# Difference between revisions of "Scoped type variables"

(how to avoid scoped type variables) |
(simplified sizeOfPtr) |
||

Line 50: | Line 50: | ||

E.g. if you want to determine the size of an object a pointer points to, |
E.g. if you want to determine the size of an object a pointer points to, |
||

then you might define a function like |
then you might define a function like |
||

+ | <haskell> |
||

+ | sizeOfPtr :: Ptr a -> Int |
||

+ | sizeOfPtr = sizeOf . (undefined :: Ptr a -> a) |
||

+ | </haskell> |
||

+ | <!-- provided by Lennart Augustsson in http://www.haskell.org/pipermail/haskell-cafe/2010-January/072060.html --> |
||

+ | or |
||

<haskell> |
<haskell> |
||

sizeOfPtr :: Ptr a -> a -> Int |
sizeOfPtr :: Ptr a -> a -> Int |
||

sizeOfPtr _ a = sizeOf a |
sizeOfPtr _ a = sizeOf a |
||

+ | |||

+ | sizeOf :: Ptr a -> Int |
||

⚫ | |||

. |
. |
||

</haskell> |
</haskell> |
||

⚫ | |||

== See also == |
== See also == |

## Revision as of 17:26, 14 January 2010

Scoped Type Variables are an extension to Haskell's type system that allow free type variables to be re-used in the scope of a function. They are also described in the GHC documentation.

As an example, consider the following functions:

```
{-# LANGUAGE ScopedTypeVariables #-}
...
mkpair1 :: forall a b. a -> b -> (a,b)
mkpair1 aa bb = (ida aa, bb)
where
ida :: a -> a -- This refers to a in the function's type signature
ida = id
mkpair2 :: forall a b. a -> b -> (a,b)
mkpair2 aa bb = (ida aa, bb)
where
ida :: b -> b -- Illegal, because refers to b in type signature
ida = id
mkpair3 :: a -> b -> (a,b)
mkpair3 aa bb = (ida aa, bb)
where
ida :: b -> b -- Legal, because b is now a free variable
ida = id
```

Scoped type variables make it possible to specify the particular type of a function in situations where it is not otherwise possible, which can in turn help avoid problems with the Monomorphism restriction.

This feature should be better documented in the Wiki, but this is a start.

## Avoiding Scoped Type Variables

Although Scoped Type Variables are often a simple solution, they are not available in all compilers. Often there is a solution that is Haskell 98. First, there is

```
asTypeOf :: a -> a -> a
asTypeOf a b = a
.
```

It is used like `x `asTypeOf` y`

and has the same value like `x`

,
but type inference asserts that `x`

and `y`

have the same type.

Sometimes it helps to divide a big function into smaller ones and give each of the small functions a signature. This also helps reading the program.

If this does not help, too, then use a helper function. E.g. if you want to determine the size of an object a pointer points to, then you might define a function like

```
sizeOfPtr :: Ptr a -> Int
sizeOfPtr = sizeOf . (undefined :: Ptr a -> a)
```

or

```
sizeOfPtr :: Ptr a -> a -> Int
sizeOfPtr _ a = sizeOf a
sizeOf :: Ptr a -> Int
sizeOf ptr = sizeOfPtr ptr undefined
.
```

## See also

- Haskell-Cafe on sizeOf on a type
- Haskell-Cafe on What is a rigid type variable?
- Haskell-Cafe on asserting the type of a binding in a "do" expression