DominoScript

Current version 0.2.0

Have you ever wanted to write code using domino pieces? No?

Well, now you can! Introducing DominoScript!

A recreational stack-oriented concatenative two-dimensional non-linear self-modifying int32-based esoteric programming language that uses the dots on domino pieces to represent code.

This repository contains the reference implementation written in TypeScript as well as all the documentation and examples for the language.

It's still very much a work-in-progress. Not everything is fully fleshed out yet.

Feel free to open issues for clarification requests, feature suggestions, bugs etc. I am grateful for any interest and help in eliminating bugs, edgecases and improve the documentation. I'd love for people to make their own Interpreters or Compilers and will link to all of them. Just be warned about potential breaking changes this early on!

Table of Contents

• Core Concepts

• How to run DominoScript

• How does it work

  • Text Format
  • About the stack
  • How to represent Strings
  • How to represent floating point numbers
  • How the Instruction Pointer Moves
  • How Navigation Modes work
  • How to read DominoScript manually

• Instructions

  • Stack Management
  • Arithmetic
  • Comparison & Logical
  • Bitwise
  • Control Flow
  • Input & Output
  • Misc

• Navigation Modes

  • Basic Three Way
  • Basic Two Way
  • Basic One Way
  • Cycle Three Way
  • Cycle Two Way
  • Cycle One Way
  • Flip Flop

• Other References

  • Unicode To Domino
  • Error Types
  • Domino Modes
  • Examples

Core Concepts

• Recreational Esolang: This isn't a serious programming language. I got inspired after watching "The Art of Code" by Dylan Beattie where I discovered "Piet" and eventually went down the esolang rabbit hole. I wanted to create a language that is not only weirdly powerful but can also look good when hanged on a wall.

• Stack-Oriented: There is a global data stack that all instructions operate on. Internally every item on the stack is a signed 32-bit integer. Strings are just null-terminated sequences of integers representing Unicode char codes. Floats are not supported. No other data structures exist.

• Concatenative: DominoScript at its core is just another concatenative reverse-polish language. The following code: 0—1 0—5 0—1 0—6 1—0 0—3 1—2 5—1 is the same as 5 6 + dup * . in Forth.

• Two-Dimensional: The code is represented on a rectangle grid. The instruction pointer can move in any cardinal direction. One domino takes up 2 cells on the grid. Direction changes are performed by placing dominos in a certain way (IP always moves from one half to the other half of the same domino) as well as the current Navigation Mode.

• Self-Modifying: The code can override itself similar to befunge.

• Òbfuscated: Because all code is represented using domino pieces, reading it is somewhat like reading machine code. To "de-obfuscate" it you would need to replace the domino pieces with their corresponding instructions and literal values. The following: NUM 5 NUM 6 SUM DUPE MULT NUMOUT is a readable pseudocode representation of DominoScript.

How to run DominoScript

Be warned that the interpreter is still in its early stages and might not always work as expected.

The reference interpreter is here in this repo. See the here for details.

If you want to use dominoscript via the command line, you can install it globally like this:

npm install -g dominoscript

Then you can run it like this:

dominoscript path/to/your/file.ds

Or you can use npx to run it without installing it:

npx dominoscript path/to/your/file.ds

Eventually there will be an online editor where you can write and run it directly in the browser.

Maybe even a repository of user submitted DominoScript programs.

How does it work

DominoScript by default uses Double-Six (aka D6) dominos to represent code. Double-six here means that each domino has 2 sides with up to 6 dots on each side.

Everything is either:

• an instruction
• a number literal
• or a string literal

By default, everything will be parsed using base7. This behaviour can be changed using the BASE instruction. This means that with a higher base you can use dominos with more dots to represent larger numbers with fewer dominos.

The Grid

• The grid is a rectangle of cells which can contain domino pieces.
• The grid can contain up to 65408 cells (soft limit)
• One domino takes up 2 cells and can placed horizontally or vertically.
• The top-left cell is address 0. The bottom-right cell is address width * height - 1.
• When playing domino game variants you can usually place pieces "outside" the grid when both sides have the same number of dots: 🁈🁳🁀 - this is not allowed in DominoScript (Maybe in future versions but for now not worth the extra complexity)

Each cell needs to be indexable using an int32 popped from the stack, so in theory you could have something crazy like a 300k rows and columns. However, the interpreter will likely not be able to handle that. The artifical limit I decided on for now is a total of 65408 cells. That allows a square grid of 256x256 or various rectangular grids like 64x1024, 128x512, or 949x69 as long as the total cell count is 65408 or less. This limit will likely be configurable in future versions.

Text Format

A text based format is used to represent domino pieces.

This format is used as source code. At the beginning it will be the only way to write DominoScript until a visual editor is created that shows actual dominos. Eventually I want to be able to convert images of real dominos on a (reasonably sized) grid into the text format.

• The digits 0 to f represent the dots on half of a domino. To indicate an empty cell, use a dot .
• The "long hyphen" character — indicates a horizontal domino (regular hyphen - also accepted to make it easier to type). It can only appear on even columns and odd rows.
• The "pipe" character | indicates a vertical domino. It can only appear on odd columns and even rows.
• Any other line before and after the actual code is ignored.
• It is just a text format, so the file extension doesn't matter for now. You can make it .md and comment using markdown if you want! See example

Example:

TITLE
=====

You can write the soure code as a normal text file (.ds extension recommended) or as a .md file with markdown comments like here.

Be aware of the following rules:
> 1. You cannot start a non-code line with a dot `.`
> 2. You cannot start a non-code line with a number `0 to f`
> 3. You cannot comment within the code. Only above and below it.

For comments starting with any non-allowed character, add a space or any other allowed char before it.

## DominoScript

The below code NO-OPs forever because
The IP can always move to a new domino

. . . . . . . .

. 6 6 6—6 6 6 .
  | |     | |
. 6 6 6 6 6 6 .
      | |
. 6—6 6 6 6—6 .

. 6—6 6—6 6—6 .

. . . . . . . . 


## Some Notes

Bla bla bla
  

When the source code is parsed it ignores everything except the actual code:

. . . . . . . .

. 6 6 6—6 6 6 .
  | |     | |
. 6 6 6 6 6 6 .
      | |
. 6—6 6 6 6—6 .

. 6—6 6—6 6—6 .

. . . . . . . . 

