How to stack the State and IO monads

Question

In a game of Connect4:

• we start with an empty grid
• two players place pieces x and o on the grid
• the first player to achieve 4 pieces in a line wins!
• this is a text based console game

My thinking is that:

• at each step of the game the Grid is transitioning from one state to another
• hence i need to use the State monad
• and because this is a console based app that involves io
• in this case I also need to use the IO monad

Is this thinking correct?

Assuming the above is correct, which one of these is correct?

• type StateInIO[S,A] = IO[State[S,A]]
• type IOInState[S,A] = State[S,IO[A]]

I favour the second option, makes more sense to me.

Can I stack these monads (State, IO) in this way?

Answer 1

I think you should just write a working program and generalize/refactor it afterwards instead of overingeneering upfront. You need State[S, A] for business logic and IO[A] for interacting with the console. But you don't have to mix it.

Then in the main class where you write a user-interaction loop you could leverage StateT which unifies both of your types.

By the way IO[State[S,A]] doesn't look like a very useful type. It says that you can read State[S, A] from the real world. State is a function. There are no sensible ways to read a function from the real world. So most probably you need something less powerful. Of course this type makes sense in the context of monad transformers.