haskell - State Monad with multiple state values -

consider following:

do   x1 <- new 2   set x1 3   x2 <- x1   y1 <- new 10   set y1 20   y2 <- y1   return (x2 + y2) 

i want result in 23. there way implement in pure haskell, , if how? understand stref this, want in ordinary haskell (not worried efficiency @ moment). presume i'll have make data type , make instance of monad, i'm not sure of details, working example helpful.

this allows more 1 value, it's hairier :) nicely simplified daniel's suggestion of dynamic.

import data.dynamic import data.maybe import control.monad.state import data.map m  newtype ref = ref {ref :: int}  type mutstate = state (int, map int dynamic)  val :: typeable => ref -> mutstate val r = snd `fmap` >>=          return . fromjust . (>>= fromdynamic) .  m.lookup (ref r)  new :: typeable => -> mutstate (ref a) new =   (curr, binds) <-   put (curr + 1, m.insert (curr + 1) (todyn a) binds)   return . ref $ curr + 1  set :: typeable => ref -> -> mutstate () set (ref i) =   (c, m) <-   put (c, m.insert (todyn a) m)  runmut :: mutstate -> runmut = flip evalstate (0, m.fromlist []) 

then use it

default (int) -- lazy signatures :) test :: int test = runmut $   x1 <- new 2   set x1 3   x2 <- val x1   y1 <- new 10   set y1 20   y2 <- val y1   return (x2 + y2) 

refs ints type information attached , val appropriate dynamic , attempt force correct type.

if real code, should hide implementations of ref , mutstate. convenience, i've fromjusted return of val bur if want safe implementation suppose layer state , maybe monads deal unbound variables.

and in case worried typeable constraints, shown above trivially derived.