date and time created 89/08/15 11:53:45 by bostic
No VAX G_floating support was included in the "float.h" I just sent you.
All constants rely heavily upon a correctly-rounded decimal-binary
conversio
date and time created 89/08/15 11:53:45 by bostic
No VAX G_floating support was included in the "float.h" I just sent you.
All constants rely heavily upon a correctly-rounded decimal-binary
conversion routine; at least the following 6 constants *must* be converted
exactly to what the comments say they ought to be:
#define FLT_EPSILON 1.19209290E-07F /* 2^-23 */
#define FLT_MIN 2.93873588E-39F /* 2^-128 */
#define FLT_MAX 1.70141173E+38F /* (1-2^-24)*2^127 */
#define DBL_EPSILON 2.775557561562891351E-17 /* 2^-55 */
#define DBL_MIN 2.938735877055718770E-39 /* 2^-128 */
#define DBL_MAX 1.701411834604692294E+38 /* (1-2^-56)*2^127 */
DBL_MAX/FLT_MAX may generate spurious floating-point overflows if your
decimal-binary conversion routine is deficient; likewise DBL_MIN/FLT_MIN
may underflow inadvertently to zero after conversion.
The constants of type float rely on the ANSI-invented suffix of "F" to
precisely specify their type.
show more ...
|