Safe Haskell | None |
---|---|

Language | Haskell98 |

# Documentation

The datatype of Keys.

- x
- The scope of this key. This can either be
`T`

for top-level keys created with`createKey`

or an existential type for keys introduced by`withKey`

(or with the Key monad`KeyM`

). - a
- The type of things that can be sorted at this key.

For example, `Key T Int`

is a top-level key that can be used to store values
of type `Int`

in a heterogenous map.

withKey :: (forall x. HKey x a -> b) -> b Source

*O(1)*. Scopes a key to the given function
The key cannot escape the function (because of the existential type).

The implementation actually *creates* a key, but because the key cannot escape
the given function `f`

, there is no way to observe that if we run
`withKey f`

twice, that it will get a different key the second time.

# Key Monad

type KeyM s a = KeyT s Identity a Source

A monad that can be used to create keys
Keys cannot escape the monad, analogous to the ST Monad.
Can be used instead of the `withKey`

function if you
need an statically unknown number of keys.

The applicative instance is more non-strict than
the standard `ap`

:

let hang = getKey >> hang
in snd $ runIdentity $ runKeyT $ pure (,) * hang * (getKey >> return 2)
does not hang, but with `ap`

it does.

keyTSplit :: KeyT s m a -> KeyT s m (m a) Source

Split of a keyT computation.

As an analogy, think of a random number generator some random number generator can be split, from one random number generator we obtain two distinct random number generator that are unrelated.

The KeyT monad gives us access to a name source, this operation allows us to split the name source. The generated name from both this and the split off computation have the same scope, but are otherwise underlated.

Notice that the sharing of the same scope is not a problem because the monad ensures referential transparency.