Implicit conversion from float -8.112058465286225 to int loses precision