Implicit conversion from float -40.27777777777777 to int loses precision