Principles of Functional Progrmming in Scala

• Moor’s Law

• Clock frequency

has not raised.

• More Complex

programs.

• Growing in data

non-determinism

Interference

Horizontal-scaling

among vertical-scaling

Cuncurrency and Parallelism

var x = 0

async{ x = x+1 }

async{ x = x*2 }

// the answer could be 0, 1 or 2

Non-determinism = parallel processing+ mutable values

Example

In science, a paradigm describes distinct concepts or

thought patterns in some scientific discipline.

Main programming paradigms:

imperative programming

functional programming

logic programming

Paradigm

modifying mutable variables,

using assignments

and control structures such as if-then-else, loops, break, continue, return.

Imparative paradigm

Immutable states

Functions without side-effects

Everything is function

Functions are expressions not statements

Fucntional Paradigm

once a variable is given a value it never changes.

no side effects are possible.

Functions are building blocks functions are free side-effect

function call does not affect other functions and it just computes its own output for a specific input

sin (pi) = 0

Functions are expressions not statements

Referential transparency

Neither theories nor functions have mutable values.

You can combine functions to make more complex functions

The result of a function or a theory does not change for different input.

(a+b)^2 = a^2 + b^2 + 2*a*b

This state doesn’t change for different inputs.

FP is inspired by mathematics

Pure(restricted) Functional languages ( no imperative structure) like: Pure Lisp, XSLT, XPath, XQuery, FP,Haskell (without I/O Monad or UnsafePerformIO)

Wider sense : Lisp, Scheme, Racket, ClojureSML, Ocaml, F# , Haskell (full language) , Scala , Smalltalk, Ruby

Functional Languages

Scala is a statically typed JVM language that has transparent interoperability with Java.

Both object oriented and functional language

What is scala?

Lightweight syntax

Combines functional and object-oriented approaches

Advanced type system: everything has a type

Strong type inference

Performance comparable to Java

Fully interoperable with Java

Why Scala?

Scala formclass person(val name: String , val age: Int) { ...}

Java formclass person{private String name;private int age;

Peron(String name, int age){

this.name = name;this.age = age;

}}

Example of Scala High-level syntax

val list = List(2,5,4)

val newList = (list map(x => x*x)) sortWith(_ > _)

//newList = list(4,16,25)

Java form ???

val people: Array[Person]

val (minors,adults) = people partition (_.age < 18)

Java form ???

continue

The idea underlying this model is that all evaluation does is reduce an expression to a value.

It can be applied to all expressions, as long as they have no side effects.

Lambda calculus

Subsitution Model

A non-primitive expression is evaluated as follows:

1. Take the leftmost operator

2. Evaluate its operands (left before right)

3. Apply the operator to the operands

(2 * pi) * radius

(2 * 3.14159) * radius

6.28318 * radius

6.28318 * 10

62.8318

Non-primitive expression

1. Evaluate all function arguments, from left to right

2. Replace the function application by the function’s right-hand side, and, at the same time

3. Replace the formal parameters of the function by the actual arguments

Parameterized functions substitution

1. sumOfSquares(3, 2+2)

2. sumOfSquares(3, 4)

3. square(3) + square(4)

4. 3 * 3 + square(4)

5. 9 + square(4)

6. 9 + 4 * 4

7. 9 + 16

8. 25

Example

What if right hand side does not terminate ?

two elements for storing expressions in Scala: def and val

def loop: Boolean = loop

def x = loop //it is ok

val x = loop // it will lead to an infinite loop.

As you see the difference between val and def becomes apparent when the right hand side does not terminate. Val is changed to its value when it is defined but the value of def is substituted where it is used.

Scala Substitution model

def:

1. can have parametersdef cube(x: Int): Int = x*x*x

2. Call-by-name

val:

1. The value of it is substituted as it is defined. In other words: call-by-value

val & def continue

sumOfSquares(x: Int , y: => Int)

sumOfSquares(3, 2+2)

square(3) + square(2+2)

3 * 3 + square(2+2)

9 + square(2+2)

9 + (2+2) * (2+2)

9 + 4 * (2+2)

9 + 4 * 4

25

Call-by-name function parameters

Functions are expressions so they can be:

Defined anywhere included in other functions

sent to other functions as parameters

Returned from other functions

def fact(x: Int) = if(x ==0) 1 else x*fact(x-1)

def cube(x: Int) = x*x

def processOnSums( f: Int => Int , a: Int , b: Int) = f(a) + f(b) processOnSums(cube, 2 , 3 ) // Output = 2*2 + 3*3

anonymous function: x=> x*x

processOnSums(fact , 2 , 3 ) // Output: 2*1 + 3*2*1

Higher-Order functions

def sum(f: Int => Int, a: Int, b: Int): Int = if (a > b) 0

else f(a) + sum(f, a + 1, b)

def sumInts(a: Int, b: Int) = sum(x=>x, a, b)

def sumCubes(a: Int, b: Int) = sum(x=> x*x, a, b)

def sumFactorials(a: Int, b: Int) = sum(x=>fact(x), a, b)

Can we make it even shorter???

