As discussed in https://lists.isocpp.org/std-proposals/2026/04/17921.php, it would be helpful if there was a more detailed traits type than numeric_limits. This could be used to identify whether a floating-point type has binary32 format, and obtain all sorts of implementation properties such as the number of mantissa bits, exponent bits, etc. that are hard or impossible to obtain from numeric_limits.
This would have been pretty helpful in the implementation of std::big_int.
As discussed in https://lists.isocpp.org/std-proposals/2026/04/17921.php, it would be helpful if there was a more detailed traits type than
numeric_limits. This could be used to identify whether a floating-point type has binary32 format, and obtain all sorts of implementation properties such as the number of mantissa bits, exponent bits, etc. that are hard or impossible to obtain fromnumeric_limits.This would have been pretty helpful in the implementation of
std::big_int.