A 10-Minute Introduction to Scala

Scala is a programming language released in 2004 by Martin Odersky. It provides support for functional programming and is designed to be concise and compiled to Java bytecode so that a Scala application can be executed on a Java Virtual Machine (JVM).

Let’s check at the core features of the language.

Hello World

First, let’s see how to implement a hello world in Scala:

We defined a HelloWorld object containing a main method. This method takes an array of String as an input.

In the main, we called the method println which takes an object as an input to print something in the console.

Meanwhile, HelloWorld is also part of the io.teivah.helloworld package.


We can name the result of an expression using the val keyword. In the following example, both expressions are a valid way to define a value:

The type is optional. In the example, v1 and v2 are both typed as a String.

The Scala compiler can infer the type of a value without having to explicitly declare it. This is known as type inference.

Last but not least, a value in Scala is immutable.


A variable is a mutable value. It is declared with the var keyword.

Just like with values, the type is optional.

Meanwhile, Scala is a statically typed language. The following code, for example, is invalid as we try to map an Int in a String:


In Scala, we can combine expressions by surrounding them with {}. Let’s consider the println() function which takes an object as an input. The two following expressions are similar:

Note that for the second println the last expression (i + 2) is the result of the overall block.

When we call a function with a single argument just like println, we can also omit the parenthesis:

Basic Types

Scala is considered a pure object-oriented language because every value is an object. Hence, there is no primitive in Scala (like Java int for example).

There are 8 basic types in Scala:

  • Byte
  • Short
  • Int
  • Long
  • Float
  • Double
  • Char
  • Boolean

Scala type hierarchy

Every basic Scala type inherits from AnyVal. On the other side, AnyRef is an alias for java.lang.Object. Lastly, both AnyVal and AnyRef inherits from Any.

String Interpolation

Scala provides an elegant way to embed variable/value references directly in processed string literals. As a concrete example:

This is made possible by the s interpolator before the quotation mark. Otherwise, it would print Hello $name!.

There are few interpolators provided by Scala but it is a customizable mechanism. We can create for example our own interpolator to handle JSON conversions like this: println(json"{name: $name}").

Array and List

An array is also handled in Scala as an object:

Two things to highlight here.

Firstly, the way to set elements. Instead of using a[0] like in many languages, we use the syntax a(0). This is a syntactic sugar to let us call an object just as if it was a function. Under the hood, the compiler is calling a default method called apply() taking a single input (an Int in our case) to make it possible.

Secondly, despite being declared as a val in this example, the Array object is mutable so we can change the value of indexes 0 and 1. val just enforces to not mutate the reference, not the corresponding object.

An array can also be initialized this way:

This expression is similar than above. Moreover, because it is initialized with 5 and 2, the compiler infers a as an Array[Int].

To manage multi-dimensional arrays:

This code creates a two-dimensional array and initializes the very first element to 5.

There are many different data structures composing the Scala standard library. One of them is the immutable List:

Compared to Array, modifying an index after having initialized a List will lead to a compilation error.


A map can be initialized like this:

Note the -> operator to associate a color key to its corresponding hexadecimal value.

Map is an immutable data structure. Adding an element means creating another Map:

Meanwhile, the elements cannot be modified. In the case we need a mutable structure, we can use scala.collection.mutable.Map:

In this example, we mutated the AK key.

Methods/Functions: Basics

We have to make the distinction between methods and functions. A method is a function that is a member of a class, trait or object (we are going those notions).

Let’s see a basic method example:

Here we defined an add method with the def keyword. It took two Int as an input and returned an Int. Both inputs are immutable (in the sense that they are managed just like if they were declared as val).

The return keyword is optional. The method will automatically return the last expression. Moreover, it’s worth mentioning that in Scala (compared to Java), return exits the current method, not the current block.

One last thing to add, the return type is optional. The Scala compiler is also able to infer it. Yet, for the sake of code maintainability, it might be a good option to set it explicitly.

Furthermore, a method without output can be written is both ways:

We can also return multiple outputs like this:

This prevents us from having to wrap a set of outputs in a specific object.

