Implicit conversion from float -43.04794520547944 to int loses precision