Testing for NaN

[Rudy Ray Moore]

How do I test a float for NaN (or other weird garbage) without messing up my
FPU or throwing exceptions or other bizarro things?

I'm using VS6

Thanks,

Rudy

===

float f;

SetFloatToFunnyValue(&f);

if( IsFunnyValue(f) )
{
f = 0.0f;
}

 

[Carl Daniel [VC++ MVP]]

See _isnan() in the VC++ documentation. It's delcared in <float.h>

-cd

 

[Sean Cavanaugh]

How do I test a float for NaN (or other weird garbage) without messing up my
FPU or throwing exceptions or other bizarro things?

I'm using VS6

Thanks,

Rudy

===

float f;

SetFloatToFunnyValue(&f);

if( IsFunnyValue(f) )
{
f = 0.0f;
}

_finite()
_isnan()
_fpclass()

Google around for IEEE 754 will get some interesting links if you want
to do stuff manually with some lower level code (bitwise integer
operations, or asm or something)

http://research.microsoft.com/~hollasch/cgindex/coding/ieeefloat.html