Let's do Befunge-93


Hello, this is my try to teach you a little bit of Befunge-93.


You may ask why someone should learn an esoteric language like Befunge-93 ? It has close to no real world application and is far away from every other language you may already know.

But wait, Befunge has one really neat property (at least in my opinion): its really fun to write in it. And that because it's totally different from every other language. Writing code in Befunge has a lot to do with planning and laying out your code. Yes physically planning how your code will look and where you have to write specific subroutines. But enough, let's go look at some code.

Chapter 1: A whole new dimension

Choosing the right tools

So first you will need an interpreter. For the beginning I recommend to simply use an javascript interpreter - they are missing a lot of important features, like manually jump in your program, breakpoints, or high speed. But they are easy too use, and for now you will write programs the size you don't need all these fancy features. So just google for an online-interpreter - there are many out there, and if you need an more advanced interpreter later you can take a look at my own interpreter BefunExec.

The simplest program

Okay let's start. You will need an empty ASCII-encoded text file - this will be our code. You have to understand that Befunge operates in an 2-dimensional space - like in normal programming you have a PC (Program Counter) that describes your current position. In normal programs (you know, C, Python, Java etc.) this PC is 1-dimensional, in Befunge it's 2dimensional. At the beginning its positioned in the top-left corner and with every tick its moving one field to the left. So every Character in your text-document is a command - the most simple command is a space. The space is the NOP Command - the No operation


  • Befunge operates in a 2-dimensional Grid
  • The program-counter starts top-left and initially moves right
  • Every character in your document is then an individual command

So our first program can be as simple as that:

And it loops around

So an empty file is indeed a valid Befunge Program. So this program does only execute NOP over and over again in an endless loop. A Befunge-93 program has the size of 80x25, and when the PC reaches the right edge it just wraps around and comes back in on the left edge.


  • Befunge-93 programs have a fixed size of 80x25
  • The Befunge space is infinite - it wraps around its edges

OK, but now we want to write real commands to our file - there are 4 basic commands ( v , > , ^ and < ). With these "arrows" you are able to change the direction of the PC. So with these commands we are now able to write a rather simple endless-loop program:

>      v

^      <


The stack

We now have a Grid where we can layout our program and a PC that runs through this grid to execute our program.

Befunge has one more element you should know about: The stack. Every befunge program has a Stack you can manipulate, you can do all the normal stack operations on it, push, pop, peek etc. The commands to push something on the stack are 0 - 9. So if you write

0 1 9 9

it will push a zero, a one, and two nines on the stack. If you want to end the program after that you can use the @ Command.


  • You can push a digit to the stack with the commands 1, 2, 3, 4, 5, 6, 7, 8, 9, 0
  • The Command @ stops the program

You also easily push values to the stack with the so called stringmode With the command " you start the stringmode and with the same command " you can also end it again. While the stringmode is active every Character the PC encounters will be pushed as an ASCII-Value to the stack.

The Program

"abc" @

will leave the stack in the following state after completion:

| 99 |
| 98 |
| 97 |


  • A " starts and ends the stringmode
  • While in stringmode every Command will be pushed as its ASCII Value to the stack

A real program

I know I promised you a Hello-World program. And we are nearly at the point where we can write it. Only one last command is missing, the character-output (,) command. This command takes one value of the stack - interprets it as an ASCII-Value, and outputs it to wherever the interpreter wants to show the output.

So now - lets write this program:

"dlroW olleH" ,,,,,,,,,,, @


And that's it - a really simple Hello World. As you can see the program has two parts, first we fill the stack with the ASCII values for the two words "Hello World", and then we print each char to the output. You may ask why Hello World is written backwards - that's because we work with a stack, so the last pushed value is the first popped value (LI-FO principle ). And - to output something in the correct order - we have to input it in the reversed order

Chapter 2: Let's golf

What we did wrong

Some of you probably know Codegolf . These are programming challenges with the target, to write program-code with the least amount of bytes.

Because of the limited amount of space in Befunge, this is here more important than ever. In our last example you saw how to output a string, but the longer the string gets the more commas you have to write and sooner or later you will have filled the whole space. So now let's try to optimize our Hello World a little bit

"Good code is small code"

Decision, Decisions, Decisions

One of the missing things is the possibility to do conditional logic. Our program has to react to something and react differently depending on the results.

