Implicit conversion from float -117.39549839228297 to int loses precision