Remove superfluous CoreFloat imports

[PoignardAzur]

No description provided.

[PoignardAzur]

   error[E0599]: no method named `abs` found for type `f32` in the current scope
   --> parley/src/util.rs:7:13
    |
  7 |     (x - y).abs() < f32::EPSILON
    |             ^^^ method not found in `f32`
    |

Wait, I'm confused.

cargo check -p parley --no-default-features --features libm works on my machine. Does the presence of abs in no_std mode depend on the target?

[nicoburns]

That is indeed weird. The docs show that this has been stable since 1.0, but I think that's a lie for nostd. I believe this was moved from std to core relatively recently. But this compiles locally for me with your command and Rust 1.82.

[PoignardAzur]

But this compiles locally for me with your command and Rust 1.82.

So this should compile for CI too, right?

[tomcur]

Changelog claims these were moved into core in 1.84: https://releases.rs/docs/1.84.0/.

[PoignardAzur]

I've tried Rust 1.82 (indeed fails) and Rust 1.84 (indeed works).

Do you think it's worth bumping MSRV two versions for this? I guess it wouldn't solve the "warning fires on latest Rust but not MSRV" problem.

[nicoburns]

Do you think it's worth bumping MSRV two versions for this?

Probably not for this, given that there are other functions which still need to be polyfilled on latest Rust.

Although #424 proposes to bump MSRV beyond this point anyway. If it wasn't required for SIMD support, then I would campaign for keeping Parley's MSRV <= 1.85, because that's what Debian stable is currently shipping. But given that fearless_simd is already 1.86 and we may want to use it in Parley, then I'm inclined to allow MSRV bumps if people want them.

[PoignardAzur]

Alright, I'm changing the status to draft until we either bump the MSRV or find a better way to gate this warning.