Thank you! Feel free to add the assertions too!

That would require the check float >= 1.0, which is non trivial to do. Are you okay with adding a dependency on approx? Or do you know another solution

What's wrong with float <= 1.0?

Exact equality is not possible with floats, it won't even compile in rust since floats only implement PartialEq

Hmm... My compiler doesn't seem to complain about this:

assert!(0.0 >= r, r <= 1.0);
assert!(0.0 >= g, g <= 1.0);
assert!(0.0 >= b, b <= 1.0);
assert!(0.0 >= a, a <= 1.0);

I understand precision issues are a thing, but I am not sure if we should be worrying about that here. AFAIK, both 0.0 and 1.0 should be perfectly representable.

Oh that's weird? I have no idea why that is the case. Maybe 1.0 is special cased?

A type implementing only PartialEq still allows to check for equality. Eq is basically PartialEq but with reflexiveness (in the case of floats: NaN != NaN). Also, the relevant trait for <= is PartialOrd.

Should from_rgb_u32 check for values larger than 0xFFFFFF too? Or just ignore those?