Implicit conversion from float -112.71226415094338 to int loses precision