Implicit conversion from float -44.03819918144612 to int loses precision