Implicit conversion from float -40.96446700507616 to int loses precision