Archives | Categories


Table of Contents


1. Introduction

Kittle Buffer, or KBF for short, is an extended Brainf**k implementation. KBF could also be explained as KDr2’s Brainf**k.

This is a project I use to teach my kid, that's why I renamed it to eliminate the dirty word.

This document doesn't contain any introduction or manual of BF, but you must know what BF is before you use KBF. BF has only 8 instructions, it is a simple yet intresting programming language, you will easily find many materials for learning it if you do a search on the internet.

2. Extensions to the vanilla BF

I added a few extensions to the vanilla BF with the intention of making it more expressive, more interesting, and richer for kid teaching.

Among these extensions, the most important ones are ^ and @.

^ is for multiple pointers support. A pointer stack is added besides the data cells (a.k.a, the data buffer), you can push into or pop out a pointer from the pointer stack. Each pointer has a integer ID. When a BF program starts, the default pointer whose ID is 0 is pushed into the pointer stack automatically. Afterwards, you can push a pointer into the pointer stack by ^ID where ID is the numerical ID of the pointer. If you want to pop the top pointer out from the pointer stack, just use ^ without any digit following it. As a special case, the last pointer in the pointer stack can't be popped out: ^ does nothing when there's only one pointer in the pointer stack.

The current pointer which instrcutions like +-<> are manipulating is always the top pointer of the pointer stack.

@ is added for function call. The vanilla BF has no function related facilities and it is hard to do calculations like comparison and division. To cope with this issue, I add a function table to KBF. Each function in that table has an integer ID, for example:

  • 0 coresponds to function identify, a simple function returns its only argument untouched.
  • 1 converts its argument to a string and output it.
  • 21 is + who adds all its arguments up.

A function call is usually expressed as ret = func(args...) in common programming languages, so do we here:

  • first, we use ^N to push a pointer into the pointer stack, the data cell to which this pointer points to will be used to hold the return value of he function, this step can be thought ss ret =
  • then, again we use ^N to push the pointer who points to the data cell with the target function id in it into the pointer stack, this step coresponds to func.
  • then, at push the pointer(s) to the arguments orderly.
  • at last, we use @ARITY to trigger the function call, here ARITY is count of the arguments.
  • @ARITY doesn't change the pointer stack, you should pop out the pointers manually if needed.

Let's see an example: ^0^1>+21^2>>+^3>>>++@2 calculates that cell(0) = cell(2) + cell(3), which is cell(0) = 1 + 2.

  • ^0 pushes pointer 0 to the pointer stack, which is for the return value.
  • ^1 pushes pointer 1 to the pointer stack, > moves it right to cell(1), +21 adds it up to 21. This is for function +.
  • ^2>>+^3>>>++ prepares the arguments, 1 in cell(2), 2 in cell(3).
  • @2 calls the function, after this step, cell(0) will become 3.

There are also some trivial extensions besides the two main extensions:

  • Using a number for repetition, +23 means repeating + for 23 times. This works for +-<>;
  • ! for input data, like what this BF visualizer does. The characters after a ! will not be treated as code, but as data. , will pick character from this data circularly.
  • ? to set a break-point under GUI mode. Under cli-mode, it does nothing, but under GUI mode, it will pause when an ? instruction is encountered if the "Pause at ?" checkbox is checked.

Here is a simple demonstration of the GUI window:

3. How to obtain KBF

There are pre-built binaries hosted here:

Or you can download the source from github and run it from source:

  • install racket
  • clone the source, run raco pkg install in the project direcotry
  • run kbf.rkt
    • racket kbf.rkt will start with the GUI;
    • racket kbf.rkt will run the BF/KBF code in the file;
    • racket kbf.rkt - will read BF/KBF source from stdin.

4. The builtin function-table

  • 0: @1, identify, a function returns its only argument untouched.
  • 1: @1, outputs the number literally, instead of converting it to a char, returns the count of chars it has printed out.
  • 21: @N (N>=1), +
  • 22: @N (N>=1), -
  • 23: @N (N>=1), *
  • 24: @2, /
  • 25: @2, modulo
  • 26: @2, power, returns arg-0 raised to the power of arg-1.
  • 27: @2, =, returns 0 or 1 which stands for false and true respectively.
  • 28: @2, >
  • 29: @2, <
  • 30: @2, >=
  • 31: @2, <=
  • 32: @2, bitwise or
  • 33: @2, bitwise and
  • 34: @2, bitwise xor
  • 35: @2, bitwise shift
  • 36: @2, random

5. Changelog

  • v0.2.1
    • change icon on macOS and Windows
  • v0.2
    • use ? to set a break-point under GUI mode
    • colorize related cells while executing a function call instruction (i.e., @N)
    • cacth the error when brackets mismatch
  • v0.1
    • The first release
    • An extended BF interpreter with the following extensions: ^, @, <N>, !
    • A graphic user interface to visualize the execution process of BF

6. TODO Examples

The below code:


outputs a.

  • ^0 pushes the pointer 0 to the pointer stack;
  • ^1>+21 pushes the pointer 1 to the pointer stack, moves it right for 1 step, then adds its value to 21 (21 is the function +);
  • ^2>>+50 pushes the pointer 2 to the pointer stack, moves it right for 2 steps, then adds its value from 0 to 50;
  • ^3>3+47 pushes the pointer 3 to the pointer stack, moves it right for 3 steps, then adds its value from 0 to 47;
  • @2 uses the top 2 pointers (^2 and ^3) as arguments, calls the function 21 to which the pointer 1 is pointing, places the return value (97 here) into the cell to which pointer 0 is pointing;
  • ^0. selects pointer 0 and outputs the value it points to.


Built with Emacs 29.0.50 (Org mode 9.5.2).

Last updated: 2024-06-08 Sat 07:25.