/* floating-point mod function without infinity or NaN checking */
#include <math.h>
double
fmod (double x, double y)
{
int sign = 0, yexp;
double r, yfr;
if (y == 0)
return 0;
if (y < 0)
y = -y;
yfr = frexp (y, &yexp);
if (x < 0) {
sign = 1;
r = -x;
} else
r = x;
while (r >= y) {
int rexp;
double rfr = frexp (r, &rexp);
r -= ldexp (y, rexp - yexp - (rfr < yfr));
}
if (sign)
r = -r;
return r;
}
|