Implicit conversion from float -5.298913043478279 to int loses precision