Implicit conversion from float -59.01315789473685 to int loses precision