Implicit conversion from float -9.139072847682115 to int loses precision