Creating Parse Tree from 4thEntrance.lua

So, here we are, at the podcast talking about the compiler that gives the name to this blog. I hope I make it justice. I thought the podcast would talk about how a general compiler works, with added specs about gcc and how it used certain languages or features to make it as famous as it is today (and was almost 12 years ago). I was pleasantly taken by surprise when it got a whole lot more technical and went through pretty much every single step of the way, including how it dealt with special characteristics of languages like Java’s JIT compilation and C++ classes, even going as far as to explain (more or less) the structure of the GCC code. It was kind of hard to take it all in, but I think it’s worth it at the end.

I did not know much about the gcc compiler before listening to the podcast. I thought that it could only compile natively on Linux and could only compile languages directly derived from C, such as C++ and C#. When I was younger I used to use it on a virtual machine running Ubuntu to compile simple C programs. I didn’t have a lot of technical knowledge back then, so I could never get it to work properly on my native Windows system. I thought that it was just not possible, but now I know that I need a small C compiler to make it all work.

I really liked 2 things from GNU Compiler Collection. The first is how it can be so modular by just changing either the front or the back end to change language and architecture. It sounds like a Jack of All Trades, anyone can go and use the base code in order to make a compiler for a new or specific language or for a new processor. The second one is how it optimizes things, first at a general high-level by using trees and then using RTL to make it at a low-level. I think this makes things easier to plan and modify at a design point, because they clearly separate different kinds of optimizations and can add, modify or delete at any point without worrying for unplanned changes or bugs.