C++ check input floats (-0.0f, +0.0f)

Question

I am currently trying to build a program which prints an input float value to binary.

The first bit is 0 if positive or 1 if negative, but with an input value of e.g.: -0.0g my if statement always prints 1, also for a positive input. How would I check that correctly?

string sign = "sign: ";
if(value <= -0.0f)
 sign.append("1\n");
else if (value >= 0.0f)
 sign.append("0\n");

...

Simple method, use the [signbit](https://en.cppreference.com/w/cpp/numeric/math/signbit) function. — john, Nov 22 '22 at 12:40
[`std::signbit` on cppreference](https://en.cppreference.com/w/cpp/numeric/math/signbit) — kotatsuyaki, Nov 22 '22 at 12:40
@ChrisMM: More importantly, in IEEE754 positive and negative zero are _equal_. That's why the code fails `+0.0f <= -0.0f` is true, because `+0.0f == -0.0f` — MSalters, Nov 22 '22 at 13:59

chux - Reinstate Monica · Accepted Answer · 2022-11-22T15:13:58.583

+0.0 and -0.0 have the same value, yet different signs.

value <= -0.0f is true for value as +0.0 and -0.0. @MSalters

To distinguish the sign, use std::signbit(). @john

if(std::signbit(value)) {
  sign.append("1\n");
} else {
  sign.append("0\n");
}

Note that signbit() also applies to infinities and NaN.

Even though std::signbit reports "Along with std::copysign, std::signbit is one of the only two portable ways to examine the sign of a NaN.", there may be others.

C++ check input floats (-0.0f, +0.0f)

1 Answers1