Currying

def sum(f: Int => Int): (Int, Int) => Int = {def sumF(a: Int, b: Int): Int =if (a > b) 0else f(a) + sumF(a + 1, b)sumF}Sum returns another functionWe can define like this:def sumInts = sum(x => x)def sumCubes = sum(x => x * x * x)def sumFactorials = sum(fact)sumCubes(1, 10) + sumFactorials(10, 20)

Even shorter???

continue

The answer is yes.

sum (cube) (1, 10)

Write once, program many times. ..

continue

Two kinds of instant field initialization:Class myClass(input1: Int, input2: Int)

{

Val a = input1

Vab b = input2

}

class myClass(val input1: Int, val input2: Int){…}

Classes

In order to create auxiliary constructors, define method this with required parameters.

Class myClass(input1: Int, input2: Int)

{

def this(input1: Int) = this (input1, 0)

}

Auxiliary Constructors

Inheritance rules so similar to java

Traits are alternative for interfaces

Why do we need a new structure ?

In java classes can have only one super class, but what if a class need to inherit from several supertypes ???

This is why scala introduces traits.

Inheritance

trait super

{

val a = 5

def b = a*2

def c(x: Int) = a*x

}

class sub extends Object with super{…}

Trait

Class Hierarchies

Int scala.Int

Boolean scala.Boolean

Object java.lang.Object

require scala.Predef.require

assert scala.Predef.assert

Imports

The same as java. with new notationVal myList = new List(1,2,3)

The other form: Val myList = List(1,2,3)

What happened? Whre is new? We will see…

Object Definition

What is singleton object and why do we need it?

The reason we used singleton classes was to create just one object of a certain type.

Scala gives us a powerful tool called singleton object that ease the procedure we used to do with singleton classes in java.

Singletone Objects

object myObject

{

def +(x: java.lang.String) = x + " and "

}

def names = myObject + "Ehsan" + "Sana"

println(names)

//output Ehsan and Sana

Example

Still don’t know how compiler translates this:

Val myList = List(1,2,3)

We can create a new object just as we call a function with implenting apply method.

Reason: We are just trying to show the use of apply method as a handy way of closing the gap between functions and objects in Scala.

Apply Method

object List extends List

{

def apply[A](xs: A*): List[A]

//Creates a list with the specified elements.

}

Implementation of Apply Method for List

Lists

Maps

a fundamental structure in many functional languages is the immutable linked list.

Immutable Collections

List(List(1, 2)),List(3)

List Implementation

1. val city= List(“Mashhad”,”Tehran”,”Abadan)

2. val city = “Mashhad”::”Tehran”::”Abadan::Nil

3. val city = “Mashhad”::(”Tehran”::(“Abadan”::Nil)))

4. val city = Nil.::(“Abadan).::(“Tehran”).::(“Mashhad)”

continue

def sum(xs: List[Int]): Int = if (xs.isEmpty) 0

else xs.head + sum(xs.tail)

Sums the elements of the list

Example

val map = Map("Tehran" -> "Iran" , "Paris" -> "France")

println(map.keys) // output: Set(Tehran, Paris)

Does the String class has a method ->? The answer is no.

We will see how this is translated…

Immutable Maps

partial functions obey pattern matching rules and syntax.

Instead of switch case structure here we have match and case

wide range of usage in collections and actors

Partial function is a trait

and its Constructor takes two types; Input type and output type.

It has two abstract values too; apply(v1 a: A): B and isDefinedAt(x : A): Boolean

Partial Functions

val p = new PartialFunction[Int, Any] { def apply(xs: Int) = xs match { case xs => 1/xs.doubleValue() case _ => 1/xs} def isDefinedAt(xs: Int) = xs match { case 0 => false case _ => true } }println(p(2)) //output: 0.5 println(p(0))// output: Infinity No exception is thrown

Example

Threads in java: shared objects and lock model

Synchronized and concurrency library

always control threads to not to let them have access on a shared object at a time, using blocks

Result: race conditions, deadlocks and non-determinism.

Threads

Scala suggest shared-nothing, message-passing model.

object firstActor extends Actor {

def act() {

for (i <- 1 to 3) {

println("act" + i)

Thread.sleep(1000)

}

}

firstActor.start() //output: act1 act2 act3

Actors

If actors do not have shared objects how do they connect each other???

So simple: sending message…

Problem

val echoActor = actor {

while (true)

{

receive { case msg => println("received message: "+ msg) }

}

}

console: echoActor ! “hi”

output: hi

Receive Method

Partial function

Does not stay on a receive line.

Messages goes to inbox waiting for actor to call receive.

receive invoke IsDefinedAt method to make sure the input is valid.

If input is valid, receive sends the it’s body to apply method otherwise it throw an exception.

How receive works?

val reciever = actor {

receive {

case x: Int => // Only Int is valid as input

println("Got an Int: "+ x)

}

}

reciever ! “hello world” //output: Nothing

receiver ! 4 // output: Got an Int: 4

Example

Low performance implementation of imperative structures compared to java.

Disadvantage