sven just pointed me to some surprising bug:
create a = (Value).
observe that the output is 1 (as 0 equals 0)
now drag one of the input pins to something like 4 and drag it back to 0.000
observe that the output is 0 (as if 0 not equals 0)
when opening the xml code of the patch you will notice that the input value is now something like -1.00267016911459E-015
i agree that its great that values with a higher precision than the typical 4 digits can be used in vvvv and can be entered into a pin. i also can understand that incrementing and decrementing floating point values needs to accumulate rounding errors.
but i think if you scroll the values with the mouse, the internal value should become exactly the same as the displayed value (although almost no numbers can exactly be represented as floating point numbers, 0 actually can)