Implicit conversion from float -56.89490445859872 to int loses precision