Implicit conversion from float -9.05660377358491 to int loses precision