yagooar locus

My Take On Computer Science

What Is a Compiler?

Much of my work has come from being lazy.

John Backus Creator of Fortran

A compiler is a translator.

Simple, huh? Indeed, the concept behind a compiler is pretty straight forward. We have a program written in one language, and ask our hero – The Compiler – to translate it into another language.

But when it comes to the very process of translating, things get a little bit involved. A compiler takes care of a lot of different processes that put together allow to create a new version of a program. The most common use case is translating a high-level programming language into a low-level one (hint: assembly). Sometimes the translation is done straight into machine or byte code.

Why do we even need code compilation?

In order to answer that question, we have to go back in history. Back in the 50s, programming was quite different than now. Resources were limited and software costs passed those of hardware. Scientists and engineers began to worry about reducing programmers’ productivity in order to cut those huge costs.

At this point, the name of John Backus became popular because he was the creator of the first high-level programming language for an IBM computer, called “Speedcoding”. While the language itself improved programmers’ productivity, translating the source code “on the fly” came at a huge cost of memory and execution speed. With this approach, Backus created an interpreter, which I am going to cover in another article.

Soon enough, Backus realized that what would greatly improve speed of execution and reduce memory consumption would be a program that would translate a a complete program written in a high-level language into machine code, and than execute the translated program. This is how Fortran and its compiler were created.

What’s interesting is, that even today, programming languages that are compiled run faster and consume less memory than the interpreted ones, so it still holds true after 60 years of computer science.

The benefits of compilers are immense

Compilers not only allow to program in a more user-friendly language, but also allow to abstract machine differences. We write the code once, and compilers translate our code into different machine languages and dialects.

But after finding out all those benefits, we still don’t actually know how compilers do their job. Let’s change that.

The compilation process

Take a look at the following graph:

What we see here is that first of all, the compiler takes care of the translation of the source code into a target code. After that, we just take the target code, also called target program, and run it by providing it some data. In this scenario, the compiler’s work actually finishes after the first step. So let’s try to find out how this Compiler box actually works.

How compilation works

Explaining the whole process of compilation is definitely too much for one single blog post, so let’s just get a quick overview of the different steps:

  • Lexical Analysis: the source code is read and converted into meaningful units, called lexemes. Those are then converted into tokens.
  • Syntax Analysis or Parsing: the tokens are parsed into an intermediate representation, called syntax tree.
  • Semantic Analysis: The syntax tree is used to check the semantic consistency with the language grammar.
  • Intermediate Code Generation: the compiler creates a low-level representation of the program. It is often built for an abstract machine, but can easily be translated into real machine code.
  • Code Optimization: optimization rules are applied to the intermediate code, in order to increase speed of execution or minimize memory or resource consumption. This is one of the most complex and unique parts of the compilation process.
  • Code Generation: the intermediate, optimized code is finally translated into the target language.


We have seen that a compiler is actually a translator, which makes programmers more productive and happy and the programs more universal and portable. To accomplish this task, there are some required steps like analyzing the input and generating the new output.