Implicit conversion from float -17.02247191011236 to int loses precision