Has anyone used existential types to represent items on a schedule in a scheduled lazy data structure?

On Aug 11, 2012 4:15 AM, <oleg@okmij.org> wrote:

> data A = A deriving Show
> data B = B deriving Show
> data C = C deriving Show
>
> data Foo = forall a. Show a => MkFoo a (Int -> Bool)
>
> instance Show Foo where
>    show (MkFoo a f) = show a

I'd like to point out that the only operation we can do on the first
argument of MkFoo is to show to it. This is all we can ever do:
we have no idea of its type but we know we can show it and get a
String. Why not to apply show to start with (it won't be evaluated
until required anyway)? Therefore, the data type Foo above is in all
respects equivalent to

> data Foo = MkFoo String (Int -> Bool)

and no existentials are ever needed. The following article explains
elimination of existentials in more detail, touching upon the original
problem, of bringing different types into union.

        http://okmij.org/ftp/Computation/Existentials.html



_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe@haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe