The atan2 maths function does not seem to handle the singularity correctly.

The whole point of atan2(y,x) as opposed to atan(y/x), is to be able to give the correct angle (+/- pi/2) when x is zero, and the parameter to atan would become infinite.

However I find that the BCD(float) version (?atan2f) actually returns zero if the second parameter is zero, and the IEEE version (?atan2) returns NAN.

I have tested it over the specified range of -pi...pi by
y = sin(a);
x = cos(a);
b = atan2( y, x);
then comparing a with b. For all values other than +/- pi/2 it correctly returns the original angle, (except that pi returns -pi, which is understandable).

I am using Ride BN 746-51

regards

Tim Jackson