Haskell State Monad使用修改表示

Question

I need to express put of the State Monad with use of its modify function however I'm getting a 'Could not deduce' error on my following code:

class Monad m => MonadState m s | m -> s where
    get :: m s
    get = modify (\s -> s)

    put :: s -> m ()
    put = modify $ const ()

    modify :: (s -> s) -> m s
    modify f = undefined  --do { x <- get; put (f x) }

(*I did not yet define modify because my implementation in the comment beside it also returned a similar error but then on an expected return type ms instead of the actual return type m () )

The exact error I'm getting is:

Could not deduce (MonadState m ()) arising from a use of 'modify'
from context (MonadState m s)
  bound by the class declaration for 'MonadState'
  at <..>
Possible fix: add an instance declaration for (MonadState m ())
In the expression: modify
In the expression: modify $ const ()
In an equation for 'put': put = modify const ()

However I'm not sure what this error is trying to tell me and what I'm doing wrong here. If anyone could help me out it'd be much appreciated!

Best regards, Skyfe.

Answer 1

What const () would do is transform the current state from whatever it is to () . Since you want this to work on all states, this won't work since it means that it would be changing the state type from s to () . It also wouldn't quite do what you want it to do. You can use the const function, but it'd be better to write it without first. I'd suggest making it pointful and using an explicit lambda:

put :: s -> m ()
put newState = modify (\currentState -> ???)

As a side note, your commented implementation of modify won't work, you'd really need

modify f = do
    x <- get
    let newX = f x
    put x
    return x

However, I would recommend leaving modify out of the class definition, as mtl does, since it's really more desirable to just implement get and put . This helps you prevent circular definitions.