Implicit conversion from float -42.663316582914575 to int loses precision