Re: [Haskell-beginners] Return a Foldable instance

On Wed, May 4, 2011 at 1:22 PM, Chaddaï Fouché wrote:

On Wed, May 4, 2011 at 2:21 PM, Federico Mastellone wrote:

...

So, is it even possible to write a function with this type: getValues :: Foldable f => k -> MultiMap k v -> f v ???

It isn't since there's nothing in the Foldable class that allows you to build one (only to consume one, do a fold over it, no "unfold").

As it's impossible to build a function like this one below: getValues :: Foldable f => k -> MultiMap k v -> f v I know that it may be too difficult, but wouldn't be great to make this type illegal in Haskell ? You can't make a function that returns a class that has no method that builds an instance of that class unless another parameter of the function is of that class or allows you to build one instance of that class.

But since this signature is _not_ the one you really want (it says you will deliver any Foldable your caller asks for), this isn't a problem, what you really want is :

...

data ExFoldable a = forall f . (Foldable f) => ExFoldable (f a)

This will do the trick and it is better than maintaining an intermediate structure as I previously said. Thank you and Alexander for this tip.

...

instance Foldable ExFoldable where   foldr f z (Exfoldable xs) = foldr f z xs

getValues :: k -> MultiMap k v -> Exfoldable v getValues k mm = ExFoldable (yourLookupFunction mm k)

or something like that. With this signature you're really saying you will deliver a Foldable without telling which one, like in OO.

Now I'm not sure if this is such a good idea since as other have pointed, delivering a list is a pretty good and mostly simple option where the list itself is often optimized away. But if you really want it, Haskell can do it.

(You'll need to activate some extensions to allow this)

-- Jedaï

-- Federico Mastellone Computer Science Engineer - ITBA ".. there are two ways of constructing a software design: One way is to make it so simple that there are obviously no deficiencies, and the other way is to make it so complicated that there are no obvious deficiencies. The first method is far more difficult." Tony Hoare, 1980 ACM Turing Award Lecture.