Personal tools

Generalised algebraic datatype

From HaskellWiki

(Difference between revisions)
Jump to: navigation, search
(Adding more links to papers on GADTs)
m (:@ constructor symbol -- infix also in data declaration)
Line 17: Line 17:
 
Without GADTs we would have to write something like this:
 
Without GADTs we would have to write something like this:
 
<haskell>
 
<haskell>
data Term = K | S | :@ Term Term
+
data Term = K | S | Term :@ Term
 
infixl 6 :@
 
infixl 6 :@
 
</haskell>
 
</haskell>

Revision as of 09:27, 2 May 2006

1 Motivating example

Generalised Algebraic Datatypes (GADTs) are datatypes for which a constructor has a non standard type. Indeed, in type systems incorporating GADTs, there are very few restrictions on the type that the data constructors can take. To show you how this could be useful, we will implement an evaluator for the typed SK calculus. Note that the K combinator is operationally similar to \lambda\;x\;y\;.\;x and, similarly, S is similar to the combinator \lambda\;x\;y\;z\;.\;x\;z\;(\;y\;z\;) which, in simply typed lambda calculus, have types a \rightarrow b \rightarrow a and (a \rightarrow b \rightarrow c) \rightarrow (a \rightarrow b) \rightarrow a \rightarrow c Without GADTs we would have to write something like this:

data Term = K | S | Term :@ Term 
infixl 6 :@

With GADTs, however, we can have the terms carry around more type information and create more interesting terms, like so:

data Term x where
    K :: Term (a -> b -> a)
    S :: Term ((a -> b -> c)  -> (a -> b) -> a -> c)
    Const :: a -> Term a
    (:@) :: Term (a -> b) -> (Term a) -> Term b
infixl 6 :@

now we can write a small step evaluator:

eval::Term a -> Term a
eval (K :@ x :@ y) = x
eval (S :@ x :@ y :@ z) = x :@ z :@ (y :@ z)
eval x = x

Since the types of the so-called object language, being the typed SK calculus, are mimicked by the type system in our meta language, being haskell, we have a pretty convincing argument that the evaluator won't mangle our types. We say that typing is preserved under evaluation (preservation.) Note that this is an argument and not a proof.

This, however, comes at a price: let's see what happens when you try to convert strings into our object language:

parse "K" = K
parse "S" = S

you'll get a nasty error like so:

   Occurs check: cannot construct the infinite type: c = b -> c
     Expected type: Term ((a -> b -> c) -> (a -> b) -> a -> b -> c)
     Inferred type: Term ((a -> b -> c) -> (a -> b) -> a -> c)
   In the definition of `foo': foo "S" = S

One could, however, reason that parse has type: String -> exists a. Term a

2 Example with lists

here's another, smaller example:

data Empty
data NonEmpty
data List x y where
     Nil :: List a Empty
     Cons:: a -> List a b ->  List a NonEmpty
 
safeHead:: List x NonEmpty -> x
safeHead (Cons a b) = a

now safeHead can only be applied to non empty lists, and will never evaluate to bottom. This too comes at a cost; consider the function:

silly 0 = Nil
silly 1 = Cons 1 Nil

yields an objection from ghc:

Couldn't match `Empty' against `NonEmpty'
     Expected type: List a Empty
     Inferred type: List a NonEmpty
   In the application `Cons 1 Nil'
   In the definition of `silly': silly 1 = Cons 1 Nil