Which is the equivalent of these dominos:

The grid doesn't have to be a square but it must have a consistent number of columns and rows, otherwise an InvalidGridError will be thrown before execution starts:

GOOD ✅	BAD ❌
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .	. . . . . . . . . . . . . . . . . . . . . . . . . . .

Connecting to a domino half which is already connected results in MultiConnectionError:

GOOD ✅	BAD ❌
6—6 6—6 . 6 6—6 . . | 6 . . . .	6—6—6—6 . 6—6 . . . | 6 . . . .

Having a domino half that is not connected to anything results in MissingConnectionError:

GOOD ✅	BAD ❌
. . 6—6 . . 6 . . . | . 6 . . .	. . 6 6 . . 6 . . . . 6 . . .

Having a connection where 1 or both ends are empty results in a ConnectionToEmptyCellError:

GOOD ✅	BAD ❌
6—6 . 6—6 6 . . . 6 | | 6 . . . 6	6—. . .—6 6 . . . . | | . . . . 6

About the stack

• There is a single global stack that all instructions operate on.
• It only stores signed 32-bit Integers
• The interpreter will preallocate all the memory required to maintain the stack, therefore its size is limited to 512 items for now. (No particular reason for this rather small limit, it will likely be configurable in future versions)

Why not 64-bit integers?: No good reason really. I wanted to implement the first reference interpreter in typescript and since JS converts numbers to 32-bit when doing bitwise operations, I decided to just stick with 32-bit integers instead of having to split the lower and upper 32-bits for every bitwise operation. If there is demand, I will change the spec to support 64-bit ints but for now it is what it is.

How to represent Strings

DominoScript is a language where you cannot really tell what is going on just by looking at the code. It all depends on how the IP moves.

When the IP encounters a STR instruction, it will parse the next dominos as characters of a string. How that works exactly is explained in more detail in the description of the instruction.

It is important to understand that internally everything in DominoScript is represented as signed 32-bit integers and externally everything is represented by the dots on the domino pieces.

Internally strings are just null-terminated sequences of integers representing Unicode characters. It is your job as the developer to keep track of what items on the stack are numbers and what are characters of a string.

You can use any instruction on characters of a "string" but most of them will not distinguish between what is a number and a character. There are only 3 instructions which are specifically for handling strings: STR, STRIN, STROUT.

For convenience and clarity in examples I will often represent Unicode characters like this:

[..., 'NUL', 's', 'e', 'y']

But in reality the stack will store them as integers and look like this:

[..., 0, 115, 101, 121]

How to represent floating point numbers

Floats don't exist in DominoScript. I'd suggest to scale up numbers by a factor of 10, 100, 1000 or whatever precision you need.

(I know that pico-8 uses 32-bits for numbers but treats them as 16.16 fixed point numbers. I am not quite sure if that is just a convention or if pico8's API actually treats them as fixed point numbers. I would like to eventually add some trigonometry instructions to DominoScripts "D9-mode" but unsure what the best way would be)

How the Instruction Pointer Moves

The instruction pointer (IP) keeps track of the current cell address that will be used for the next instruction. Since DominoScript is 2D and non-linear, it isn't obvious where the IP will move to without understanding the fundamental rules and the Navigation Modes.

Before the program starts:

• the interpreter will scan the grid from top-left to top-right, move down and repeat until it finds the first domino.
• Upon reaching the first domino, the IP is placed at the address of the first found domino half.
• If no domino could be found, the program is considered finished.

During the program execution: The IP will adhere to the following rules:

• Rule_01: The IP moves in all cardinal directions, never diagonally. How dominos are parsed is all relative to that. For example the horizontal domino 3—5 can be interpreted as the base7 number 35 (IP moves eastwards) or 53 (IP moves westwards). Same thing for vertical dominos.

• Rule_02: The IP will always move from one half (entry) of the same domino to the other half (exit) of the same domino.

• Rule_03: If the IP cannot move to a new domino, the program is considered finished. If a JUMP happens to move to an empty cell, a JumpToEmptyCellError is thrown and the program exits with a non-zero code

• Rule_04: At the exit half of a domino, the IP will never move back to the entry half. It will always try to move to a new domino. That means there are 0 to 3 potential options for the IP to move.

• Rule_05: When the IP needs to move to a new domino, it is possible that there are no valid moves despite there being dominos around. The Navigation Mode decides where the IP can and cannot move next.

How Navigation Modes work

In a nutshell Navigation Modes are predefined "behaviours" that follow a specific deterministic pattern. There are a lot of different nav modes. Some of them simple and logical, and others a bit more complex and chaotic.

Change navigation modes using the NAVM instruction.

First I'm gonna bombard you with some jargon:

• Priority Directions (PDs): Primary, Secondary, Tertiary
• Relative Directions (RDs): Forward, Left, Right
• Cardinal Directions (CDs): North, East, South, West

The Cardinal directions don't matter much. It is all about the direction in relation to the exit half of the current domino (If you ever did any kind of game dev you probably know the difference between world space and local space. It's kind of like that)

When the IP moves to a new domino, the half it enters to is called the "entry" while the other half is called the "exit". Now from the perspective of the exit half, the IP can potentially move in 3 directions: Forward, Left, Right. These are the Relative Directions (RDs).

Which direction it chooses depends on the current "Navigation Mode". Here are some of the most basic Nav Mode mappings:

index	Primary	Secondary	Tertiary
0	Forward	Left	Right
1	Forward	Right	Left
2	Left	Forward	Right
3	Left	Right	Forward
4	Right	Forward	Left
5	Right	Left	Forward
...	...	...	...

The "index" here is the argument for the NAVMinstruction but also refers to the current Navigation Mode

If we imagine the 6 to be the exit half, what will be the next domino the IP moves to?:

East	West	South	North
. 2 . . . | . 2 . . . 5—6 1—1 . . 3 . . . | . 3 . . .	. . . 3 . | . . . 3 . . 1—1 6—5 . . . 2 . | . . . 2 .	. . 5 . . | 3—3 6 2—2 . . 1 . . | . . 1 . . . . . . .	. . . . . . . 1 . . | . . 1 . . 2—2 6 3—3 | . . 5 . .

All 4 snippets are exactly the same code with the difference that they are all flipped differently. This is what I mean by the cardinal direction not mattering much in DominoScript.

