Haskell is seriously packed with modern, interesting, expressive goodies.
But I have a message for the Haskell community: stop trying to “simplify” monads. Really, just stop. I have yet to see an “explanation” of monads that does not add to the confusion and fear a novice might feel at approaching the language. You don’t need to know what a monad is in order to use the language effectively. Talking about them to a beginner is about as useful as explaining loop unrolling and keyhole optimization to a beginning C programmer. Later, yes, it’s helpful. But it’s too much information at the start. A beginner just needs to know that if you want to use code that does IO, it has to go inside a
do, and if you want to pull out values and use them elsewhere, you need the
<- operator. It took me three nights of working on this one to learn that one, because the Gentle Introduction relegated I/O to chapter 7.
On that note, you’ve got to stop treating I/O in general as a red-headed stepchild. I realize that isolating side effects is one of the raisons d’être for Haskell in the first place, but if you hope to win converts from the programming world at large (and I really hope you do) then you’ve got to understand that most of us mere mortals want our programs to do something, and that means input/output. I understand, or at least I think I do, that you want to instill good habits from the start and minimize time spend in the IO monad, but honestly, the language is pretty good at doing that already. You’ve got to go and talk to people where they already are before you can lead them somewhere else.
Once peple get used to IO and Maybe and whatever other monads are out there, then you can introduce the concept because there’s somewhere for it to land. I remember taking CS theory in college, and talking with one of the TFs who I happened to know socially. He is an extremely bright guy, and as is sometimes common with bright guys, can have trouble explaining a concept to someone who doesn’t get it themselves right away. I made plea for more concrete examples to go with the theoretical explanations in the class, and his response was, “Well, we assume that you, the smart [name of school] students, can come up with the examples yourselves.” This is exactly wrong for many many people out there, including a lot of the folks who are attracted to coding in the first place. Some folks work well top-down — get the overarching concepts and then fill in the details themselves. But some people learn the other way around — experience the details, and then build up the framework from there. My impression is that the Haskell community is filled with the former type of learner. But it’s a great language, even for the more experiential/practical and less theoretical, and I hate to see it shunned as too scary.
Ranting aside, I really enjoyed writing this. For those of you who haven’t tried it yet, I highly recommend it. It will make you smarter. I stumbled on places where it differs from OCaml, and I got stuck on IO for a while, but I’m glad I persevered.
This was written against GHC 6.10.4.
import Data.List (intersperse) import System.IO (stdout, hFlush) import System.Exit (exitSuccess) -- -- Users -- data User = X | O deriving(Show, Eq) otherUser X = O otherUser O = X -- -- Squares -- data Square = Move User | Empty Int deriving(Eq) instance Show Square where show (Move x) = " " ++ show x ++ " " show (Empty x) = "(" ++ show x ++ ")" filled (Move _) = True filled _ = False -- -- Boards -- data Board = Board [[Square]] deriving(Eq) instance Show Board where show (Board ls) = "\n" ++ concat (intersperse "---+---+---\n" $ map showLine ls) ++ "\n" where showLine xs = concat (intersperse "|" $ map show xs) ++ "\n" full (Board squares) = all filled (concat squares) -- -- Results -- data Result = Continue User Board | Win User Board | Draw Board instance Show Result where show (Continue user board) = show board ++ "Select a square, " ++ show user ++ ": " show (Win user board) = show board ++ show user ++ " Wins!\n" show (Draw board) = show board ++ "It's a Draw!\n" -- -- Initial Board -- startingBoard :: Board startingBoard = Board [[Empty 1,Empty 2,Empty 3], [Empty 4,Empty 5,Empty 6], [Empty 7,Empty 8,Empty 9]] matchSquare :: User -> Int -> Square -> Square matchSquare user position (Empty x) | x == position = Move user matchSquare _ _ square = square -- This evaluates the board and determines the result of the current user's action outcome :: User -> Board -> Result outcome user board = case board of (Board [[a, _, _], [_, b, _], [_, _, c]]) | eq a b c -> Win user board (Board [[_, _, a], [_, b, _], [c, _, _]]) | eq a b c -> Win user board (Board [[a, b, c], [_, _, _], [_, _, _]]) | eq a b c -> Win user board (Board [[_, _, _], [a, b, c], [_, _, _]]) | eq a b c -> Win user board (Board [[_, _, _], [_, _, _], [a, b, c]]) | eq a b c -> Win user board (Board [[a, _, _], [b, _, _], [c, _, _]]) | eq a b c -> Win user board (Board [[_, a, _], [_, b, _], [_, c, _]]) | eq a b c -> Win user board (Board [[_, _, a], [_, _, b], [_, _, c]]) | eq a b c -> Win user board _ | full board -> Draw board | otherwise -> Continue (otherUser user) board where eq a b c = a == b && b == c move :: User -> Int -> Board -> Result move user pos board = result user pos board (place user pos board) where place user pos (Board lines) = Board $ map (placeInLine user pos) lines placeInLine user pos = map $ matchSquare user pos result user pos orig board | orig == board = Continue user orig | otherwise = outcome user board main :: IO () main = loop (Continue X startingBoard) where loop result = do putStr $ show result hFlush stdout case result of Win user board -> exitSuccess Draw board -> exitSuccess Continue user board -> getInput user board getInput user board = do pos <- readLn::IO Int loop (move user pos board)