Implicit conversion from float -117.78688524590163 to int loses precision