• index 0 the IP will move to 1—1 (Primary, Forward)
• index 1 the IP will move to 1—1 (Primary, Forward)
• index 2 the IP will move to 2—2 (Primary, Left)
• index 3 the IP will move to 2—2 (Primary, Left)
• index 4 the IP will move to 3—3 (Primary, Right)
• index 5 the IP will move to 3—3 (Primary, Right)

What if we remove the 1—1 domino? Where will the IP go to then?:

East	West	South	North
. 2 . . . | . 2 . . . 5—6 . . . . 3 . . . | . 3 . . .	. . . 3 . | . . . 3 . . . . 6—5 . . . 2 . | . . . 2 .	. . 5 . . | 3—3 6 2—2 . . . . . . . . . . . . . . .	. . . . . . . . . . . . . . . 2—2 6 3—3 | . . 5 . .

• index 0 the IP will move to 2—2 (Secondary, Left)
• index 1 the IP will move to 3—3 (Secondary, Right)
• index 2 the IP will move to 2—2 (Primary, Left)
• index 3 the IP will move to 2—2 (Primary, Left)
• index 4 the IP will move to 3—3 (Primary, Right)
• index 5 the IP will move to 3—3 (Primary, Right)

And what if we remove both the 1—1 and the 2—2 domino?:

East	West	South	North
. . . . . . . . . . 5—6 . . . . 3 . . . | . 3 . . .	. . . 3 . | . . . 3 . . . . 6—5 . . . . . . . . . .	. . 5 . . | 3—3 6 . . . . . . . . . . . . . . . . .	. . . . . . . . . . . . . . . . . 6 3—3 | . . 5 . .

• index 0 the IP will move to 3—3 (Tertiary, Right)
• index 1 the IP will move to 3—3 (Secondary, Right)
• index 2 the IP will move to 3—3 (Tertiary, Right)
• index 3 the IP will move to 3—3 (Secondary, Right)
• index 4 the IP will move to 3—3 (Primary, Right)
• index 5 the IP will move to 3—3 (Primary, Right)

Again, these are only the very basic navigation modes. See the reference for all the different modes and how they work.

How to read DominoScript manually

DS isn't meant to be human readable but there are deterministic patterns that will make it much easier to follow along once you know them.

All of these patterns revolve around how the NUM and STR instructions behave differently than any other instruction.

Once you understand their differences, reading the rest of DominoScript is mostly a matter of keeping track of how the other instructions affect:

• the Stack (most of them do)
• the Instruction Pointer (JUMP, CALL, NAVM).

Here are the pattern and some examples:

PATTERN 1:

Look out for 0—1 and 0—2 dominos.

These are the opcodes for the NUM and STR instructions respectively. They are used to push numbers or strings to the stack so you will see them a lot.

They are the ONLY 2 instructions that don't get their arguments from the stack but from the board.

PATTERN 2:

Look out for the first half of a domino right after a NUM instruction.

It will decide how many more dominos will be part of the argument before the next instruction is executed.

The below code results in the number 6 being pushed and popped of the stack:

0—1 0—6 0—0

• 0—1 is the NUM instruction (PATTERN 1)
• 0—6 is the argument for NUM
  • first half is 0 which means no more dominos will follow and only second half is parsed as number (see PATTERN 2)
  • Second half is 6 in both base7 and decimal so the number 6 is pushed to the stack
• 0—0 is the next instruction. We know that because the first half of previous domino told us that no more dominos will be part of the argument. (see PATTERN 2)

The below code results in the number 1000 being pushed and popped off the stack:

0—1 2—0 2—6 2—6 0—0

• 0—1 is the NUM instruction (see PATTERN 1)
• 2—0 2—6 2—6 is the argument for NUM representing 1000 in base7
  • the first half is 2 which means 2 more dominos will be part of the argument (see PATTERN 2)
  • the remaining 2.5 dominos are parsed as 2626 in base7 which is 1000 in decimal.
  • 0—0 is the next instruction. We know that because the first half of the domino after NUM told us that 2 more dominos will be parsed as part of the number so 3rd one after will be an instruction(see PATTERN 2).

PATTERN 3:

Look out for the first half of a domino right after a STR instruction.

For the same reason as after a NUM instruction. It will decide how many more dominos will be part of the character before the next character is parsed.

PATTERN 4:

Look out for the NULL terminator 0—0 during a STR instruction.

It indicates the the string is complete and will be pushed to the stack and that next domino will be a new instruction.

The below code results in the string "abc" being pushed to the stack.

0—2 1—1 6—6 1—2 0—0 1—2 0—1 0—0 0—1 0—6 0—0

• 0—2 is the STR instruction
• 1—1 6—6 is the Unicode value for "a"
• 1—2 0—0 is the Unicode value for "b"
• 1—2 0—1 is the Unicode value for "c"
• 0—0 is the null terminator and not the POP instruction as in the previous 2 examples. We know that because STR only ends once it encounters a 0—0 (see PATTERN 4)
• 0—1 0—6 0—0 is the code from the first example above. It will push the number 6 to the stack and then pop it off again.

All these patterns are deterministic but it doesn't mean that every 0—0, 0—1, 0—2 you see is a NULL terminator, a NUM instruction or a STR instruction respectively (as you can see in the examples above!)

It all depends on the context and how the IP moves:

• A 0—0 can be part of a number, the NULL terminator or the POP instruction
• A 0—1 can be part of a number, the NUM instruction or the ADD instruction if the IP moves in a different direction
• A 0—2 can be part of a number, the STR instruction or the NOT instruction if the IP moves in a different direction

The patterns are universal for all cardinal directions the Instruction Pointer can move in.

I only showed examples where the IP moves from left to right but you have to understand that the same domino can either mean the same thing or something completely different depending on the direction the Instruction Pointer moves in and what instructions precede it:

0—1 . 1—0 . 1 . 0 . . .
            |   |
. . . . . . 0 . 1 . . .

Instructions

The base instruction set is designed to fit on a single "double-six" domino. It consists of up to 49 instructions and is shown below on a 7x7 matrix.

Keep in mind that if you change into a higher "base", you will need to use a different domino to represent the same opcode than shown in the images you see alongside each instruction. E.g. a NOOP in base7 is 6—6 but in base16 it would be 3—0.

