From ES6 to Scala: Basics

This is a short introduction to the Scala language for those familiar with JavaScript ES6. We are comparing to ES6 instead of earlier versions of JavaScript because ES6 contains many nice features and syntax changes that bring it closer to Scala.

The best way to experiment with Scala is to use a Scala REPL, or the worksheet functionality in the Scala IDE or IntelliJ IDEA.

For more reading check out Scala Exercises, Scala School and official Scala Tutorials.

The Scala language

Scala is a modern multi-paradigm programming language designed to express common programming patterns in a concise, elegant, and type-safe way. It smoothly integrates features of object-oriented and functional languages. Scala is a pure object-oriented language in the sense that every value is an object. It is also a functional language in the sense that every function is a value and that immutability is favored over mutability.

This combination of paradigms is not alien to JavaScript developers. JavaScript is also object-oriented, although primitive values are not considered as objects. It is at least partially functional as well, since functions are values and can be passed around to other functions (such as Array.prototype.map). However, although it is possible to write with an immutable mindset in JavaScript, the language does not encourage immutability.

The biggest difference to JavaScript is that Scala is statically typed. This means that it is equipped with an expressive type system that enforces statically that abstractions are used in a safe and coherent manner, meaning the compiler will catch many typical programming errors. If you have used other statically typed languages like Java or C# before you may have noticed that type definitions are all over the place. This is not true with Scala where the compiler can infer most of the types automatically.

Variables

Let’s start with something simple, variables. Both Scala and ES6 support mutable and immutable variables.

ES6
// mutable variable
let x = 5;
// immutable variable
const y = "Constant";
Scala
// mutable variable
var x = 5
// immutable variable
val y = "Constant"

Note that the Scala compiler automatically infers the types for x and y from the values that are assigned. In Scala both mutable and immutable variables must always be initialized when declared.

Primitive types

Scala defines several primitive types, of which most have corresponding types in JavaScript as well. The following table presents the most commonly used ones (there are also Byte, Short and Float).

Scala typeJavaScript typeNotes
Stringstring
Booleanboolean
Intnumberinteger, range (-2147483648, 2147483647)
Doublenumber64-bit floating point, fully equivalent to JS number
LongN/A64-bit integer
CharN/AUTF-16 code unit
Unitundefined
Nullnull

In JavaScript all numbers are represented as 64-bit floating point internally, which may give surprising results when making some calculations. In Scala calculations are always performed using the types of operands, so dividing an Int with another Int, the result is rounded to an Int.

ES6
const x = 5 / 3; // == 1.6666666666666667
Scala
val x = 5 / 3 // == 1
val y = 5.0 / 3 // == 1.6666666666666667
val z = 5 / 3.0 // == 1.6666666666666667

Because in JavaScript every number is a number there is no need to do type conversions. In Scala, however, it is an error if you try to assign a higher precision value to a lower precision variable. You must explicitly convert it using an appropriate function.

Scala
val x: Double = 3 // Ok!
val y: Int = 3.5 // Compile error
val z: Int = 3.5.toInt // Ok!
val a: Int = x // Compile error
val b: Int = x.toInt // Ok!

Functions

Defining functions is quite similar in both languages. You just replace the function keyword with def, and add types for the parameters and the result type. Speaking of which, this is about the only place you will explicitly write types in Scala. The types of local values inside functions are usually inferred by the compiler. Actually, even the result type will usually be inferred too, but it is good practice to explicitly define it, to help catch potential type errors.

Note that there is no need for a return keyword: the last expression in the function is automatically used as the return value.

ES6
function mult(x, y) {
  return x * y;
}
Scala
def mult(x: Double, y: Double): Double = x * y

Anonymous functions

In functional programming you quite often need to provide a function as a parameter, but you don’t need it elsewhere so it can be anonymous. Both languages support the nice “fat arrow” notation for defining anonymous functions conveniently.

