most-positive-float equals to DBL_MAX most-negative-float equals to -DBL_MAX least-positive-normalized-float equals to DBL_MIN least-negative-normalized-float equals to -DBL_MIN float-epsilon equals to DBL_EPSILON float-epsilon and float-negative-epsilon can be safely calculated in temacs. We cannot set least-positive-float and least-negative-float from <float.h> constants in the portable manner. The following C program do the similar thing to cl-float-limits in calculating least-positive-float, and can generate a C header file. -- SAKIYAMA Nobuo nobuo(a)db3.so-net.or.jp #include <stdio.h> #include <float.h> #include <signal.h> void dummy() { } main () { double x, y; signal(SIGFPE,SIG_IGN); x = DBL_MIN; while(x > 0.0){ y = x; dummy(); x/= 2.0; } printf("#define LEAST_POSITIVE_FLOAT %.*e\n",DBL_DIG, y); }