Implicit conversion from float -42.76877761413846 to int loses precision