ES6
const f = (x, y) => x + y;

const p = ["Fox", "jumped", "over", "me"];
const l = p.map(s => s.length)
  .reduce((a, b) => a + b, 0); // == 15
Scala
val f = (x: Double, y: Double) => x + y

val p = Array("Fox", "jumped", "over", "me")
val l = p.map(s => s.length)
  .foldLeft(0)((a, b) => a + b) // == 15

Default, named and rest parameters

You can also define default values for parameters if they are not supplied when the function is called. For variable number of parameters, you can access those as a Seq (see the collections part of this tour for more info). Named parameters work just as you would expect in Scala, whereas in ES6 you need to supply them with the object notation.

ES6
// default value
function mult(x, y = 42.0) {
  return x * y;
}

// variable number of parameters
function sum(...args) {
  return args.reduce((a, b) => a + b, 0);
}

const s = sum(5, 4, 3, 2, 1); // == 15

// named parameters
function vec({x = 0, y = 0, z = 0}) {
  return new Vec(x, y, z);
}

const v = vec({x: 8, z: 42}); // Vec(8, 0, 42)
Scala
// default value
def mult(x: Double, y: Double = 42.0): Double =
  x * y


// variable number of parameters
def sum(args: Double*): Double =
  args.foldLeft(0.0)((a, b) => a + b)


val s = sum(5, 4, 3, 2, 1) // == 15

// named parameters (works directly)
def vec(x: Int = 0, y: Int = 0, z: Int = 0): Vec =
  new Vec(x, y, z)


val v = vec(8, z = 42) // Vec(8, 0, 42)

Again, the Scala compiler can infer all the required types in the code above, including the parameters for the anonymous function given to the foldLeft function.

if, while, for, match control structures

As you would expect, Scala has the regular if-else and while control structures found in most programming languages. The big difference to JavaScript is that if statements are actually expressions returning a value. In JavaScript you have the special a ? b : c construct to achieve the same result.

ES6
const res = (name === "") ? 0 : 1;
Scala
val res = if (name == "") 0 else 1

The for construct in Scala is quite different from the for-loop in JavaScript and also much more powerful. You can use it to iterate over numerical ranges or collections in both languages:

ES6
let x = 0;
for (let i = 0; i < 100; i++)
  x += i * i;

const p = ["Fox", "jumped", "over", "me"];
for (let s of p) {
  console.log(`Word ${s}`);
}
Scala
var x = 0
for (i <- 0 until 100)
  x += i * i

val p = Array("Fox", "jumped", "over", "me")
for (s <- p) {
  println(s"Word $s")
}

In case you have nested for-loops, you can easily combine them into one for-comprehension in Scala. Inside the for you can even filter using if expressions. In Scala a for-comprehension is just syntactic sugar for a series of flatMap, map and withFilter calls making it very handy when dealing with Scala collections.

ES6
function findPairs(n, sum) {
  for (let i = 0; i < n; i++) {
    for (let j = i; j < n; j++) {
      if (i + j == sum)
        console.log(`Found pair ${i}, ${j}`);
    }
  }
};
findPairs(20, 31);
Scala
def findPairs(n: Int, sum: Int): Unit = {
  for {
    i <- 0 until n
    j <- i until n if i + j == sum
  } println(s"Found pair $i, $j")
}
findPairs(20, 31)

The code above also serves as an example for string interpolation (in Scala) and template strings (in ES6). Both make it easier to construct strings using variables or function calls. In Scala you don’t need to enclose the variable in {} if it’s just a simple variable name. For more complex cases you’ll need to use the s"Length = ${data.length}" syntax.

Finally the match construct provides pattern matching capabilities in Scala. Pattern matching is a complex topic covered in more detail in the advanced section of this article, so here we just focus on the simple use cases like replacing JavaScript switch/case with it.

ES6
const animal = "Dog";
let description;
switch(animal) {
  case "Cat":
  case "Lion":
  case "Tiger":
    description = "It's feline!";
    break;
  case "Dog":
  case "Wolf":
    description = "It's canine!";
    break;
  default:
    description = "It's something else";
}
console.log(description);
Scala
val animal = "Dog"
val description = animal match {
  case "Cat" | "Lion" | "Tiger" =>
    "It's feline!"
  case "Dog" | "Wolf" =>
    "It's canine!"
  case _ =>
    "It's something else"
}
println(description)

In Scala you can use the |-operator to match multiple choices and there is no need (nor support) for break, as cases never fall through like they do in JavaScript. For the default case, use the ubiquitous _ syntax (it has many many more uses in Scala!) As with if, a match is an expression returning a value that you can directly assign to a variable.

