Login | Register   
LinkedIn
Google+
Twitter
RSS Feed
Download our iPhone app
TODAY'S HEADLINES  |   ARTICLE ARCHIVE  |   FORUMS  |   TIP BANK
Browse DevX
Sign up for e-mail newsletters from DevX


Tip of the Day
Language: C++
Expertise: Intermediate
Oct 2, 2009

Portable Functions to Check for NaN and Infinity

Currently, there are no portable C/C++ functions to check for NaN (not a number) and infinity (positive or negative).

Unix provides isnan() and isinf(), while Microsoft's implementation uses _isnan() and _isinf(). Moreover, some C/C++ compilers do not provide either of these functions.

There are many requests on the web to provide compatible functions that are available on all platforms. However, it is not difficult to write these functions from scratch, using the documented properties of these two special values.

For example, NaN is the only "number" that is not equal to any other number, including itself. Here's how you can use that property to construct a new function that tests for NaN:

inline int my_isnan(double x)
{
   return x != x;
}

You can write a similar function to test for infinity, differentiating between positive and negative infinities by comparing them to zero:

inline int my_isinf(double x)
{
   if ((x == x) && ((x - x) != 0.0)) return (x < 0.0 ? -1 : 1);
   else return 0;
}

There are other ways to write these tests; for example, the sum of an infinity and a finite value is still an infinity.

To help avoid compiler optimization issues, you can change these functions as follows:

inline int my_isnan1(double x)
{
   volatile double temp = x;
   return temp != x;
}

inline int my_isinf1(double x)
{
   volatile double temp = x;
   if ((temp == x) && ((temp - x) != 0.0))
      return (x < 0.0 ? -1 : 1);
   else return 0;
}
Alexander Ocher
 
Comment and Contribute

 

 

 

 

 


(Maximum characters: 1200). You have 1200 characters left.

 

 

Sitemap