Implicit conversion from float -89.1549295774648 to int loses precision