Classes

Being an object-oriented language, Scala naturally supports classes with inheritance. In addition to basic classes Scala also has:

  • case classes for conveniently storing data
  • objects for singletons
  • traits for defining interfaces and mixins

Let us look at a simple class hierarchy in both languages.

ES6
class Shape {
  constructor(x, y) {
    this.x = x;
    this.y = y;
  }

  move(dx, dy) {
    this.x += dx;
    this.y += dy;
  }

  draw() {
    console.log(`Shape at ${this.x}, ${this.y}`);
  }
};

class Circle extends Shape {
  constructor(x, y, r) {
    super(x, y);
    this.r = r;
  }

  draw() {
    console.log(`Circle at ${this.x}, ${this.y} with radius ${this.r}`);
  }
}

const c = new Circle(5, 5, 42);
const r = c.r; // == 42
Scala
// use var to make coordinates mutable
abstract class Shape(var x: Int, var y: Int) {
  def move(dx: Int, dy: Int): Unit = {
    x += dx
    y += dy
  }

  def draw(): Unit = {
    println(s"Shape at $x, $y")
  }
}

// r is immutable but accessible outside the class
class Circle(x: Int, y: Int, val r: Int)
    extends Shape(x, y) {
  override def draw(): Unit = {
    println(s"Circle at $x, $y with radius $r")
  }
}

val c = new Circle(5, 5, 42)
val r = c.r // == 42

Note that this is typically omitted in Scala, since the compiler can tell that x, y and r are properties of the enclosing class (and not local variables).

Case classes

Case classes are a particular kind of class in Scala which have a lot of compiler-generated goodies. They are particularly suited for immutable data containers whose instances are equal if and only if their fields are pairwise equal. They also automatically receive a sensible toString() representation, and they can be instantiated without the new keyword.

JavaScript doesn’t quite have a similar construct, but whenever you would use the regular object notation, consider using a case class instead.

ES6
const person = {first: "James", last: "Bond"};
Scala
case class Person(first: String, last: String)

val person = Person("James", "Bond")

Case classes enforce type safety and prevent constructing invalid objects with missing fields.

The Scala compiler automatically generates a proper equals method for case classes, making comparing them trivial. In ES6 you would typically go for a library like lodash to avoid writing the complex comparison code yourself.

ES6
const o1 = {a: 1, x: "test"};
const o2 = {a: 1, x: "test"};

if (o1 != o2) {
  // this doesn't work as expected
  console.log("They are not equal!");
}

if (_.isEqual(o1, o2)) {
  console.log("They are equal!");
}
Scala
case class AX(a: Int, x: String)

val o1 = AX(1, "test")
val o2 = AX(1, "test")
if (o1 == o2) {
  println("They are equal!")
}

Fields in case classes are public and immutable by default (unless you define them with var modifier) so you cannot make changes to instances. Instead of modifying the instance you make a copy and modify one or more fields during the copy. Scala provides a suitable copy function for each case class automatically. In ES6 you can use Object.assign to achieve the same result.

ES6
const o1 = {a: 1, x: "test"};

// start with empty object to prevent
// modification of o1
const o2 = Object.assign({}, o1, {a: 42});
Scala
case class AX(a: Int, x: String)

val o1 = AX(1, "test")
val o2 = o1.copy(a = 42)

Finally case classes can be used nicely in pattern matching, which is covered in the advanced section.

Objects

An object is a special class with only a single instance: a singleton. JavaScript also has a singleton design pattern (or actually several) even though the language itself does not have direct support for the concept. Singletons are useful for putting stuff in a shared namespace without polluting the global scope.

