NSLondon 10 -- Introduction to Functional Programming / Haskell

FP? Haskell Side Eﬀects Monads Bonus End
Introduction to Functional
Programming / Haskell
Johannes Weiß
@johannesweiss
NSLondon 10
Johannes Weiß Introduction to Functional Programming / Haskell

View Slide

FP? Haskell Side Effects Monads Bonus End Definition
Definition of Functional Programming
[...] functional programming is a programming paradigm,
a style of building the structure and elements of
computer programs, that treats computation as the
evaluation of mathematical functions and avoids state
and mutable data.
Wikipedia 1
1https://en.wikipedia.org/wiki/Functional programming

View Slide

FP? Haskell Side Eﬀects Monads Bonus End Intro Type System Laziness Purity Currying
What is Haskell?
Haskell will serve as an example for a functional programming
language. But there are many others (e.g. Lisp, Erlang, Scala, F#,
ML, Closure, etc.)
Haskell is a computer programming language. In
particular, it is a polymorphically statically typed, lazy,
purely functional language, quite diﬀerent from most
other programming languages.
Haskell Wiki 2
2http://www.haskell.org/haskellwiki/Introduction

View Slide

Haskell in one slide (credits to Simon PJ)
filter :: (a -> Bool) -> [a] -> [a]
filter pred [] = []
filter pred (x:xs) -- (1 : (2 : (3: []))) == [1,2,3]
| pred x = x : filter pred xs
| otherwise = filter pred xs
Type
signature
Higher order
Polymorphism
(works for
any type a)
Function deﬁned
by pattern
matching
Guards
distinguish
sub–cases
f x y rather
that f(x,y)

View Slide

ﬁlter in Objective-C
NSArray *filter(BOOL(^pred)(id obj), NSArray *list)
{
NSMutableArray *objsPassingTest = [NSMutableArray
array];
for (id x in list) {
if (pred && pred(x)) {
[objsPassingTest addObject:x];
}
}
return [objsPassingTest copy];
}

View Slide

ﬁlter in Swift
func filter(pred: (T -> Bool), list: T[]) -> T[] {
var filtered : T[] = []
for x in list {
if pred(x) {
filtered += x;
}
}
return filtered;
}

View Slide

Static Types
Haskell’s type system is strongly and statically typed (so is Swift’s).
• Strong type system: Fine grained set of types (characters,
booleans, and integers are not the same)
• Static type system: types known at compile time
C is weakly but statically typed. Objective-C is weakly typed and a
hybrid between static and dynamic typing (type id can be
anything without even needing a cast).

View Slide

Polymorphism
The type system supports (multiple forms) of polymorphism:
• Parametric Polymorphism (similar to Java, .Net Generics)
-- Standard Functions
map :: (a -> b) -> [a] -> [b]
length :: [a] -> Int
odd :: Integral a => a -> Bool
-- 1: map as (String -> Int) -> [String] -> [Int]
strLens = map length ["C", "Objective-C", "C++"]
strLens = [1, 11, 3] -- the result
-- 2: map as (Int -> Bool) -> [Int] -> [Bool]
oddInts = map odd [1, 2, 3, 4, 5]
oddInts = [True,False,True,False,True] -- the result

View Slide

Polymorphism
• Ad–hoc Polymorphism (similar to Operator Overloading)
-- (==) :: Eq a => a -> a -> Bool
okIntegers :: Bool
okIntegers = 1 == 2 -- is False
okStrings :: Bool
okStrings = "foo" == "foo" -- is True
compileTimeError = "foo" == True
{- Couldn’t match expected type ‘[Char]’
with actual type ‘Bool’
In the second argument of ‘(==)’, namely ‘True’
In the expression: "foo" == True
In an equation for ‘compileTimeError’:
compileTimeError = "foo" == True
-}

View Slide

Type Classes
class Eq a where
(==) :: a -> a -> Bool
(/=) :: a -> a -> Bool -- not strictly needed
data Maybe a = Just a | Nothing
-- example: Just "Hello World!" :: Maybe String
instance Eq a => Eq (Maybe a) where
mL == mR =
case (mL, mR) of
(Just l , Just r ) -> l == r
(Nothing, Nothing) -> True
_ -> False

View Slide

Maybe in Swift
enum Maybe {
case Just(A)
case Nothing
}
func ==(mL:Maybe,mR:Maybe)->Bool {
switch ((mL, mR)) {
case (.Just(let l), .Just(let r)):
return l == r
case (.Nothing, .Nothing):
return true
default:
return false
}
}

View Slide

Laziness
Haskell is a language which does lazy evaluation. This means: An
expression is evaluated when some other computation needs the
value.
{- important to know: The cons (:) operator:
[1, 2, 3] == 1 : 2 : 3 : []
Standard Library:
take :: Int -> [a] -> [a]
-}
endlessFrom :: Integer -> [Integer]
endlessFrom n = n : endlessFrom (n+1)
first5 :: [Integer]
first5 = take 5 (endlessFrom 1)
first5 = [1,2,3,4,5] -- the result

View Slide

Purity
• Pure computations yield the same value each time they are
invoked.
• No actions (often called side effects) allowed!
• That allows laziness (can be evaluated at any time)
• Consequences:
• No I/O (user input, random values, etc.) in pure computations
• No state
• No variables (writing to variables is a side effect)
• No side effects
• Advantages
• Concurrency — no problem with pure computations
• Referential transparency gives room for compiler optimisations
• Types speak: Same function parameters ⇒ same return value

