Make Vec* methods const

[SUPERCILEX]

What problem does this solve or what need does it fill?

I'd like to use Vec2 in a static declaration, but that's currently not possible.

What solution would you like?

Make Vec* methods const.

[mockersf]

This is not possible as Vec2 takes floats, and floats can't be const...

However, there is a macro const_vec2 that does that 🙂

[alice-i-cecile]

Very reasonable. This is fundamentally blocked on float math being usable in const functions. Once that's in, I expect glam to rapidly migrate away from the mentioned macros.

[SUPERCILEX]

Floats can't be const? TIL, though I guess that makes sense if a machine doesn't adhere to 754.

[SUPERCILEX]

Actually, are you sure? This compiles just fine for me:

#[derive(Debug)]
struct Blah(f32, f32);

const BLAH: Blah = foo(1., 2.);

const fn foo(blah: f32, boo: f32) -> Blah {
    Blah(blah, boo)
}

pub fn use_it() {
    println!("{:?}", BLAH);
}

[SUPERCILEX]

Related issue: bitshifter/glam-rs#76

[alice-i-cecile]

const fn add_two() -> f32 {
  2. + 2.
}

This should fail to compile :( Here's the Rust issue: rust-lang/rust#57241

[SUPERCILEX]

Hmmm, all I'd want is Vec2::new. Would it be reasonable to only allow construction? Or is there some extra magic happening that would prevent that?

[alice-i-cecile]

That's ultimately a question for upstream. Maybe try it out in a glam fork?

[mockersf]

Hmmm, all I'd want is Vec2::new.

If you just want that in a const context, the macro should work for you?

[SUPERCILEX]

Good point, I guess I'll just use the macro though it seems weird to require that.

[alice-i-cecile]

Closing out, although once the linked glam issue is resolved I'll be thrilled to be able to migrate.