Saturation, nonmonotonic reasoning and the closed-world assumption