Implicit conversion from float -28.016528925619838 to int loses precision