On 8 Jan 2008, at 3:38 PM, Achim Schneider wrote:
---$<--- or, while I'm at it ---$<---
moveBall :: State' -> State' moveBall state = state {ballPos' = (bpx+bvx, bpy+bvy)} where (bpx, bpy) = ballPos' state (bvx, bvy) = ballVel' state
idle'' :: StateRef -> IdleCallback idle'' st = st $~ moveBall
---$<---
With the multiple IORef-Model, moveBall looks like this:
moveBall :: Vec2 -> Vec2 -> Vec2 moveBall (bpx, bpy) (bvx,bvy) = (bpx+bvx, bpy+bvy)
You can use this with the single IORef model, using the lifting function liftMove :: (Vec2 -> Vec2 -> Vec2) -> IORef State -> IO () liftMove move r = withIORef r $ \ st -> st{ballPos = moveBall (ballPos st) (ballVel st) } liftMove and moveBall can then be maintained separately; liftMove is part of your state framework (the outer layer of your program); moveBall is part of the algorithm specification (the inner layer of your program). jcc