Implicit conversion from float -112.96208530805686 to int loses precision