[Haskell-cafe] Trouble understanding records and existential types

[Brandon S. Allbery KF8NH]

On Jan 25, 2007, at 2:08 , John Ky wrote:

> On 1/25/07, Brandon S. Allbery KF8NH <allbery at ece.cmu.edu> wrote:
> data ANode = Branch { name :: String, description :: String,
> children :: [AnyNode] }
>             | Leaf { name :: String, value :: String } -- this reuse
> is legal
> -- leaving Node available if you still need it
>
> Would I be able to this?
>
>    getLeaves :: ANode -> [Leaf]

Leaf is a data constructor, not a type.  Your second one:

>    getLeaves :: ANode -> [ANode]

is correct.  If you want the type system to ensure they are only  
leaves, then indeed you can't use this method.

> (b) I think you *can* do this with a class:
>
> class Node a where
>    name :: a -> String
>
> data Branch = Branch { brName :: String, ... }
> data Leaf = Leaf { lName :: String, ... }
>
> instance Node Branch where
>    name = brName
>
> instance Node Leaf where
>    name = lName
>
> Okay, though it's a lot more wordy.

How so?  You were declaring the class and instances anyway; I simply  
defined a new method to go into it and renamed the constructor fields  
to obey Haskell's rules, but you will probably be using the class  
method so your code won't care about the latter.

-- 
brandon s. allbery    [linux,solaris,freebsd,perl]     allbery at kf8nh.com
system administrator [openafs,heimdal,too many hats] allbery at ece.cmu.edu
electrical and computer engineering, carnegie mellon university    KF8NH