	0	1	2	3	4	5	6	CATEGORY
0	POP	NUM	STR	DUPE	ROLL	LEN	CLR	Stack Management
1	ADD	SUB	MULT	DIV	MOD	NEG	_	Arithmetic
2	NOT	AND	OR	EQL	GTR	_	_	Comparison & Logical
3	BNOT	BAND	BOR	BXOR	LSL	LSR	ASR	Bitwise
4	NAVM	BRANCH	LABEL	JUMP	CALL	IMPORT	WAIT	Control Flow
5	NUMIN	NUMOUT	STRIN	STROUT	_	_	_	Input & Output
6	GET	SET	_	BASE	EXT	TIME	NOOP	Misc

(DominoScript isn't limited to these 49 instructions though. The way the language is designed, it can theoretically be extended to up to 1000 instructions)

Stack Management

POP

Discards the top of the stack.

Errors:

• If the stack is empty, a StackUnderflowError is thrown.

NUM

Switch to "number mode". The first half of the next domino will indicate how many dominos to read as a number. Then the other halfs will all be read as base7 digits (in D6 mode) to form the number that will be pushed to the stack.

With 7 dominos, 13 out of 14 halfs are used for the number. You can theoretically represent a number much larger than the max int32 value. However, if the number exceeds the maximum int32 value, it will wrap around from the minimum value, and vice versa (exactly the same as when doing bitwise operations in JS --> (96889010406 | 0) === -1895237402).

You might think that since internally numbers are int32s, that we parse from base7 to two's complement. That is not the case. We simple push the decimal version of the positive base7 number to the stack

For example:

• 0—0 represents the number 0 in both deciamal and base7
• 0—6 represents the number 6 in both decimal and base7
• 1—6 6—6 represents the number 342 in decimal and 666 in base7
• 2—6 6—6 6—6 represents the number 16,806 in decimal and 6,666 in base7
• 6—6 6—6 6—6 6—6 6—6 6—6 represents the number 1,977,326,742 in decimal and 66,666,666,666 in base7 (about 92.1% of the max int32 value)
• 6—0 1—0 4—1 3—4 2—1 1—1 6—1 represents the number 2,147,483,647 in decimal and 104,134,211,161 in base7 (exactly the max int32 value)
• 6—6 6—6 6—6 6—6 6—6 6—6 6—6 represents the number -1,895,237,402. WHY?: The actual decimal number the dominos represent is 96,889,010,406 which is ~45x larger than the max int32 value. It wraps around about that many times before it reaches the final value.

What if numbers are read from the other direction?

• 1—1 1—1, 2—2 2—2 2—2 for example will be exactly the same numbers (216 in decimal) eastwards and westwards.
• 1—2 3—1 when parsed backwards is 1—3 2—1 and can therefore represent different numbers if the IP moves to the east or to the west.
• 1—6 6—6 represents 666 in base7 (342 in decimal) but when parsed backwards the interpreter will raise an UnexpectedEndOfNumberError. Remember that the first half of the first domino indicates how many more will follow. In this case it expects to read 6 more dominos but the number ends prematurely after 1 domino.

To push the number 10 and 5 to the stack you would use the following dominos:

• In pseudo code: NUM 10 NUM 5
• In DominoScript: 0—1 1—0 1—3 0—1 0—5
  • 0—1 is NUM
  • 1—0 1—3 is the number 13 in base7 which is 10 in decimal
  • 0—1 is NUM again
  • 0—5 is the number 5 in both base7 and decimal

To push the number -10 and -5 to the stack you would use the following dominos:

• In pseudo code: NUM 10 NEG NUM 5 NEG
• In DominoScript: 0—1 1—0 1—3 1—5 0—1 0—5 1—5
  • 0—1 is NUM
  • 1—0 1—3 is 13 in base7 which is 10 in decimal
  • 1—5 is NEG
  • 0—1 is NUM again
  • 0—5 is 5 in both base7 and decimal
  • 1—5 is NEG

Errors:

• UnexpectedEndOfNumberError is thrown when the IP cannot step as many times as indicated by the first half of the first domino.

STR

With STR you switch to "string mode" and can push multiple integers to the stack to represent Unicode characters.

The way the dominos are parsed to numbers is identical to NUM: First half of first domino indicates how many more will follow for a single number.

The difference is that it doesn't stop with just one number. It will keep reading numbers until it encounters the NULL character represented by domino 0—0.

Only once the interpreter does encounter the NULL character, will it push the characters to the stack in reverse order.

(Note: I decided to parse strings like this because I wanted a single int32 based stack and, out of all options I could think of, this one felt the least annoying. If you can think of better ways, I am open for suggestions!)

This is how you push the string "hi!" to the stack and output it:

0—2 1—2 0—6 1—2 1—0 1—0 4—5 0—0 5—3

It equals the following pseudo code: STR "hi!" STROUT

• 0—2 is the STR instruction
• 1—2 0—6 is the Unicode value 105 representing the character h
• 1—2 1—0 is the Unicode value 105 representing the character i
• 0—0 4—5 is the Unicode value 33 representing the character !
• 0—0 is the Unicode value for the NULL character which terminates the string.
• 5—3 is the STROUT instruction. It will pop items from the stack, parse them as Unicode chars and once it encounters the NULL character, it will output the string to stdout all at once.

This is the resulting stack:

Imaginative	Reality
[..., 'NUL', '!', 'i', 'h']	[..., 0, 33, 105, 104]

Keep in mind that the IP can move in 4 cardinal direction so the following variations would also push the string "hi!" to the stack:

IP moves right to left:

3—5 0—0 5—4 0—1 0—1 2—1 6—0 2—1 2—0

IP moves in multiple directions:

0 . . . . 0 4—5
|         |
2 . . . . 1 . 0
              |
1 . . 2 1—0 . 0
|     | 
2 0—6 1 . . 3—5

DUPE

Duplicate the top item on the stack.

Stack Before	Stack After
[a, b]	[a, b, b]

ROLL

Pops one argument from the stack to be used as "depth".

• With a negative depth, the item at the top is moved to the nth depth
• With a positive depth, the item at the nth depth is moved to the top

With roll you can implement common stack operations like SWAP and ROT:

Roll Depth	Equivalent to	Stack Before	Stack After
-3	-	[a, b, c, d]	[d, a, b, c]
-2	ROTR	[a, b, c]	[c, a, b]
-1	SWAP	[a, b]	[b, a]
0	NOOP	[a]	[a]
1	SWAP	[a, b]	[b, a]
2	ROTL	[a, b, c]	[b, c, a]
3	-	[a, b, c, d]	[b, c, d, a]

LEN

Pushes the number of items on the stack.