So now it's time to introduce 2 new commands | and _ . These commands are called "Decision making". They pop a value from the stack and change the PC direction depending on its value. The value is interpreted as an boolean, the conversion used is the same you probably know from C. If the value is Zero, its false - otherwise its true. The | Command routes the PC to the top if the value is true and to the bottom if the value is false. Respective behaves the _ command with left and right.

An other command is the :, what it does is pretty simple - it duplicates the top-stack value. And the last command is $: It pops a value from the stack and does nothing with it. With this knowledge we can now finally optimize our Hello World:


                   v  ,  <
>    "dlroW olleH" >  :  |
                         >  $  @


So what happens here ?

  • First we put "Hello World" reversed on the stack
  • The we go in a loop, in every loop we do
    • Duplicate the TOS1
    • Test if the TOS is != 0 (true for every ASCII Character)
    • Output the TOS
  • This loop goes on until the stack is empty. When we now try to access the empty stack (with the duplicate command) it results in two zeros on the stack. This is so because an access to an empty stack will always result in a zero
  • The returned zero is now interpreted as false and we exit the loop, the remaining zero on the stack is removed with the pop command and the program exits on the @ command


  • _ and | change the PC direction depending on a value popped from the stack
  • : duplicates the top-of-stack-value
  • $ removes the top-of-stack-value
  • Accessing an empty stack will result in a zero

One step further

So our Hello World is now a lot smaller. Especially if you remove all the unneeded NOPs in the previous example.

So now lets go to one of the more interesting commands - the trampoline #. After your PC encounters the trampoline it "jumps" over the next command, this results in a few very neat tricks you can do in your code.

The next Hello World I will show you is really optimized. This is the standard code to output a string of variable length and I don't believe that there is a more optimized way to do it.

`"dlroW olleH">:#,_@


Here you can see another neat feature of Befunge - the jump command is used twice, one time from left to right and the second time from right to left, so its used to skip two different commands (: and ,) although only being one command.

Chapter 3: A little bit math

Simple Calculus 101

Our programs are still rather static, we want them to actually do some work, for example calculate something. Luckily Befunge has a few operational commands in its repertoire.

  • + The ADD command: Pops two values from the stack and pushed the result of the addition
  • - The SUB command: Pops two values from the stack and pushed the result of the subtraction
  • * The MULT command: Pops two values from the stack and pushed the result of the multiplication
  • / The DIV command: Pops two values from the stack and pushed the result of the division
  • % The MOD command: Pops two values from the stack and pushed the result of the modulo operation
  • ! The NOT command: Pops one value from the stack, interprets it as a boolean and pushes the negation
  • ´ The COMP command: Pops two values from the stack and pushed the result of the compare "a>b"

So if you want to calculate 4+5 just write



  • Push 4 to stack
  • Push 5 to stack
  • Pop 4 and 5 -> Push (4+5=)9 to the stack

And if you want (4+5)*6+7 you write




Some of you may say that this notation seems familiar. That's true - it's the so called postfix2 notation. This is one of Befunge's very neat "features" writing a mathematical expression will always result in a postfix notation and you can simply write a postfix notation in Befunge.


  • +, *, - ,/, %, !, ` are the operator commands
  • These pop their arguments from the stack and push the result back
  • You can write mathematical expressions in postfix notation

... Eight, Nine, Ten ?

Perhaps you remember Chapter 2, we learned how to push digits to the stack. But probably you will one day feel the need to have values greater 9 on the stack. Now you can't just simply write 12 to push a twelve.


would just push a 1 and a 2 on the stack. The only option we have is two write a formula which results in a twelve on the stack, for example:






You see there are a lot of ways displaying the same value and we sure would like to know the best (= shortest) way to do so.

Now there are a few recipes how to do so:


The most simple way of displaying numbers is expressing them with addition. So

12 => 93+
19 => 991+
100 => 999999999991+

As you can see this method does not scale very well, and you will reach the acceptable limit pretty fast when the number becomes high enough.


