Implicit conversion from float -119.41558441558442 to int loses precision