ES6
const RandomGen = {
  _privateMethod() {
    console.log("I am private");
  },

  _rnd() {
    return Math.random()
  },

  publicMethod() {
    console.log("The public can see me!");
    this._privateMethod();
  },

  name: "RandomGen",

  getRandomNumber() {
    return this._rnd()
  }
}

const r = RandomGen.getRandomNumber();
Scala
import scala.util.Random

object RandomGen {
  private def privateMethod(): Unit = {
    println("I am private")
  }

  private val rnd = new Random()

  def publicMethod(): Unit = {
    println("The public can see me!")
    privateMethod()
  }

  val name = "RandomGen"

  def getRandomNumber: Double = rnd.nextDouble()
}

val r = RandomGen.getRandomNumber

As you can see, defining singleton objects in Scala is quite trivial thanks to the native support in the language.

Another common use for objects in Scala is using them as companion objects for classes to store static variables and methods shared by all instances of the class.

Traits

Scala traits are similar to the mixin design pattern in JavaScript by allowing developer to define behaviors for class composition. Because of Scala’s strict type system, traits are commonly used to describe common interfaces for a group of implementation classes. JavaScript itself has no need for interfaces, but some extensions like TypeScript support them for the same purpose as Scala.

ES6
class Circle extends Shape {
  constructor(x, y, r) {
    super(x, y);
    this.r = r;
  }

  draw() {
    console.log(`Circle at ${this.x}, ${this.y} with radius ${this.r}`);
  }
}

const Clickable = {
  onClick() {
    console.log("Clicked!");
  }
};

class ClickableCircle extends Circle {}
Object.assign(ClickableCircle.prototype, Clickable);

const cc = new ClickableCircle(0, 0, 42);
cc.onClick();
Scala
class Circle(x: Int, y: Int, val r: Int)
    extends Shape(x, y) {
  override def draw(): Unit = {
    println(s"Circle at $x, $y with radius $r")
  }
}

trait Clickable {
  def onClick(): Unit = {
    println("Clicked!")
  }
}

class ClickableCircle(x: Int, y: Int, r: Int)
    extends Circle(x, y, r) with Clickable

val cc = new ClickableCircle(0, 0, 42)
cc.onClick()

Note that there are many ways for defining mixins in JavaScript, using Object.assign is just one of them supported by ES6.

Option, the type safe undefined

The notorious undefined type in JavaScript can be a blessing or a curse. On the other hand it makes life easy by allowing you to drop function parameters or leave variables undefined. But then it also masks many errors and makes you write extra code to check for undefined. Quite often undefined is used to make a distinction between an existing value (of any type) and a missing value.

Scala doesn’t have undefined (it does have null but its use is discouraged), but instead it has an Option trait for representing optional values. In Scala.js the undefined type exists to support interoperability with JS libraries, but even there it is recommended to use Option whenever possible.

Option[A] is a container for an optional value of type A (note that Option[A] is Scala’s notation for type parameters, which most programming languages, including TypeScript, write as Option<A>). If the value of type A is present, the Option[A] is an instance of Some[A], containing the present value of type A. If the value is absent, the Option[A] is the object None.

ES6
function log(msg, context) {
  let s;
  if (context !== undefined)
    s = `[${context}] ${msg}`;
  else
    s = msg;
  console.log(s);
};

// produces: First message
log("First message");
// produces: [debug] Second message
log("Second message", "debug");
Scala
def log(msg: String,
    context: Option[String] = None): Unit = {
  val s = context match {
    case Some(c) => s"[$c] $msg"
    case None => msg
  }
  println(s)
}

log("First message")
log("Second message", Some("debug"))

Pattern matching works nicely with Option, but there are more powerful ways to use it. Let’s rewrite the previous function another way giving us the same result.

Scala
def log(msg: String, context: Option[String] = None): Unit = {
  val s = context.map(c => s"[$c] $msg").getOrElse(msg)
  println(s)
}

Whoa, quite a reduction in code size! Next let’s see how we can process a sequence of option values.

ES6
const data = [1, 2, 3, undefined, 5, undefined, 7];
const res = data.filter((x) => x !== undefined);
Scala
val data = Array(Some(1), Some(2), Some(3),
  None, Some(5), None, Some(7))
val res = data.filter(x => x.isDefined)

Option provides many collection like methods like map, filter and flatMap, which are discussed in the next chapter.