The original version of Exercise B.ii asked you to use PyMonad's State Monad to implement RPN. In fact, RPN is used in LearnYouAHaskell to solve the problem of errors rather than a state problem. So just use MayBe (again).