Another syntactic sugar to mention. Let’s consider a bar method without arguments. We can call this method in both ways:

The best practice is to keep the parenthesis only if bar introduces a side-effect. Otherwise, we call bar like in the second expression.

Also, Scala allows us to indicate that a method argument can be repeated. Just like in Java, this repeatable argument must be the last parameter:

Here, we iterate over each args element and we return an aggregated sum.

Last but not least, we can also define default parameters value:

Invoking default without providing a value for x can be done in two ways.

First, using the _ operator:

Or, using named arguments like this:

Methods/Functions: Advanced

Nested Methods

In Scala, we can nest method definitions. Let’s consider the following example:

In this case, mergesort2 method is used solely by mergesort1. To restrict its access, we may decide to set it private (we’ll see later on the different visibility levels).

Yet, in Scala we can also decide to nest the second method into the first one like that:

mergesort2 becomes available only in the scope of mergesort1.

Higher-Order Functions

Higher-order functions take as parameters a function or return a function as a result. As an example of a method taking a function as a parameter:

f is a function taking an Int as an input and returning an Int. In our example, foo delegates the execution to f by passing i to it.

Function Literals

Scala is considered as a functional language in the sense that every function is a value. It means we can express a function in a function literal syntax like that:

increment is a function with an Int => Int type (which could have been inferred by the Scala compiler). For each integer x it returns x + 1.

If we take again the previous example, we could pass increment to foo:

We can also manage so-called anonymous functions:

The second parameter is a function without any name.


A closure in a function literal which depends on the value of one or more variable/value declared outside this function.

A simple example:

Here, foo depends on Pi which is declared outside of foo.

Partial Functions

Let’s consider the following method to compute the speed from a distance and a time:

Scala allows us to partially apply speed by calling it only with a subset of the mandatory inputs:

Note that in this example, none of the parameters of speed have a default value. So in order to call it, we need to fill all the parameters.

In this example, partialSpeed is a function of type Float => Float.

Then, in the same way as we were calling increment, we can call partialSpeed like this:


A method can define multiple parameter lists:

This method is doing exactly the same job than:

Yet, the way to call multiply is different:

Like requested by the method signature, we call it with two lists of parameters. Then, what if we call multiply with only one list of parameters?

In this case, we partially applied multiply which gives us in return an Int => Int function.

What are the benefits? Let’s consider a function to send a message given a particular context:

As you can see, there’s an effort made to make this function pure. Instead of depending on an external context, we make it available as a parameter of the send function.

Yet, it might be somewhat tedious to have to pass this context during every single call of send. Or maybe a function does not need to know about the context.

One solution may be to partially apply send with a predefined context and to manage an Array[Byte] => Unit function.

Another solution is to curry send and make the context parameter implicit like this:

How can we call send in this case? We can define an implicit context before to call send:

The implicit keyword means that for every function managing an implicit Context parameter, we don’t even need to pass it. It will automatically be mapped by the Scala compiler.

In our case, send manages the Context object as a potential implicit (we can also decide to pass it explicitly). So, we can simply call send with the first argument list.


A class in Scala is a similar concept than in Java:

Point exposed a default (Int, Int) constructor because of the syntax line 1. Meanwhile, x and y are two members of the class.

A class can also contain a collection of methods just like move in the previous example.

We can instantiate Point with the new keyword:

A class can be abstract meaning it cannot be instantiated.

Case Classes

Case classes are a particular kind of classes. If you are familiar with DDD (Domain Driven Design), a case class is a value object.

By default, a case class is immutable:

The value of x and y cannot be changed.

A case class must be instantiated without new:

Case classes (compared to regular classes) are compared by value (and not by reference):


An object in Scala is a singleton:

Objects are defined with the object keyword (👏).


Traits are in a way similar to Java interfaces. They are used to share interfaces between classes but also fields. As an example:

A trait method can also have a default implementation.

Traits cannot be instantiated but they can be extended by classes and objects.


In Scala, every member of a class/object/trait is public by default. There are two other access modifiers:

  • protected: members are only accessible from sub-classes
  • private: members are only accessible from the current class/object

Furthermore, we can also have a more granular way to restrict access by specifying a package in which the restriction is applied.

