Implicit conversion from float -88.67036011080333 to int loses precision