Standard says that implementation of math functions that have floating-point-type parameter should provide an "overload for each cv-unqualified floating-point type".
Standard says that implementation of math functions that have floating-point-type parameter should provide an "overload for each cv-unqualified floating-point type".