Implicit conversion from float -23.55337078651685 to int loses precision