Scala Introduction

Scala Introduction
Constantine Nosovsky

Agenda
 Motivation
 Imperative vs Functional
 What is Scala?
 Scala syntax intro
 Case classes, tail recursion
 Putting it all together
 Pure functional theory implementation
 Collections
 Advanced topics
 Resources

Motivation
 Top “Scala” queries for the past 12 months
Google trends (http://www.google.com/trends/)

Imperative vs Functional
 Imperative
– Mutable variables
– Assignments
– Control structures (if-else, loops, break, continue, return)
– Sequence of instructions executed one-by-one
– Strong correspondence to the computer components

Imperative vs Functional
 Functional
– Immutable variables
– Functions are treated as a first-class data types
– Production, consumption and composition of functions
– Corresponds to math theory where data, operations and relationships are
described, e.g. Matrix theory

Functional programming benefits
 Better programming abstraction: combine complex operations by
composing primitives
 Looser coupling: no side effects only input and output
parameters => better contract conformation
 Better scalability: immutable (therefore thread-safe) objects

What is Scala?
 Designed by Martin Odersky, 2003
 Multi-paradigm: Functional and Imperative
 Object oriented
– Even primitive types are objects
– Singleton objects instead of static methods and fields
 Platform: JVM, LLVM
 Compatible with Java

Scala syntax intro: value declaration
 Static strong inferred typing
val a: Int = 5
is similar to
val a = 5 // type will be inferred

Scala syntax intro: final
 Subsequent assignment are not allowed
val a: Int = 5
a = 10 // error: reassignment to val
var a: Int = 5
a = 10 // OK for var, avoid using vars!

Scala syntax intro: method, expression
 Expressions instead of statements
def abs(x: Int): Int =
if (x >= 0) x else –x
Or even
val result = {
if(f(x)) x else -x
} * 2

Scala syntax intro: high-order functions
 Functions might be passed as parameters
def map(xs: List[Int],
m: Int => Int): List[Int] =
if(xs.isEmpty) List()
else List(m(xs.head)) ++ map(xs.tail, m)
// pass function as a parameter
def square(x: Int) = x * x
map(List(1, 2, 3), square)
map(List(1, 2, 3), x => x * x)

Scala syntax intro: classes
class Class {
def method = println("Class")
}
new Class().method
// singleton object instead of static method
object Singleton {
def method = println("Singleton")
}
Singleton.method

Scala syntax intro: traits
 Trait is similar to interface, but may contain implementation
trait Drawable { def draw() }
trait Cowboy extends Drawable {
override def draw() = println("Bang!")
}
trait Artist extends Drawable {
override def draw() = println("A painting")
}
class CowboyArtist extends Cowboy with Artist
// "A painting" will be used

Scala syntax intro: operators
 Easy to build DSL
class Complex {
def add(y: Complex) = ???
def +(y: Complex) = add(y)
def unary_- = ???
}
val x = new Complex(); val y = new Complex()
x.add(y); x add y; x.+(y); x + y // equivalent
-x // unary operator

Case classes and pattern matching
// class applicable for pattern-matching
case class Complex(r: Int, i: Int)
// implicitly generated companion factory class
val v = Complex(1, 2); val v2 = Complex(2, 3)
// match object value and constructor arguments
v match {
case Complex(r, _) => println("Real " + r)
case null => println("No value")
}

Recursion
 Recursion is often used for list processing
def sum(xs: List[Int]): Int =
if(xs.isEmpty) 0
else xs.head + sum(xs.tail)
Problem: list of 1 million elements causes stack overflow

Tail recursion
 If a function calls itself as its last action, the function’s stack
frame can be reused
def sum(xs: List[Int]): Int = {
// will be compiled to a loop in byte code
@tailrec
def loop(acc: Int, xs: List[Int]): Int =
if(xs.isEmpty) 0
else loop(acc + xs.head, xs.tail)
loop(0, xs)
}

Putting it all together
 No mutable data sharing problem
– Variables are final by nature
– Classes are designed to be immutable
 Easy composition
– Everything is an expression returning result
– Mixing functions, methods and operators
– No side effects, i/o parameters only
– Closure
 Advanced type safety
– Inferred: automatic deduction of the data type
 OOP
– Singleton classes instead of static methods/fields
– Primitive types are objects as well
– Traits for abstraction and multiple inheritance
– Case classes and pattern matching

Pure functional theory implementation
 Build a Set theory
 Operations
– Check whether an element belongs to a set
– Union, Intersection, Difference
– Filter set elements with a predicate
 Sets are immutable, each modification produces a new set
 Functional way: no data structures, only rules

// Set is a function checking element presence
type Set = Int => Boolean
def contains(s: Set, elem: Int): Boolean =
s(elem)
// Returns the set of the one given element
def singletonSet(elem: Int): Set =
(x: Int) => x == elem

 Union, intersect, diff operations
def union(s: Set, t: Set): Set =
(x: Int) => s(x) || t(x)
def intersect(s: Set, t: Set): Set =
(x: Int) => s(x) && t(x)
def diff(s: Set, t: Set): Set =
(x: Int) => s(x) && !t(x)

 Filter elements
// Returns the subset of s for which p holds
def filter(s: Set, p: Int => Boolean): Set =
intersect(s, p)

 Usage example
val oddNumbers = (x: Int) => x % 2 == 1
val range = (x: Int) => x > 0 && x < 100
filter(range, oddNumbers)

Collections
 Mostly immutable data structures
 Modifications will create a new instance
 Optimized for language structures
– Functional way of processing and construction
– Pattern matching
– For expressions

Collections: List
 scala.collection.immutable.List
 Linked list implementation
 Case class, has a factory
val list = List(1, 2, 3)
 Prepending element will create a new list (shallow copy)
val newList = 0 :: list

Collections: List
 Methods head and tail used for one by one processing
def contains(l: List[Int], e: Int): Boolean =
if(l.isEmpty) false
else l.head == e || contains(l.tail, e)
 Applicable for pattern matching
def contains(l: List[Int], e: Int): Boolean =
l match {
case List() => false
case head :: tail =>
head == e || contains(tail, e)
}

Collections: Vector
 Faster access by index than linked list
 Supports almost similar operations
 Hierarchical data structure
– Array of 32 elements
– After all elements are filled becomes an array of 32 references to arrays of
elements
– Then next level, again and again
references
references … references
elements elements … elements

Collections: common methods
 exists, forall – contains by predicate
 zip, unzip – transform two collection into a single one
chaining corresponding items into a tuple
 map, flatMap – apply transformation to each element
 to[List, Map, Set, Stream]
 sum, product, max, min

Advanced topics
 OOP in Scala:
– Traits
– Generics, bounding
– Subtyping (Any, Nothing)
– Polymorphism
 Language structures
– For expressions
– Lazy evaluation
– Currying
– Implicit parameters
– Tuples
 Concurrency
 Scala frameworks: akka, play

Resources
 Coursera class by Martin Odersky
https://class.coursera.org/progfun-004
 Scala API documentation
http://www.scala-lang.org/api/current
 Programming in Scala
Martin Odersky, Lex Spoon, Bill Venners. 2nd ed. Artima 2010
 Scala tutorial by Twitter
http://twitter.github.io/scala_school/

Scala Introduction

More Related Content

What's hot

Similar to Scala Introduction

Recently uploaded

Scala Introduction