[Haskell-cafe] Re: Re[2]: [Haskell] Proposal: unification of styleof function/data/type/class definitions

[Brian Hulley]

Taral wrote:
> On 9/11/06, Bulat Ziganshin <bulat.ziganshin at gmail.com> wrote:
>>       Expr Bool = Eq (Expr t) (Expr t) | forall t . Eq t
>
> Still confusing, but less so.
>
> The problem is that it's really backwards. The symbol being defined
> is "Eq".
> Eq (Expr t) (Expr t) | Eq t = Expr Bool
>
> but that doesn't fit well, does it? In my mind, the current mode is a
> functional one:
>
> Eq :: Eq t => Expr t -> Expr t -> Expr Bool
>
> And this is perfectly readable. The arrow means that the argument is
> an implicit propositional one, but it's still an argument.

Perhaps the proposed new style of function signature could also be used in a 
GADT eg:

    -- There is no need for the keyword "Data" because we know it's part of
    -- a GADT because the function name is a conid - all such decls in a 
module would
    -- then be gathered together to form the GADT(s)

    If    :: Expr Bool -> Expr t -> Expr t -> Expr t
    Eq  :: Expr t -> Expr t -> Expr Bool | Eq t
    Lit  :: Int -> Expr Int
or
    data
            Expr t         = If (Expr Bool) (Expr t) (Expr t)
            Expr Bool   = Eq (Expr t) (Expr t) | Eq t
            Expr Int      = Lit Int

The second option is of course backwards compared to a function declaration, 
but this is natural because there can only ever be one clause for each value 
constructor but mutiple clauses per result type, whereas in a function decl 
there can be multiple clauses for each pattern but only one result type, 
therefore function decls are inherently backwards wrt data decls...

Regards, Brian.
-- 
Logic empowers us and Love gives us purpose.
Yet still phantoms restless for eras long past,
congealed in the present in unthought forms,
strive mightily unseen to destroy us.

http://www.metamilk.com