CLR

Removes all items from the stack.

Arithmetic

ADD

Pops 2 numbers from the stack. The sum is pushed to the stack.

SUB

Pops 2 numbers from the stack. The result of numberA - numberB is pushed to the stack.

MULT

Pops 2 numbers to multiply. The result is pushed to the stack.

DIV

Pops 2 numbers. The result of the division of numberA by numberB is pushed to the stack.

Keep in mind that DominoScript is integer based and any remainder is discarded.

Pseudocode:

• NUM 5 NUM 3 DIV is 5 / 3 and equals 1
• NUM 5 NEG NUM 3 DIV is -5 / 3 and equals -1

MOD

Pops 2 numbers. The remainder of division of numberA / numberB is pushed to the stack.

When numberA is positive modulo behaves identical in most languages (afaik). However, there are some differences across programming languages when numberA is negative. In DominoScript modulo behaves like in JavaScript, Java, C++ and Go and NOT like in Python or Ruby!

Pseudocode:

• NUM 5 NUM 3 MOD is 5 % 3 and equals 2
• NUM 5 NEG NUM 3 MOD is -5 % 3 and equals -2 (in python, ruby and calculators it would equal 1)

NEG

Pops the top item off the stack. Negates it. Then pushes the negated version back onto the stack. Essentially a num * -1 operation.

RESERVED_1_6

Unmapped opcode. Will throw InvalidInstructionError if executed.

Comparison & Logical

NOT

Pops the top item off the stack. If it is 0, it pushes 1 to the stack. Otherwise it pushes 0.

AND

Pops the top 2 items off the stack, performs the logical AND operation and pushes the result back onto the stack.

OR

Pops the top 2 items off the stack, performs the logical OR operation and push the result back onto the stack.

EQL

Pops the top 2 items off the stack, compares them and pushes the result back onto the stack. If the items are equal, it pushes 1 to the stack, otherwise 0.

GTR

Pops the top 2 items off the stack, compares them and pushes the result back onto the stack. If the first item is greater than the second, it pushes 1 to the stack, otherwise 0.

RESERVED_2_5

Unmapped opcode. Will throw InvalidInstructionError if executed.

RESERVED_2_6

Unmapped opcode. Will throw InvalidInstructionError if executed.

Bitwise

BNOT

Bitwise NOT. Pops the top item off the stack, inverts all bits and pushes the result back onto the stack.

BAND

Bitwise AND. Pops the top 2 items off the stack, performs bitwise AND and pushes the result back onto the stack.

BOR

Bitwise OR. Pops the top 2 items off the stack, performs bitwise OR and pushes the result back onto the stack.

BXOR

Bitwise XOR. Pops the top 2 items off the stack, performs bitwise XOR and pushes the result back onto the stack.

LSL

Logical Shift Left. Performs the equivalent of argA << argB and pushes the result back onto the stack.

LSR

Logical Shift Right. Performs the equivalent of argA >>> argB and pushes the result back onto the stack.

ASR

Arithmetic Shift Right. Performs the equivalent of argA >> argB and pushes the result back onto the stack.

Control Flow

NAVM

Changes the Navigation Mode. The default Mode is 0.

See Navigation Modes to see all possible nav modes and their indexes.

BRANCH

Like an IF-ELSE statement. It pops the top of the stack as a condition:

• When true: The IP will move LEFT
• When false: The IP will move RIGHT

It ignores the current Navigation Mode, so you can be assured that it will always either go left or right.

Keep in mind that: all non-zero numbers are considered true. Only 0 is false! -1, -2 etc. is true

Here we push 1 to the stack which will cause the IP to move LEFT:

. . . . . 6 . .
          |
. . . . . 6 . .

0—1 0—1 4—1 . .
          
. . . . . X . .
          |
. . . . . X . .

Here we push 0 to the stack which will cause the IP to move RIGHT:

. . . . . X . .
          |
. . . . . X . .

0—1 0—0 4—1 . .
          
. . . . . 6 . .
          |
. . . . . 6 . .

LABEL

A label is like a bookmark or an alternative identifier of a specific Cell address. You can also think of it as a pointer. It can be used by the JUMP, CALL, GET and SET instructions.

Labels are probably not what you expect them to be.

• They are not strings, but negative numbers.
• They are auto generated and self decrementing: -1, -2, -3, etc. ...

Executing the LABEL instruction pops the address of the cell you want to label from the stack and assigns it to the next available negative number label.

The negative number label will NOT be pushed to the stack. First label will be -1, second label will be -2 and so on. You need to keep track of them yourself.

For clarity, I'd generally recommend adding comments like the following to your source files:

## Label Mappings
| Label | Address | Function name |
|-------|---------|---------------|
| -1    | 340     | main          |
| -2    | 675     | update        |
| -3    | 704     | whatever      |

It is not mandatory to use labels. The 4 mentioned instructions that can use them also work with addresses directly!

Labels are mandatory only in the following cases:

• Calling JS functions: The "official" js interpreter exposes an API where you can define functions in JS and call them from DominoScript via a label (Not implemented yet!)
• Calling imported functions: DominoScript will have an IMPORT instruction that allows source files to be imported into others. The imported functions can only be called via labels, so in that regard a label also acts like an export. (Not implemented yet!)

JUMP

Moves the IP to a labeled address on the grid. If the IP cannot move anymore, the program will terminate.

If label is unknown it will throw an UnknownLabelError.

CALL

Like the name suggests, it is similar to a function call.

Exactly like JUMP with one crucial difference: When it cannot move anymore, the IP will return to where it was called from instead of terminating the program.

Internally there is another stack that keeps track of the return addresses.

IMPORT

Pop a "string" from the stack to indicate the file name of the source file to import.

On import the interpreter will load the file and start running it until its Instruction Pointer cannot move anymore.

Labels defined in the imported file are accessible from the importing file. That means you can call functions from the imported file via the CALL instruction.

If the importer file defined label before the import, the labels from the imported will have different identifiers. For example:

• FileChild.ds defines a label -1.
• FileAParent.ds defines labels -1, -2, then imports FileB.ds, then defines another label -6.

