Implicit conversion from float -189.92805755395682 to int loses precision