Implicit conversion from float -38.22141560798548 to int loses precision