The label -1 in FileChild.ds will be -3 in FileAParent.ds because labels are always auto decrementing. Why? Because it is the simplest way to avoid conflicts and be able to use labels internally and externally.

The data stack is shared between parent and all imported files. Apart from that they parent and child imports run in their own contexts. Imported files can have imports themselves but you should avoid circular dependencies.

If you import the same file into more than one other file, it will result in multiple instances of the imported file. This is not a problem as long as you are aware of it.

WAIT

Pops the top item off the stack and waits for that many milliseconds before continuing.

(You could simulate a delay without using WAIT using a 'busy loop' like in example 011_basic_game_loop but it is not recommended)

Input & Output

NUMIN

Prompt the user for a number. The user input will be pushed to the stack.

NUMOUT

Pop the top item from the stack and output it to stdout.

STRIN

Prompt the user for a string. The user input will be pushed to the stack as individual Unicode characters in reverse order.

So if the user inputs "yes", the stack will look like this:

[..., 0, 115, 101, 121]

For convenience you might often see the stack represented But remember that in reality it just stores int32s.

[..., 'NUL' 's', 'e', 'y']

STROUT

Pops numbers (representing Unicode char codes) from the stack until it encounters a null terminator (number 0). It will then output the string to stdout.

RESERVED_5_4

Unmapped opcode. Will throw InvalidInstructionError if executed.

RESERVED_5_5

Unmapped opcode. Will throw InvalidInstructionError if executed.

RESERVED_5_6

Unmapped opcode. Will throw InvalidInstructionError if executed.

Misc

GET

Get the value of a single whole domino at a specific address on the grid.

Pops an item from the stack representing the address of the first domino half, then gets the number of dots on that first half and the other half. Out of both halves it forms a number the same way it does when parsing opcodes. The resulting decimal value will then be pushed to the stack.

For example:

• 0—6: First half 0, second half 6 -> 6 in base7 -> 6 in decimal pushed to stack
• 1—0: First half 1, second half 0 -> 10 in base7 -> 7 in decimal pushed to stack
• 6—6: First half 6, second half 6 -> 66 in base7 -> 48 in decimal pushed to stack

Errors:

• If the address is out of bounds, an InvalidAddressError is thrown.
• If the address references an empty cell the number -1 will be pushed to the stack. No error.

SET

Sets the value of 2 cells (a whole domino) on the grid. This one is similar to the p instruction in befunge used to modify the grid.

It requires 2 arguments from the stack:

• The address of the first cell (just like GET)
• and then the value to set the domino to. 0-48 range is allowed for D6 mode. If

For example here we have a 10x3 grid:

Before SET	After SET
0—1 1—0 6—0 0—1 1—0 . . . . . . 1—6 0—1 . . . . . . . . . .	0—1 1—0 6—0 0—1 1—0 . . 0—6 . . 1—6 0—1 . . . . . . . . . .

What has happened?:

• 0—1 1—0 6—0 We push the number 42 to the stack
• 0—1 1—0 1—0 We push the number 13 to the stack (last domino wraps around on second line)
• 6—1 We execute the SET instruction
  • it pops 13 for the address
  • it pops 42 for the value

The decimal number 42 is 60 in base7. The cell at address 13 is set to the value 6, the cell at the address 12 is set to the value 0:

But why is the set domino "backwards"? Because the IP was moving to the west when SET was executed. Remember that the IP can move in all 4 cardinal directions, so there is no "backwards" or "forwards" in the traditional sense. The address argument indicates the first cell only. The other one depends on the direction the IP is moving.

Errors:

• If the address argument is out of bounds, an InvalidAddressError is thrown.
• If the value argument is not within 0-48, an InvalidDominoValueError is thrown.
• If the address of the other domino half is out of bounds, an AddressError is thrown.

RESERVED_6_2

Unmapped opcode. Will throw InvalidInstructionError if executed.

BASE

Pops one number from the stack to use as the "base" for future parsing of dominos (opcodes, number literals, string literals)

By default, DominoScript uses double six (D6) dominos to represent everything, so the default base is 7.

The max cell value of half of a domino is always 1 less than the Base. So in base 7, the max value is 6. In base 10, the max value is 9. In base 16, the max value is 15 (aka f).

If the number of dots on a domino half exceeds the base, it is clamped.

In below table you can see how the same domino sequence results in different decimal numbers depending on the base:

Domino Sequence	Base 7 (D6)	Base 10 (D9)	Base 16 (D15)
0—6	6	6	6
0—9	6	9	9
0—f	6	9	15
1—6 6—6	342	666	1638
1—9 9—9	342	999	2457
1—f f—f	342	999	4095
2—6 6—6 6—6	16806	66666	419430
2—9 9—9 9—9	16806	99999	629145
2—f f—f f—f	16806	99999	1048575

With a higher Base, you have access to higher opcodes without needing to switch to extended mode.

Base	Opcode Range
7	0 to 48
10	0 to 99
16	0 to 255

While the opcode-to-instruction mapping never changes, the domino-to-opcode mapping is completely different in each base.

The below table shows how the domino 3—0 is mapped to different opcodes depending on the base.

Base	Opcode	Instruction
7	21	BNOT
8	24	BOR
9	27	LSL
10	30	BRANCH
11	33	IMPORT
12	36	NUMIN
13	39	undefined
14	42	GET
15	45	BASE
16	48	NOOP

EXT

Toggle extended mode on or off. If extended mode is active the interpreter will use 2 dominos instead of 1 for each instruction which extends the opcode range from 0-48 to 0-2400.

The range 0-1000 is reserved for "real" instructions.

The range 1001-2400 is used to call functions by label. It is essentially "Syntactic Sugar" to execute CALL with a label. Making function calls look like actual instructions.

To call a function with the label -1 you'd normally do 0—1 0—1 1—5 4—4 which is equivalent to NUM 1 NEG CALL. In extended mode you could do the same 0—0 0—1 0—1 0—0 1—5 0—0 4—4 BUT You can also do the exact same using 2—6 3—0 which is the opcode 1001 and is mapped to the label -1.

TIME

Pushes the milliseconds since program start to the stack.

Useful for things like a gameloop, animations, cooldowns etc.

NOOP

No operation. The IP will move to the next domino without executing any instruction.

