Hi, have the maintainers of the standard library considered adding wrapper structs for f32/f64 that ensure the values are not nan, +inf, or -inf? Finite floats are implemented by a number of crates, but adding them to the standard library would allow Option<FiniteF32> and Option<FiniteF64> to occupy just 4 bytes and 8 bytes respectively, by using nan, +inf, or -inf as the None value. This is similar to how std::num::NonZero* use rustc_layout_scalar_valid_range_start to use 0 as the None value. Thanks!
5 Likes
Perhaps (Pre-?)Pre-RFC: Range-restricting wrappers for floating-point types is the closest to relevant. Searching for NonNan in particular gets a few more.
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.