[Haskell-cafe] Re: Help understanding type error

[Benjamin Franksen]

Derek Elkins wrote:
> On Sat, 2007-09-08 at 12:24 +1000, Stuart Cook wrote:
>> On 9/8/07, Ryan Ingram <ryani.spam at gmail.com> wrote:
>> > This does what you want, I think:
>> >
>> > {-# LANGUAGE ExistentialQuantification #-}
>> > module Exist where
>> >
>> > data Showable = forall a. (Show a) => Showable a
>> > instance Show Showable where
>> >    showsPrec p (Showable a) = showsPrec p a
>> >    show (Showable a) = show a
>> >    -- You have to use the default implementation of showList
>> >    -- because a list could be heterogeneous
>> >
>> > data T a = forall b. (Show b) => T b a
>> >
>> > extShow :: T a -> Showable
>> > extShow (T b _) = Showable b
>> 
>> Wow, I'm impressed! Making the existential wrapper an instance of its
>> own typeclass solves quite a few problems.
>> 
>> While the idiom is obvious in hindsight, I don't think I've seen it
>> documented before. (Looking around just now I found some oblique
>> references to the technique, but nothing that really called attention
>> to itself.)
> 
> It's documented on the old wiki... and in the papers that introduce
> local existential types I believe.

E.g. http://citeseer.ist.psu.edu/aufer95type.html, this is where I first
read about it.

Cheers
Ben