Error handling with MonadPlus

Question

Got stuck again while trying to learn some Haskell. What I'm trying to do is implementing a combined head/tail method for lists with error handling. The signature must look like this:

head' :: MonadPlus m => [a] -> m (a,[a])

However I'm a bit lost at how error handling with MonadPlus works. I tried the following:

head' xs = if (length xs > 0) then Just(head xs, tail xs) else Nothing

but it complains: Expected type: m (a, [a]) Actual type: Maybe (a, [a]). Just for fun I also tried:

head' xs = if (length xs > 0) then Just(head xs, tail xs) else Nothing `mplus` Nothing

but not only does that look redundant, it also does not work either.

Any hints appreciated!

Answer 1

Try this:

head' :: MonadPlus m => [a] -> m (a,[a])
head' xs = if null xs then mzero then return (head xs, tail xs)

mzero is the "nothing" or (indeed) "zero" value, which you can use here to model the absence of a result. It has type m x for any type x and monad with zero m, so it fits here.

Instead, return wraps any value in any monad. In particular, you can insert our (a,[a]) pair inside m. (This does not even require the monad to be a MonadPlus)

Note how in the code above we did not mention any specific monad: the code is indeed generic, so that it can work with any MonadPlus.

Lastly, I used null instead of length ... > 0 since null only examines the first constructor in the list, while length ... > 0 has to traverse it all.

You can actually remove the check by switching to pattern matching:

head' :: MonadPlus m => [a] -> m (a,[a])
head' []     = mzero
head' (x:xs) = return (x, xs)