Implicit conversion from float -72.54180602006687 to int loses precision