Implicit conversion from float -125.2777777777778 to int loses precision