Implicit conversion from float -30.815450643776813 to int loses precision