(If you have 10 NOOPs in a row it will do 10 steps without doing anything. Over time, the interpreter may optimize this and do an implicit jump to the end of the NOOP chain when it things you are within a loop and the navigation mode doesn't change)

Navigation Modes

(F=Forward, L=Left, R=Right)

There are 49 total navigation modes in DominoScript. This section is a reference for all of them.

• Basic: The IP will prioritize moving in specific directions
  • Basic Three Way
  • Basic Two Way
  • Basic One Way
• Cycle: The IP will prioritize moving in specific directions but the priority will change every cycle.
  • Cycle Three Way
  • Cycle Two Way
  • Cycle One Way
• Flip Flop: The IP will alternate between two primary directions.
  • Flip Flop

Basic Three Way

Out of three directions, the IP will prioritize moving to the one with the highest priority.

Index	Priorities	Domino ->
0	Forward Left Right	0—0
1	Forward Right Left	0—1
2	Left Forward Right	0—2
3	Left Right Forward	0—3
4	Right Forward Left	0—4
5	Right Left Forward	0—5
6	RANDOM	0—6

Basic Two Way

Out of two directions, the IP will prioritize moving to the one with the highest priority.

Index	Priorities	Domino ->
7	Forward Left	1—0
8	Forward Right	1—1
9	Left Forward	1—2
10	Left Right	1—3
11	Right Forward	1—4
12	Right Left	1—5
13	RANDOM	1—6

Basic One Way

IP can only move in one direction.

Index	Only Direction	Domino ->
14	Forward	2—0
15	Forward	2—1
16	Left	2—2
17	Left	2—3
18	Right	2—4
19	Right	2—5
20	RANDOM	2—6

Cycle Three Way

The direction with the highest priority becomes the least prioritized in the next cycle.

All 3 directions are available in all cycles.

Index	Cycle 1	Cycle 2	Cycle 3	Domino ->
21	F L R	L R F	R F L	3—0
22	F R L	R F F	L F R	3—1
23	L F R	F R F	R L F	3—2
24	L R F	R F L	F L R	3—3
25	R F L	F L R	L R F	3—4
26	R L F	L F R	F R L	3—5
27	(unmapped)	(unmapped)	(unmapped)	3—6

Cycle Two Way

The direction with the highest priority becomes the least prioritized in the next cycle.

Only 2 directions are available in a single cycle.

Index	Cycle 1	Cycle 2	Cycle 3	Domino ->
28	F L	L R	R F	4—0
29	F R	R F	L F	4—1
30	L F	F R	R L	4—2
31	L R	R F	F L	4—3
32	R F	F L	L R	4—4
33	R L	L F	F R	4—5
34	(unmapped)	(unmapped)	(unmapped)	4—6

Cycle One Way

The direction with the highest priority becomes the least prioritized in the next cycle.

Only 1 direction is available in a single cycle.

Index	Cycle 1	Cycle 2	Cycle 3	Domino ->
35	F	L	R	5—0
36	F	R	L	5—1
37	L	F	R	5—2
38	L	R	F	5—3
39	R	F	L	5—4
40	R	L	F	5—5
41	(unmapped)	(unmapped)	(unmapped)	5—6

Flip Flop

The priority flip flops between 2 primary directions.

Index	Flip	Flop	Domino ->
42	F	L	6—0
43	F	R	6—1
44	L	F	6—2
45	L	R	6—3
46	R	F	6—4
47	R	L	6—5
48	(unmapped)	(unmapped)	6—6

Other References:

Unicode to Domino Lookup Table

Wiith DominoScript you can output Unicode characters to the console. Here is a lookup table for the ASCII range.

Control characters (ASCII 0-31)

CHARACTER	UNICODE (Hex)	DECIMAL	BASE7	DOMINO -->
NUL (null character)	U+0000	0	0	0—0
SOH (start of heading)	U+0001	1	1	0—1
STX (start of text)	U+0002	2	2	0—2
ETX (end of text)	U+0003	3	3	0—3
EOT (end of transmission)	U+0004	4	4	0—4
ENQ (enquiry)	U+0005	5	5	0—5
ACK (acknowledge)	U+0006	6	6	0—6
BEL (bell)	U+0007	7	10	1—0 1—0
BS (backspace)	U+0008	8	11	1—0 1—1
HT (horizontal tab)	U+0009	9	12	1—0 1—2
LF (line feed)	U+000A	10	13	1—0 1—3
VT (vertical tab)	U+000B	11	14	1—0 1—4
FF (form feed)	U+000C	12	15	1—0 1—5
CR (carriage return)	U+000D	13	16	1—0 1—6
SO (shift out)	U+000E	14	20	1—0 2—0
SI (shift in)	U+000F	15	21	1—0 2—1
DLE (data link escape)	U+0010	16	22	1—0 2—2
DC1 (device control 1)	U+0011	17	23	1—0 2—3
DC2 (device control 2)	U+0012	18	24	1—0 2—4
DC3 (device control 3)	U+0013	19	25	1—0 2—5
DC4 (device control 4)	U+0014	20	26	1—0 2—6
NAK (negative acknowledge)	U+0015	21	30	1—0 3—0
SYN (synchronous idle)	U+0016	22	31	1—0 3—1
ETB (end of transmission block)	U+0017	23	32	1—0 3—2
CAN (cancel)	U+0018	24	33	1—0 3—3
EM (end of medium)	U+0019	25	34	1—0 3—4
SUB (substitute)	U+001A	26	35	1—0 3—5
ESC (escape)	U+001B	27	36	1—0 3—6
FS (file separator)	U+001C	28	40	1—0 4—0
GS (group separator)	U+001D	29	41	1—0 4—1
RS (record separator)	U+001E	30	42	1—0 4—2
US (unit separator)	U+001F	31	43	1—0 4—3

