Rich, The nan's and inf's show up when the ComplexPower() routine in mpmath_c.c is called. This is the routine that gets called by dStkPwr() in parser.c. I added a printf() statement to show the inputs and outputs. We have a line in unixscr.c that looks like this: signal(SIGFPE, fpe_handler); The fpe_handler() routine in general.c just sets overflow to 1. This doesn't appear to be working either. The value of overflow in the while loop in StandardFractal() in calcfrac.c is always equal to 0 when running this formula. Jonathan
The DOS version must be catching the math errors, setting the result to a default value, and then continuing with the calculation. This removes the nan's and inf's I'm seeing in Xfractint and allows the calculation to continue until the bailout value is reached.
Is it the formula evaluation that results in nan's and inf's?