Implicit conversion from float -8.167539267015698 to int loses precision