Print to console and update Monad state in same function - haskell

Question

I want to define a function that expects an Int, prints an error in the console depending on the number (x) and then updates the State with Nothing.
How can I join those commands in one function?
Here's what I got:
type Env = [(Variable,Int)]
newtype StateError a = StateError { runStateError :: Env -> Maybe (a, Env) }
class Monad m => MonadError m where
throw :: Monad m => a -> m a
instance MonadError StateError where
throw x = StateError (\s -> Nothing)
But I can't figure out how to perform the IO side-effect and then the State update in the same function definition

Answer 1

No
A function in the state monad, such as a -> State s b, is a pure function (no IO) that happens to have an extra function argument s hidden though some handy plumbing.
You can't print to the console from the state monad.
However, Yes!
However! You can use a monad transformer to get both State and some underlying monad such as IO.
I'll provide an example using transformers instead of a custom monad and mtl as it appears you were using. With mtl you can use classes like MonadError to leverage a throw that works well with other libraries that use the mtl classes. On the other hand, if you are the end consumer of this transformer it is less important.
First we'll import the modules that give us MonadIO, StateT, MaybeT, and use newtype deriving so we don't have to type out the monad instance boilerplate:
{-# LANGUAGE GeneralizedNewtypeDeriving #-}
import qualified Control.Monad.Trans.State as S
import Control.Monad.IO.Class
import Control.Monad.Trans.Maybe
import Control.Monad.Trans
Just to be complete we'll spell out the types useful to your abstraction:
type Variable = String
type Env = [(Variable,Int)]
Now we can get to the interesting part - a Monad definition and functions for plumbing. The monad stack is StateT MaybeT IO:
newtype StateError a = StateError { unStateError :: S.StateT Env (MaybeT IO) a }
deriving (Monad, Applicative, Functor)
And we can run it by first unwrapping the newtype, then running the state, and finally the MaybeT:
run :: StateError a -> IO (Maybe (a, Env))
run = runMaybeT . flip S.runStateT [] . unStateError
Usually you'll write an army of functions that provide your monad abstraction. For this question it's just "update the state" and "print to stdout":
modify :: (Env -> Env) -> StateError ()
modify = StateError . S.modify
emit :: Show a => a -> StateError ()
emit = StateError . liftIO . print . show
Armed with our Monad of Power, we can do fancy things like update state and emit IO messages and track failure or success:
updateAndPrint :: Variable -> Int -> StateError ()
updateAndPrint v i =
do emit (v,i)
modify ((v,i):)
Oh, and failure is pretty simple - just fail in our MaybeT monad:
throw :: a -> StateError b
throw _ = fail "" -- same as 'MaybeT (pure Nothing)'
We can use this monad as expected:
> run $ updateAndPrint "var" 1
"(\"var\",1)"
Just (() -- ^ return value of `updateAndPrint`
,[("var",1)]) -- ^ resulting state