Implicit conversion from float -42.6057298772169 to int loses precision