Implicit conversion from float -139.51807228915663 to int loses precision