Implicit conversion from float -59.791376912378325 to int loses precision