This post first appeared on my old blog in January 2021. It is preserved, but maybe not updated, here.
The repo is on Github here, and in case that turns out to be a lie in the future, the code as of the initial writing of this post is here.
About brainfuck
Brainfuck is an esoteric programming language with only eight commands (i.e. it's meant to be fun/challenging, not useful). Memory is represented as cells on an infinitely long tape and accessed with a single pointer. Typically each cell is a single byte, and by convention "infinitely long" means 30,000 cells. The eight commands are:
>move cell pointer to the right,<move cell pointer to the left,+increment value in current cell,-decrement value in current cell,[enter loop if current cell value is not 0, else jump to end,]end loop,,accept one byte of input and store in current cell,.output byte in current cell.
All other characters in brainfuck source code are comments.
As an example, the following program writes "Hello World!" followed by a newline to stdout.
++++++++[>++++[>++>+++>+++>+<<<<-]>+>+>->>+[<]<-]>>.>---.+++++++..+++.>>.<-.<.++
+.------.--------.>>+.>++.
Thoughts
Writing an interpreter and checking that other people's brainfuck programs run in it was a lot easier than actually learning brainfuck and writing it. That said, I do want to learn it well enough to write some simple things.
There are a couple simple optimizations I'd like to try implementing that are documented in the README. The most obvious is to coalesce repeated instructions. I intend on writing a script to measure the improvement these optimizations have.
