[Haskell] Barnsley Fern

[jiegillet]

Seeing as #704 was reviewed recently, I'm pushing forward with Barnsley Fern.

I also fixed a typo in the chapter: 022 -> 0.22.

Output for 100k points:

[NikulinVs]

{-
​
Here is alternative code with IO monad instead of explicit Random Generator.
You do random generation in sequence anyway (through g, g' and g''),
so why not let IO monad care about the sequential stuff for you?
Then you won't even need newStdGen in main function, you work in IO already.
​
(This is the code without taking into account my other comments)
​

​
chaosGame :: Int -> [Double] -> Array Int (Point -> Point) -> IO [Point]
chaosGame n probabilities hutchinson = do
x <- randomIO
y <- randomIO
​
picks <- replicateM n $ randomRIO (0, 1)
let cumulProbabilities = scanl1 (+) probabilities
to_choice x = (+ 1) $ length $ takeWhile (x >) cumulProbabilities
​
points = Point x y : zipWith (hutchinson !) (map to_choice picks) points
return points
-}

[jiegillet]

I usually try to stay away from IO until the very end. I feel code is more portable when it's pure. Yes, it's true that I'm doing IO in the end, but I want to focus more on the algorithm, and less on the task...

[leios]

What do you think? Is this close to mergable?

[jiegillet]

@leios I probed Seva a couple of times, but I guess he was too busy to check. He told before me that the initial code was already fine, and I've integrated all of his suggestions but one (with a valid reason) so I think this can be approved and merged, honestly. By you probably :)
Sorry, looks like all the Haskell PRs are irregular somehow.

[leios]

Ok, I thought that was what happened. Happy to merge it then!