Implicit conversion from float -35.242718446601955 to int loses precision