Implicit conversion from float -10.139664804469277 to int loses precision