View Slide

Currying
-- Standard Library
-- foldl :: (a -> b -> a) -> a -> [b] -> a
-- | Adds four numbers.
sum4 :: Num a => a -> a -> a -> a -> a
sum4 a b c d = foldl (+) 0 [a, b, c, d]
-- | Adds three numbers.
sum3 :: Num a => a -> a -> a -> a
sum3 = sum4 0
In fact, all Haskell functions have one parameter!
sum4 1 2 3 4 == ((((sum4 1) 2) 3) 4) == 10
Prelude> :type sum4
sum4 :: Num b => b -> b -> b -> b -> b
Prelude> :type sum4 1
sum4 1 :: Num b => b -> b -> b -> b
Prelude> :type sum4 1 2
sum4 1 2 :: Num b => b -> b -> b
Prelude> :type sum4 1 2 3
sum4 1 2 3 :: Num b => b -> b
Prelude> :type sum4 1 2 3 4
sum4 1 2 3 4 :: Num b => b

View Slide

FP? Haskell Side Eﬀects Monads Bonus End Real–world Intro Example
Real–World Programs
That looks nice but how to write real–world programs?
Real programs need side eﬀects!
(Otherwise it’s pointless to even run them)

View Slide

Side Effects
There are many different kinds of side effects:
• Global side effects (such as I/O)
• Local side effects (such as reading and writing to local
variables)

View Slide

Side Eﬀects
First idea (like in most languages):
putStr :: String -> ()
-- like void putStr(NSString *)
But, that would mean filter can do arbitrary things as well, e.g.
filterBad :: (a -> Bool) -> [a] -> [a]
filterBad pred [] = []
filterBad pred (x:xs)
| pred x = x : filter pred xs
| otherwise = launchTheMissiles
And what does the following mean?
[putStr "foo", putStr "bar"]
Keep in mind: order of evaluation, laziness!

View Slide

Fun
$ xcrun swift
Welcome to Swift! Type :help for assistance.
1> :version
lldb-320.3.100
1> let xs = [println("foo"), println("bar")]
Segmentation fault: 11 (core dumped)

View Slide

YO!

View Slide

The main idea
A value of type IO t is an “action” that, when performed, may do
some input/output before delivering a result of type t.
putStr :: String -> IO ()
• An action is a first class value
• Evaluating an action has no effect; performing the action has
an effect
• Approximation: type IO a = World -> (a, World) i.e.
putStr :: String -> World -> ((), World)

View Slide

Somewhat Real World Program
-- getLine :: IO String
-- putStr :: String -> IO ()
-- main is the entry point of a Haskell program
main :: IO ()
main =
do putStr "Hey there, what’s your name? "
name <- getLine
putStr ("Hello " ++ name ++ "!\n")
• The do–notation looks deliberately imperative.

View Slide

FP? Haskell Side Eﬀects Monads Bonus End Intro What? STM
Special Case I/O?
• So did Haskell just special case I/O operations with the do
notation? No!
• In fact, the do notation is syntactical sugar for monadic
computations.

View Slide

What is a Monad?
A monad is deﬁned by the following type class.
class Monad m where
-- | Sequentially compose two actions, passing
-- any value produced by the first as an
-- argument to the second. (>>= aka "bind")
(>>=) :: m a -> (a -> m b) -> m b
-- | Inject a value into the monadic type.
return :: a -> m a
instance Monad IO where
[...]

View Slide

What’s that all about?
No worries, monads are an abstract concept which looks
complicated at ﬁrst. Bear with me for some real world examples.
For now, one of the monad analogies are enough. I’d go for 1 or 2.
1 Warm, fuzzy things (Simon PJ “feels that the term Monad is
far too imposing”)
2 Programmable semicolons
3 Burritos, space suits, ... (the infamous Monad tutorials)
Examples for Monads
• State — local side efects (like local variables)
• Maybe (like Swift’s optional chaining)
• Parsers (parser combinators)
• I/O

View Slide

Tackling concurrency with STM (Software
Transactional Memory)
atomically :: STM a -> IO a
newTVar :: a -> STM (TVar a) -- new tx var
readTVar :: TVar a -> STM a -- read contents
writeTVar :: TVar a -> a -> STM () -- write contents

View Slide

A classic example: Accounting
type Account = TVar Int
deposit :: Account -> Int -> STM ()
deposit k amount =
do bal <- readTVar k
writeTVar k (bal + amount)
withdraw :: Account -> Int -> STM ()
withdraw k amount = deposit k (- amount)
transfer :: Account -> Account -> Int -> IO ()
transfer k1 k2 amount =
atomically (do deposit k2 amount
withdraw k1 amount)

View Slide

Composability of STM
splitTransfer :: Account -> Account -> Account
-> Int -> IO ()
splitTransfer k1 k2 k3 amount =
atomically $
do withdraw k1 (2 * amount)
deposit k2 amount
deposit k3 amount

View Slide

Bonus Material :-)
Here some bonus material. Added that after the talk because some
nice guy made me aware I totally missed adding some Haskell
resources. Thanks for that! The books are free to read online btw.
• Book: Learn You a Haskell for a Great Good 3
• Book: Real World Haskell 4
• FPComplete’s School of Haskell 5
• NSLondon Demo Project (runnable in the browser) 6
3http://learnyouahaskell.com/
4http://book.realworldhaskell.org/
5https://www.fpcomplete.com/school
6https://www.fpcomplete.com/user/johannesweiss/adopted/nslondon10demo

View Slide

Thank you! Questions?

View Slide

NSLondon 10 -- Introduction to Functional Programming / Haskell

NSLondon 10 -- Introduction to Functional Programming / Haskell

Johannes Weiss

More Decks by Johannes Weiss

Other Decks in Programming

Featured

Transcript