Implicit conversion from float -42.58575197889181 to int loses precision