Implicit conversion from float -42.170818505338076 to int loses precision