Sign check for NaN value

Question

My program during calculation can generate nan or -nan values. I check if the values are nan/-nan using isnan method.

I also have to distinguish if the nan value is positive or negative (nan or -nan). How can I do this?

Added:I need crossplatform solution for WIN and for Unix/Linux

Answer 1

Nearly all systems today use either IEEE single or double precision floating-point. So in that case you could (bitwise) convert it to an integer and read the sign-bit.

Here's one approach that uses unions. Although it's not fully standard-compliant, it should still work on nearly all systems.

union{
    double f;
    uint64_t i;
} x;

x.f = ... //  Your floating-point value (can be NaN)

//  Check the sign bit.
if ((x.i & 0x8000000000000000ull) == 0){
    //  positive
}else{
    //  negative
}

Answer 2

Try signbit from <math.h>:

Description

signbit() is a generic macro which can work on all real floating-point types. It returns a nonzero value if the value of x has its sign bit set.

...

NaNs and infinities have a sign bit.

It's apparently part of C99 and POSIX.1-2001, but you could write a macro/function yourself if you don't want to use/conform to either of the two.

Answer 3

You could use the copysign function (C99, in <math.h>);

double sign = copysign(1.0, your_nan);

From C99 §7.12.11.1:

Description

The copysign functions produce a value with the magnitude of x and the sign of y. They produce a NaN (with the sign of y) if x is a NaN. On implementations that represent a signed zero but do not treat negative zero consistently in arithmetic operations, the copysign functions regard the sign of zero as positive.

Returns

The copysign functions return a value with the magnitude of x and the sign of y.