ASCI Printable Characters

CHARACTER	UNICODE (Hex)	DECIMAL	BASE7	DOMINO -->
space	U+0020	32	44	1—0 4—4
!	U+0021	33	45	1—0 4—5
"	U+0022	34	46	1—0 4—6
#	U+0023	35	50	1—0 5—0
$	U+0024	36	51	1—0 5—1
%	U+0025	37	52	1—0 5—2
&	U+0026	38	53	1—0 5—3
'	U+0027	39	54	1—0 5—4
(	U+0028	40	55	1—0 5—5
)	U+0029	41	56	1—0 5—6
*	U+002A	42	60	1—0 6—0
+	U+002B	43	61	1—0 6—1
,	U+002C	44	62	1—0 6—2
-	U+002D	45	63	1—0 6—3
.	U+002E	46	64	1—0 6—4
/	U+002F	47	65	1—0 6—5
0	U+0030	48	66	1—0 6—6
1	U+0031	49	100	1—1 0—0
2	U+0032	50	101	1—1 0—1
3	U+0033	51	102	1—1 0—2
4	U+0034	52	103	1—1 0—3
5	U+0035	53	104	1—1 0—4
6	U+0036	54	105	1—1 0—5
7	U+0037	55	106	1—1 0—6
8	U+0038	56	110	1—1 1—0
9	U+0039	57	111	1—1 1—1
:	U+003A	58	112	1—1 1—2
;	U+003B	59	113	1—1 1—3
<	U+003C	60	114	1—1 1—4
=	U+003D	61	115	1—1 1—5
>	U+003E	62	116	1—1 1—6
?	U+003F	63	120	1—1 2—0
@	U+0040	64	121	1—1 2—1
A	U+0041	65	122	1—1 2—2
B	U+0042	66	123	1—1 2—3
C	U+0043	67	124	1—1 2—4
D	U+0044	68	125	1—1 2—5
E	U+0045	69	126	1—1 2—6
F	U+0046	70	130	1—1 3—0
G	U+0047	71	131	1—1 3—1
H	U+0048	72	132	1—1 3—2
I	U+0049	73	133	1—1 3—3
J	U+004A	74	134	1—1 3—4
K	U+004B	75	135	1—1 3—5
L	U+004C	76	136	1—1 3—6
M	U+004D	77	140	1—1 4—0
N	U+004E	78	141	1—1 4—1
O	U+004F	79	142	1—1 4—2
P	U+0050	80	143	1—1 4—3
Q	U+0051	81	144	1—1 4—4
R	U+0052	82	145	1—1 4—5
S	U+0053	83	146	1—1 4—6
T	U+0054	84	150	1—1 5—0
U	U+0055	85	151	1—1 5—1
V	U+0056	86	152	1—1 5—2
W	U+0057	87	153	1—1 5—3
X	U+0058	88	154	1—1 5—4
Y	U+0059	89	155	1—1 5—5
Z	U+005A	90	156	1—1 5—6
[	U+005B	91	160	1—1 6—0
\	U+005C	92	161	1—1 6—1
]	U+005D	93	162	1—1 6—2
^	U+005E	94	163	1—1 6—3
_	U+005F	95	164	1—1 6—4
`	U+0060	96	165	1—1 6—5
a	U+0061	97	166	1—1 6—6
b	U+0062	98	200	1—2 0—0
c	U+0063	99	201	1—2 0—1
d	U+0064	100	202	1—2 0—2
e	U+0065	101	203	1—2 0—3
f	U+0066	102	204	1—2 0—4
g	U+0067	103	205	1—2 0—5
h	U+0068	104	206	1—2 0—6
i	U+0069	105	210	1—2 1—0
j	U+006A	106	211	1—2 1—1
k	U+006B	107	212	1—2 1—2
l	U+006C	108	213	1—2 1—3
m	U+006D	109	214	1—2 1—4
n	U+006E	110	215	1—2 1—5
o	U+006F	111	216	1—2 1—6
p	U+0070	112	220	1—2 2—0
q	U+0071	113	221	1—2 2—1
r	U+0072	114	222	1—2 2—2
s	U+0073	115	223	1—2 2—3
t	U+0074	116	224	1—2 2—4
u	U+0075	117	225	1—2 2—5
v	U+0076	118	226	1—2 2—6
w	U+0077	119	230	1—2 3—0
x	U+0078	120	231	1—2 3—1
y	U+0079	121	232	1—2 3—2
z	U+007A	122	233	1—2 3—3
{	U+007B	123	234	1—2 3—4
|	U+007C	124	235	1—2 3—5
}	U+007D	125	236	1—2 3—6
~	U+007E	126	240	1—2 4—0

Error Types

The spec doesn't define a way to "catch" errors in a graceful way yet. For now, whenever an error occurs, the program will terminate and the interpreter will throw an error to stderr.

• SyntaxError: Unexpected token '{token}' at line {line}, column {column}
• InvalidGridError: Invalid grid. All lines containing code must be the same length (for now)
• MultiConnectionError: {type} connection at line {line}, column {column} is trying to connect a cell that is already connected
• MissingConnectionError: Non-empty cell at line {line}, column {column} does not have a connection
• ConnectionToEmptyCellError: Connection to an empty cell at line {line}, column {column}
• ConnectionToEmptyCellsError: There are connectors that are not connected to anything (Cannot give you the exact location of the error atm)
• InterpreterError: Something wrong with the Interpreter: {message}
• UnexpectedEndOfInputError: Unexpected end of input at line {line}, column {column}
• AddressError: Address '{address}' out of bounds
• InvalidLabelError: Label {name} is not a valid label
• StepToEmptyCellError: Trying to step from cell {currentAddress} to empty cell {emptyAddress}
• JumpToItselfError: Jumping to itself at address {address} is forbidden as it results in an infinite loop
• CallToItselfError:Calling to itself at address {address} is forbidden as it results in an infinite loop
• UnexpectedEndOfNumberError: Unexpected end of number at address {address}
• EmptyStackError: Cannot pop from an empty stack
• FullStackError: Cannot push to a full stack
• InvalidInstructionError: Invalid instruction opcode {opcode}
• InvalidNavigationModeError: Invalid navigation mode {mode}
• InvalidValueError: Invalid value {value}
• InvalidInputError: Invalid input {reason}
• MissingListenerError: NUMIN, NUMOUT, STRIN or STROUT instructions were called and the DominoScript "runtime" did not provide a way on how to handle input or output

Examples

A list of examples to help you understand the language better.

1. Hello World minimal
2. Hello World Commented
3. Hello World 2D
4. Loop Simple
5. Loop using jump
6. Loop using jump and label
7. Call function by address
8. Call function by label
9. Recursion: Factorial
10. Navigation Mode changes
11. Basic game loop
12. Number Input
13. String Input
14. Import script into another
15. Call imported function
16. Ansi clear screen
17. Using delay
18. Reverse String