back
3 comments
Writing brainfuck compilers has been one of mine "let's learn this language" things for a long time.

It quickly turns into a rabbit hole though: I have a compiler somewhere that does all kinds of funny things: I got stuck at trying to figure out independent code blocks that I could run in parallel. I was 16, and started by thinking "this shouldn't be too hard" and ended up giving up on it about 3 years later.

Anyway, in the end it did: fusing/folding expressions, bounds check ellision, and partial evaluation.

I ended up running the mandelbrot.b (which is probably the same as in the article) in just under 1.5 second, but I was outputting machine code directly, not running it in go. It was almost 2x faster when compiling to C and letting GCC do the rest...

There's something so tantalizing about ideas like brainfuck.

Depending on your skill level, you can easily make a solid implementation in maybe an hours time.

But once you figure that out, you now have a playground to do things like you're describing on. Optimize execution, map it's memory to a graphics display or network card, build an assembler, etc. Whatever weird stuff pops into your head.

I don't know how useful it ultimately is for most people, but for myself I've always found this stuff an interesting learning tool to spring-board off of.

The same goes for electronics. I've always wanted to build a minimal computer from scratch or discrete components, even if from a kit. But I know once I do this, it'll end up being a huge rabbit hole. It sounds fun, but so does playing Factorio again. Maybe once I retire. :)