OrderedFloat comparison with f32/f64 is not total

[orlp]

Is it intended that OrderedFloat(x) == y where x, y are float types is not a total comparison? This only matters for NaN: OrderedFloat(f32::NAN) == OrderedFloat(f32::NAN) is true, but OrderedFloat(f32::NAN) == f32::NAN is false.

On another note, why is PartialEq<Rhs=T> for OrderedFloat<T> implemented, but the same does not hold true for PartialOrd<Rhs=T> for OrderedFloat<T>?

[mbrubeck]

Is it intended that OrderedFloat(x) == y where x, y are float types is not a total comparison? This only matters for NaN: OrderedFloat(f32::NAN) == OrderedFloat(f32::NAN) is true, but OrderedFloat(f32::NAN) == f32::NAN is false.

No, I don’t think this is intended; as far as I know we just never thought about which result would be better.

On another note, why is PartialEq<Rhs=T> for OrderedFloat<T> implemented, but the same does not hold true for PartialOrd<Rhs=T> for OrderedFloat<T>?

Also just an oversight, apparently.