Nurul Choudry, a prolog guru chum of mine, used to say that there was no point in his resolution system maintaining the stack of partially resolved results as prolog should be able to deduce backwards; I think he was a bit optimistic, but I sort of understood. -- Robin Becker