Implicit conversion from float -100.32033426183841 to int loses precision