Implicit conversion from float -9.579831932773118 to int loses precision