Base-9 is by far my favourite way of expressing numbers. You probably know about different bases in number representation. 14 is 14 when written in base-10 (our everyday standard). But its also D in base-16 (or hexadecimal as you may know it). Or its 16 in base-8 (= octal), or its 1110 in base-2 (= binary).

Now we want to encode 105 (base-10). 105 is 126 in base-9 and you can easily convert bases with this formula:

(((1*9) + 2) * 9) + 6  =>  105

or in Befunge (Postfix):

69 29 1 *+ *+

or shorter:


Now you see why we use base-9, because we need to write the 9 in Befunge and 9 is the highest number we can push onto the stack in a single command.

This representation is nice because its really fast (and easy) to calculate the representation - even for big numbers, an because you can easily tell for every number you want how long its base-9-representation will be.


Base-9 is nice in most cases, but often its not the optimal way. Factorization is often more compact.

The idea here is that you can split a number into its factors (and the factors should hopefully be <10 ) and then multiply them.

So 196 (=4*7*7) becomes


Its said that this only works nice when you can factorize the number with factors smaller ten. Otherwise you would need to incorporate other mathematical operation to get to your result, finding the optimal combination here can be very CPU-intensive.

So for 107 a way of displaying it is:



At last we can use a neat little trick. When we printed "Hello World", we put the ASCII values of the individual chars onto the stack. We can use that to express greater values. For example expressing 107 is as easy as "k".

And for greater values you can even go on and write expressions based on the ASCII values of character:

"~~)'"*++  (= 1851)


Number Method Code Number Method Code Number Method Code
0 Boolean 0 40 Factorization 58* 400 Factorization 25*5*8*
1 Boolean 1 41 Stringmode ")" 401 Base9 59894*+*+
2 Digit 2 42 Factorization 67* 402 Factorization 79*4+6*
3 Digit 3 43 Stringmode "+" 403 Base9 79894*+*+
4 Digit 4 44 Stringmode "," 404 Base9 89894*+*+
5 Digit 5 45 Factorization 59* 405 Factorization 59*9*
6 Digit 6 46 Stringmode "." 406 Base9 19095*+*+
7 Digit 7 47 Stringmode "/" 407 Factorization 59*9*2+
8 Digit 8 48 Factorization 68* 408 Factorization 98+3*8*
9 Digit 9 49 Factorization 77* 409 Factorization 59*9*4+
10 Factorization 25* 50 Stringmode "2" 410 Base9 59095*+*+
11 Factorization 92+ 51 Stringmode "3" 411 Base9 69095*+*+
12 Factorization 26* 52 Stringmode "4" 412 Base9 79095*+*+
13 Factorization 94+ 53 Stringmode "5" 413 Factorization 69*5+7*
14 Factorization 27* 54 Factorization 69* 414 Base9 09195*+*+
15 Factorization 35* 55 Stringmode "7" 415 Factorization 99*2+5*
16 Factorization 28* 56 Factorization 78* 416 Factorization 94+4*8*
17 Factorization 98+ 57 Stringmode "9" 417 Base9 39195*+*+
18 Factorization 29* 58 Stringmode ":" 418 Base9 49195*+*+
19 Base9 192*+ 59 Stringmode ";" 419 Base9 59195*+*+
20 Factorization 45* 60 Stringmode "<" 420 Factorization 25*6*7*
21 Factorization 37* 61 Stringmode "=" 421 Base9 79195*+*+
22 Base9 492*+ 62 Stringmode ">" 422 Base9 89195*+*+
23 Base9 592*+ 63 Factorization 79* 423 Factorization 59*2+9*
24 Factorization 38* 64 Factorization 88* 424 Factorization 59*8+8*
25 Factorization 55* 65 Stringmode "A" 425 Factorization 98+5*5*
26 Base9 892*+ 66 Stringmode "B" 426 Factorization 79*8+6*
27 Factorization 39* 67 Stringmode "C" 427 Factorization 69*7+7*
28 Factorization 47* 68 Stringmode "D" 428 Base9 59295*+*+
29 Base9 293*+ 69 Stringmode "E" 429 Base9 69295*+*+
30 Factorization 56* 70 Stringmode "F" 430 Base9 79295*+*+
31 Base9 493*+ 71 Stringmode "G" 431 Base9 89295*+*+
32 Factorization 48* 72 Factorization 89* 432 Factorization 68*9*
33 Stringmode "!" 73 Stringmode "I" 433 Factorization 68*9*1+
34 Base9 793*+ 74 Stringmode "J" 434 Base9 29395*+*+
35 Factorization 57* 75 Stringmode "K" 435 Base9 39395*+*+
36 Factorization 49* 76 Stringmode "L" 436 Base9 49395*+*+
37 Stringmode "%" 77 Stringmode "M" 437 Factorization 68*9*5+
38 Stringmode "&" 78 Stringmode "N" 438 Factorization 89*1+6*
39 Stringmode "'" 79 Stringmode "O" 439 Factorization 68*9*7+

