Implicit conversion from float -100.05563282336581 to int loses precision