Slide 1

Slide 1 text

The Joys of Functional Programming Stephen Mizell

Slide 2

Slide 2 text

First, Some Caveats 1. I do not do 100% functional programming 2. I will be showing examples in Javascript 3. I do not currently use functional reactive programming

Slide 3

Slide 3 text

What is Functional Programming? 1. Imperative vs. Declarative 2. Persistent and Immutable Data 3. No Side Effects 4. Noun vs. Verb 5. First class and higher order functions 6. Recursive over iterative

Slide 4

Slide 4 text

Common Functions 1. Map (transform to some) 2. Filter 3. Reduce

Slide 5

Slide 5 text

Recursive vs. Iterative

Slide 6

Slide 6 text

Iterative Factorial var num = 5, total = 1; if (num !== 0) { for (i = 1; i <= num; i++) { console.log(i); total = total * i; } } console.log(total);

Slide 7

Slide 7 text

Recursive Factorial var num = 5; function factorial(n) { if (n === 0) return 1; return n * factorial(n - 1); } console.log(factorial(num));

Slide 8

Slide 8 text

Imperative vs. Declarative Imperative you express how to do it, declarative you express what you want done

Slide 9

Slide 9 text

Imperative var list = [1, 2, 3, 4, 5], total = 0; for (i = 0; i < list.length; i++) { total += list[i]; }

Slide 10

Slide 10 text

Declarative (Part 1) var list = [1, 2, 3, 4, 5]; var total = list.reduce(function(result, num) { return result + num; });

Slide 11

Slide 11 text

Declarative (Part 2) var list = [1, 2, 3, 4, 5]; function add(a, b) { return a + b; } var total = list.reduce(add);

Slide 12

Slide 12 text

Declarative (Part 3) var list = [1, 2, 3, 4, 5]; function add(a, b) { return a + b; } function sum(items) { return items.reduce(add); } var total = sum(list);

Slide 13

Slide 13 text

Using a Functional Language Clojure (def items `(1 2 3 4 5)) (reduce + items)

Slide 14

Slide 14 text

One More Example!

Slide 15

Slide 15 text

Here's Some Data var players = [ { totalPoints: 2031, games: 30 }, { totalPoints: 4268, games: 45 }, { totalPoints: 2235, games: 24 }, { totalPoints: 1221, games: 22 }, { totalPoints: 5420, games: 40 } ];

Slide 16

Slide 16 text

Requirements 1. Get a list of the average points for each player 2. Get an average for all players who have played 30 or more games

Slide 17

Slide 17 text

Imperative var playerAverages = [], allPoints = 0, allGames = 0; for (i = 0; i < players.length; i++) { var average = players[i].totalPoints / players[i].games; playerAverages.push(average); if (players[i].games >= 30) { allPoints += players[i].totalPoints; allGames += players[i].games; } } var totalAverage = allPoints / allGames;

Slide 18

Slide 18 text

Declarative (Part 1) var playerAverages = players.map(function(player) { return player.totalPoints / player.games; }); var players30Games = players.filter(function(player) { return player.games >= 30 }); var points = players30Games.map(function(player) { return player.totalPoints; }); var games = players30Games.map(function(player) { return player.games; }); var totalPoints = sum(points), totalGames = sum(games); totalAverage = totalPoints / totalGames;

Slide 19

Slide 19 text

Declarative (Part 2) function pluck(items, prop) { return items.map(function(item) { return item[prop]; }); } function average(a, b) { return a / b; } var playerAverages = players.map(function(player) { return player.totalPoints / player.games; }); var players30Games = players.filter(function(player) { return player.games >= 30 }); var totalPoints = sum(pluck(players30Games, 'totalPoints')), totalGames = sum(pluck(players30Games, 'games')); totalAverage = average(totalPoints, totalGames);

Slide 20

Slide 20 text

Currying Currying is the technique of translating the evaluation of a function that takes multiple arguments (or a tuple of arguments) into evaluating a sequence of functions, each with a single argument (partial application). –Wikipedia

Slide 21

Slide 21 text

Currying function add(a) { return function(b) { return a + b; } } var add10 = add(10); console.log(add10(5)); // 15

Slide 22

Slide 22 text

Currying (Better Example) var fullName = _.curry(function(first, middle, last) { return [first, middle, last].join(' '); }); var johnA = fullName('John', 'A.'); var johnASmith = johnA('Smith'); // John A. Smith

Slide 23

Slide 23 text

Currying (From Player Data Example) // Uses a library called Ramda var players30Games = R.filter(R.pipe(R.get('games'), R.lte(30))); var totalPoints = R.pipe(players30Games, R.pluck('totalPoints'), R.sum); totalPoints(players)

Slide 24

Slide 24 text

How is declarative helpful? 1. Testable, maintainable, evolvable code 2. Writing code at a higher level of abstraction 3. Through composition, it can lead to a very powerful way of thinking (like SQL)

Slide 25

Slide 25 text

Persistent

Slide 26

Slide 26 text

Persistent State 1. Values are immutable 2. A "change" in a value returns a new value (old is never thrown away) 3. These values persist (not necessarily on disk, but throughout time) 4. Big plus: fast!

Slide 27

Slide 27 text

Crude Example A -> B -> C Sharing all of A Z -> A -> B -> C Sharing tail of A Y -> B -> C

Slide 28

Slide 28 text

Persistence 1. Shared memory usage 2. No need to do deep clones of objects to avoid changing state 3. Allows for immutable data 4. Allows for understanding what has changed 5. Allows for no side effects

Slide 29

Slide 29 text

Side Effects

Slide 30

Slide 30 text

Side Effects var x = 10; function a() { x = 4; } console.log(x);

Slide 31

Slide 31 text

Side Effects var x = 10; function a() { x = 4; } a(); console.log(x);

Slide 32

Slide 32 text

Mutable Data

Slide 33

Slide 33 text

Mutating Data x = 4; // ...later x = 10; y = x + 10;

Slide 34

Slide 34 text

Mutating Data x = x + 1;

Slide 35

Slide 35 text

Why is persistence and immutability good? 1. We've probably all spent hours chasing down bugs from side effects 2. Code that mutates data requires following steps through code to understand what it is doing 3. Mutating lots of code can be expensive

Slide 36

Slide 36 text

Libraries for Javascript 1. Underscore.js and Lo-Dash.js provide some functions 2. Ramda - Function-first library 3. Mori - Immutable data structures in Javascript 4. React.js and Immutable.js 5. ClojureScript

Slide 37

Slide 37 text

Recap, why is this helpful? 1. Code is more declarative 2. Promotes reusability and evolvability 3. Easier to write automated tests 4. Fewer bugs

Slide 38

Slide 38 text

Bonus Functional Reactive Programming

Slide 39

Slide 39 text

FRP There is no simple definition of this on the internet, so here are some qualities. 1. Deal with events in a functional way 2. Create streams of events rather than event listeners 3. Free of side effects

Slide 40

Slide 40 text

Some Libraries 1. Bacon.js 2. Elm 3. RxJS

Slide 41

Slide 41 text

Bacon.js function movieSearch(query) { if (query.length < 3) // show no results for queries of length < 3 return Bacon.once([]); return Bacon.fromPromise(queryMovie(query)); } var text = $('#input') .asEventStream('keydown') .debounce(300) .map(function(event) { return event.target.value; }) .skipDuplicates(); // Only react to latest, in case they are out of order var suggestions = text.flatMapLatest(movieSearch); text.awaiting(suggestions).onValue(function(x) { if (x) $('#results').html('Searching...'); }); suggestions.onValue(function(results) { $('#results').html($.map(results, showMovie)); });