Implicit conversion from float -125.15267175572518 to int loses precision