Re: IEEE's NaN names

Paul Dechering <paul@cp.tn.tudelft.nl> wrote:
>Does anybody know the list of those IEEE floating point numbers that are
>NaN (Not a Number)? We like to get a standard name for the smallest
>positive float as well as the largest positive float. Any suggestions are
>welcome.


As our moderator notes, the IEEE standard doesn't name them, but I've
seen the functions


double infinity(void);
double max_normal(void);
double max_subnormal(void);
double min_normal(void);
double min_subnormal(void);
double quiet_nan(long);
double signaling_nan(long);


on several systems, most notably suns. There are also corresponding
functions for floats and long doubles.


The argument to quiet_nan() and signaling_nan() allows user data to
be embedded in the NaN that can be extracted by error handling
routines.


--------
richard henderson
richard@viz.tamu.edu
--

Return to the comp.compilers page.
Search the comp.compilers archives again.