

This function takes the expression, "derefs" it, and checks if the type is float. (i.e is_float(&&float) = is_float(float) = true, but is_float([float; N]) = false). So you have to modify this function, so that it not only checks the expressions type to be ty::Floats, but also for ty::Array(ty::Float, _)s.

Hope this helps :)

[briankabiro]

Yes, this definitely helps. I think I can work with this. Thanks!

[briankabiro]

Hey @flip1995. I've been caught up with work and this slipped under my radar; apologies for not giving an update sooner. I'd be happy to let someone else finish this if there's someone that you might recommend.

If not, I'd still be interested in completing the feature. 😃