Implicit conversion from float -89.22465208747514 to int loses precision