Chapter 4: The end

Self modification

Befunge has one pretty big feature you haven't even seen now ... self-modification. Yes you heard right, Befunge is capable of modifying its own code while running.

The responsible commands are p and g, namely put and get.

With put you can modify a specific command. put gets 3 Values from the stack x, y and value. x and y describe the position of the command and value is the new value of this field.

The get command works the other way around, it gets the value of a command at a specific position. get gets 2 values from the stack x and y, these values describe a specific field in the code, the command then gets the ASCII value of this command and pushed is onto the stack.

A more complex example

Perhaps you want to try to solve the next task on your own. You already know everything you need and at the bottom of this tutorial you will find a full table of every command.

Write a program that outputs the Fibonacci numbers, one after the other An example output would look like: 1,1,2,3,5,8 ...

I write an simple solution that does this task. Note that I didn't highly optimize it on purpose, so its easier to follow what's actually going on:

>> 100p 110p 1. ",",   1.",",     >     00g 10g: 00p + :. ",",  10p       v
                                  ^                                       <


If you want an explanation what's going on here, I also a little explanation:

[0,0] is the first "variable cell" - it contains the second last value
[1,0] is the first "variable cell" - it contains the last value

100p    Initialize cell [0,0] with '1'
110p    Initialize cell [1,0] with '1'
1.",",    Output '1,'
1.",",    Output '1,'

>       Loop Begin
00g        Hole [0,0]
10g        Hole [1,0]
:        Duplicate [1,0]
00p        Set cell [0,0] to the value of [1,0]
+        Add [0,0] + [1,0]
:        Duplicate the result
.",",      Output the result and a comma
10p        Write the result in [1,0]
v      Loop End

Note the cells [0,0] and [0,1] are used as temporary "variable" fields. We can put there values with the p command and later read them with the g command

Command overview

Character Name Description
+ ADD Adds two values from the stack together and pushes the result back
- SUB Subtracts two values from the stack from each other and pushes the result back
* MULT Multiplies two values from the stack and pushes the result back
/ DIV Divides two values from the stack and pushes the result back
% MOD Executes Modulo on two values from the stack and pushed teh result back
! NOT Gets a (boolean) value from the stack and pushes it negated back
´ GT Pushes the result of a greater than over two value from the stack, to the stack
^ PCT Set PC-Delta to up
> PCL Set PC-Delta to left
v PCB Set PC-Delta to down
< PCR Set PC-Delta to right
? PCRAND Set PC-Delta to a random direction
# JMP Jumps over the next command
_ IFH A horizontal If
| IFV A vertical If
: DUP Duplicates the TOS
\ SWAP Swaps two values from the stack
$ POP Removes the TOS
. OUT-INT Outputs the TOS as a number
, OUT-ASC Outputs the TOS as a character
& IN-INT Asks the user for a number and puts it on the stack
~ IN-ASC Asks the user for a character and puts it on the stack
p PUT Sets a field to a specific value
g GET Gets the value of a field and pushes it onto the stack
@ STOP Stops the program execution
" STRMODE Starts/Stops the stringmode
0-9 NUMBERS Pushes the respective number onto the stack

  1. TOS = top of stack 

  2. Postfix = [Postfix notation][6] 

made with vanilla PHP and MySQL, no frameworks, no bootstrap, no unnecessary* javascript