Implicit conversion from float -77.8502879078695 to int loses precision