Implicit conversion from float -5.766299745977989 to int loses precision