Hmm, what pipe was I smoking when I thought it would be possible to get a measure of consistency between 3 different readline implementations?