2011/12/28 Jon Fairbairn <jon.fairbairn at cl.cam.ac.uk>: > * non-strict semantics require that no argument is evaluated > unless needed. That's not the case on optimistic evaluation.