Ruby is beautiful

class Cycle
    def cycle(*args)
      args[((@_cycles ||= Hash.new(-1))[args.hash] += 1) % args.size]
    end
end

More beautiful code plees?

What >>22 is trying to say is that this is pointless to even try in Haskell. The type system forces you to either write a special monad specifically for the purpose of calling cycle[/cycle], or fuck around with the IO monad so at least you can intersperse it with IO calls.

Oh, and all the arguments have to be of the same type....Actually, it's even worse than that. Since you need a way to compare argument lists to each other, the type signature should say [code](Eq a) => stuff or (Ord a) => stuff, which means that a can't be a function type, which means you can't even use a list like [print 4, print 2]. So basically this (toy) function is more useless than ever in Haskell.

There is a way around this, but it means using separate functions for setting and getting the argument list (and still using a special monad). If you have setArgs :: [a] -> State [a] () and getArg :: State [a] a, the implementation is trivial. The catch is that you can only access the last arglist you set with setArgs, and if you've got different types of arglists you want to use, you have to actually thread them through different instances of State [a].