Denis Stoyanov xgrommx

DSl'и и как писать большие программы на PureScript'е

Вступление

Приветы! Меня зовут Максим Зималиев, я работаю в SlamData и уже почти два с половиной года пилю в продакшн код на PureScript'е. Как и большинство пришедших в прекрасный мир ФП, я начинал со всяких там джаваскриптов, пхп и шарпов. Соответственно, у меня есть бывшие коллеги и друзья, которые по-прежнему работают на императивных языках, используют GoF, DDD и тому подобное.

Ну так вот, пойдешь такой пиво пить, и внезапно "Слушай, а как у вас в PureScript'е, например ORM запилить, объектов же нет?" Или: "Это здорово всё, но вот я домены описываю вот так вот, у них там поведение есть, а как это PureScript'е сделать?"

ISeq (Stack Machine)

data Inst = IPlus | IMult | ICall String | IPush Int |
  ILt | INeg | IZeroJump Int | IJump Int | ILabel Int |
  ISetLocal String | IGetLocal String deriving (Show, Eq)

IPlus
IMult
ILt
Consumes 2 values and stores the result of plus/mult/less-than by the two values

Stephen Diehl (@smdiehl )

Since I wrote these slides for a little user group talk I gave two years ago they have become a surprisingly popular reference. I decided to actually turn them into a proper skimmable reference for intermediate and advanced level Haskell topics that don't necessarily have great coverage or that tend be somewhat opaque as to where to get going, and then aggregate a bunch of the best external resources for diving into those subjects with more depth. Hopefully it still captures the "no bullshit brain dump" style that seemed to be liked.

The source for all code is available here. If there are any errors or you think of a more illustrative example feel free to submit a pull request.

	test = map ((*) 2) >>> filter ((>) 15) >>> drop 3 >>> map show
	src1 = [10, 9, 8, 7, 6, 5, 4, 3, 2, 1]
	src2 = [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]

	res1 = transduce' test src1 :: [String]
	res2 = transduce' test src2 :: List String

	-- \| A “flushing” 'stream', with an additional coalgebra for flushing the
	-- remaining values after the input has been consumed. This also allows us to
	-- generalize the output away from lists.
	fstream
	:: (Cursive t (XNor a), Cursive u f, Corecursive u f, Traversable f)
	=> Coalgebra f b -> (b -> a -> b) -> Coalgebra f b -> b -> t -> u
	fstream ψ g ψ' = go
	where
	go c x =
	let fb = ψ c

	{-#LANGUAGE BangPatterns #-}
	import GHC.Magic
	import Data.IORef
	import Control.Monad
	import Control.Monad.Trans

	data Stream a m r = Yield a (Stream a m r) \| Done r \| Delay (() -> m (Stream a m r))

	instance Functor m => Functor (Stream a m) where
	fmap f (Done r) = Done (f r)

	{-#LANGUAGE DataKinds #-}
	{-#LANGUAGE DeriveDataTypeable #-}
	{-#LANGUAGE DeriveFoldable #-}
	{-#LANGUAGE DeriveFunctor #-}
	{-#LANGUAGE DeriveGeneric #-}
	{-#LANGUAGE DeriveTraversable #-}
	{-#LANGUAGE GADTs #-}
	{-#LANGUAGE GeneralizedNewtypeDeriving #-}
	{-#LANGUAGE KindSignatures #-}
	{-#LANGUAGE TypeFamilies #-}

	count:: Int -> [Int] -> Int
	count x arr = length $ filter (\i -> i == x) arr

	freq:: [Int] -> [Int] -> [(Int, Int)]
	freq [] _ = []
	freq (x:xs) acc = [(x, count x acc + 1)] ++ freq xs (x: acc)

	example:: [(Int, Int)]
	example = freq [1,2,1,2,4,2,1] []

	-- I haven't been able to actually run the program (broken archlinux packages), but it typechecks

	open import Data.String
	open import Function
	open import IO.Primitive

	Either : Set → Set → Set₁
	Either l r = {a : Set} → (l → a) → (r → a) → a

	Left : { l r : Set } → l → Either l r

	import Control.Applicative ((<$>))

	center :: String -> Int -> String
	center s n = spaces ++ s ++ spaces
	where spaces = replicate ((n - length s) `div` 2) ' '

	-- http://www.haskell.org/haskellwiki/Blow_your_mind, Ctrl-F "pascal"
	pascal :: [[Int]]
	pascal = iterate (\row -> zipWith (+) ([0] ++ row) (row ++ [0])) [1]