Implicit conversion from float -1.1320754716981298 to int loses precision