On computational geometry