Fri 17 October 2014

For a while now, I have been perplexed by the land of compilers and interpreters. Thinking of these things sparks all sorts of questions in my mind. For instance, if a compiler compiles my program, but the compiler is itself a program, doesn't something need to compile the compiler? If so, wouldn't this chain just go on forever? And, if I were to write my own computer language (which I could,theoretically, do) how would I go about doing so? And what's the difference between a compiler and an interpreter? Are interpreters and compilers things that I can write?!

In this episode, I will begin by trying to answer some of these questions. In later episodes I will describe my experiences writing a Lisp Interpreter. When I have the interpreter all set up and running, I'll present it, and explain it. But before we get to that, I think its worth while exploring what an interpreter IS, and how it differs from a compiler.

Compilers and interpreters serve a similar purpose: they take a piece of code written in some language that is foreign to your computer, and they turn it into something that can be executed on your machine. It's the way in which these two programs work that differentiates them.

A compiler works by taking the inputed piece of code (this is called the source code) and translating the entire document into machine code (ie your machines native code, written in 0s and 1s). The compiler produces a new file, called the object file which can be run directly on the machine, without any intermediate translation. So, if I write a piece of code in C++, for example, then my compiler will read through the whole document, and will re-write the set of instructions I've written in C++ into machine language instructions. (Usually some sort of optimization is also involved, but we don't have to concern ourselves with that just yet.) Since the code is then executed from the object file, and no longer interacts with the initial C++ document, the C++ document could be deleted from your computer, and the code could still be executed.

An interpreter on the other hand, executes the code directly as it reads it. It doesn't create another translated file first. An interpreter will just read a line of code, will transform it into something it understands, and then it it will execute it. I realize that this might all sound a little vague, especially the part where I say the interpreter "transforms it into something it understands". What does it understand, anyhow? Well, lets say we are writing a lisp interpreter in python (as I will be doing soon!) and the interpreter comes across a line of code that reads "( + 1 1 )". Well, python doesn't understand this, but if the interpreter changes it to "1 + 1", then python does understand this.

Another thing I'd like to mention is that languages are not always either "interpreted" or "compiled". Some languages, Python is one of them, are actually implemented with both a compiler and an interpreter. Although people often refer to Python as an 'interpreted language', its implementation has 2 parts: First, the python code is compiled into something called "byte-code". Then, an interpreter reads and executes the byte-code!

Cool, eh? Big thanks goes to Allison Kaptur for explaining this to me and discussing interpreters and compilers with me! You rock!

Category: Blog