Implicit conversion from float -6.79442508710801 to int loses precision