Let’s consider a class Foo in a bar package. If we want to make a method private only outside of bar we can do it this way:


Generics is also a feature provided by Scala:

To instantiate a generic class:


If-else syntax is similar in Scala than in several other languages:

Yet, in Scala an if-else statement is also an expression. It means we can, for example, define methods like this:


A basic loop can be implemented like this:

to means from 0 to 10 included whereas until means from 0 to 10 excluded.

We can also loop over two elements:

In this example we iterated over all the possible tuple combinations:

a=0, b=0a=0, b=1a=0, b=2a=1, b=0a=1, b=1a=1, b=2

We can also include conditions in the for. Let’s consider the following list of elements:

If we need to iterate over each element of list and consider only the even integers, we can do it this way:

Moreover, Scala provides so-called for comprehensions to create sequence of elements with the form for() yield element. As an example:

In this example, we created a collection of even integers by iterating over each element and yielding it in case it is even. As a result, sub will be inferred as a sequence of integers (a Seq object, the parent of List).

In the same way than with if-else statement, for is also an expression. So we can also define methods like this:

Pattern Matching

Pattern matching is a mechanism to check a value against a given pattern. It is an enhanced version of the Java switch statement.

Let’s consider a simple function to translate an integer into a string:

Scala adds a bit of syntactic sugar to implement an equivalent this way:

First, we removed the return statements. Then, matchB function becomes a pattern matcher as we removed the block statement after the function definition.

Anything else apart from some sugar? Pattern matching is a great addition to case classes. Let’s consider an example taken from Scala documentation.

We want to return a String depending on a notification type. We define an abstract class Notification and two case classes Email and SMS:

The most elegant way to do it in Scala is to use pattern matching on the notification:

This mechanism allows us to cast the given notification and to automatically parse the parameters we are interested in. For example, in the case of an email, maybe we are not interested in displaying the body so we simply omit it with _ keyword.


Let’s consider the concrete use case where we need to print the number of bytes from a given file. To perform the I/O operation, we are going to use java.io.FileReader which may throw exceptions.

The most common way to do it if you are a Java developer would be something like this using a try/catch statement:

The second way to implement it is somewhat similar to Java Optional. As a reminder, Optional is a container brought in Java 8 for optional values.

In Scala, Try is a container for success or failures. It is an abstract class, extended by two case classes Success and Failure.

We first wrap the creation of a new FileReader in a Try call. We use a map to convert an eventual FileReader to an Int by calling the read method. As a result, we get a Try[Int].

Then, we can use pattern matching to determine the type of tried.

Implicit Conversions

Let’s analyze the following example:

We defined two case classes Foo and Bar.

Meanwhile, an object Consumer exposes a consume method taking a Foo in parameter.

In Test, we call Consumer.consume() but not with a Foo as required by the signature of the method but with a Bar. How is this possible?

In Scala, we can define implicit conversions between two classes. In the last example, we simply need to describe how to convert a Bar to a Foo:

If this method barToFoo is imported, the Scala compiler will make sure that we can call consumer with either a Foo or a Bar.


To handle concurrency, Scala was initially based on the actor model. Scala was providing the scala.actors library. Yet, since Scala 2.10 this library became deprecated in favor of Akka actors.

Akka is a set of libraries for implementing concurrent and distributed applications. Nonetheless, we can also use Akka only at the scale of a single process.

The main idea is to manage actors as a primitive for concurrent computation. An actor can send messages to other actors, receive and react to messages and spawn new actors.

Example of communications within an actor system

Just like other concurrent computation models like CSP (Communicating Sequential Processes), the key is to communicate through messages instead of sharing memory between different threads.

Credits: baloocartoons.com

Scala is a very elegant language. Yet, the learning curve is not that small compared to other languages like Go for example. Reading an existing Scala code as a beginner might be somewhat difficult. But once you start to master it, developing an application can be done in a very efficient way.

Further Reading

Documentation | Scala Documentation

A 10-Minute Introduction to Scala was originally published in Hacker Noon on Medium, where people are continuing the conversation by highlighting and responding to this story.

Publication date: 
10/31/2018 - 19:53