Implicit conversion from float -90.72289156626505 to int loses precision