Implicit conversion from float -71.1441307578009 to int loses precision