Skip to content

Latest commit

 

History

History
185 lines (129 loc) · 8.25 KB

CONTRIBUTING.md

File metadata and controls

185 lines (129 loc) · 8.25 KB
Raw

Welcome to the Toy Programming Language

I'm extremely appreciative to have people contribute their time and expertise towards this project. Before I get into the details, let me give some high level goals:

  • Toy is intended to work as an embedded scripting language inside a game engine, tentatively titled "Box", but should be general enough to use anywhere.
  • Toy's goal is to allow for a great degree of modability for any games developed with the Box engine.
  • As such, using Toy itself should be easy for novice and veteran coders.
  • Toy is developed with a focus on speed, efficient resource management, and portability.
  • Toy should be a source of fun for everyone working on and using it.

Where to Start

First, the main development branch of Toy is 'v2', with the branches 'v1' and 'v1-docs' kept for reference.

v2 is a ground-up rewrite, with additions, changes and deletions to the language that make sense. If you want to discuss the direction things are going, the Discussions Tab is perfect for that (though, maybe a bit outdated at times).

The Issue Tracker is a good place to see what tasks and issues are currently waiting to be addressed. The toy.h source file is a quick way to see what building blocks are available in the source code. There are also a number of comments prepended with URGENT or TODO scattered throughout the source code, as reminders to myself.

The tests directory, which holds a collection of automated tests for the CI pipeline, can be a good way to see how those parts are used. Likewise, the REPL shows a practical usage of Toy.

v2 is under heavy development, and as such may not be in a working state yet. Your patience and feedback can help, but a documentation website is coming... eventually.

Building Blocks

The flow from source code to runtime is below - Toy_Bytecode is a thin wrapper for Toy_Routine, that produces the final output.

graph LR
    source[source code]@{ shape: flag }
    Toy_Lexer@{ shape: card }
    Toy_Parser@{ shape: card }

    source --->|bound to| Toy_Lexer
    Toy_Lexer --->|bound to| Toy_Parser
Loading 
graph LR
    Toy_Parser@{ shape: card }
    Toy_Ast@{ shape: card }
    Toy_Bytecode@{ shape: card }

    Toy_Parser --->|generates| Toy_Ast
    Toy_Ast --->|compiled to| Toy_Bytecode
Loading 
graph LR
    intermediate[outputted bytecode]@{ shape: flag }
    Toy_VM@{ shape: card }

    intermediate --->|used by| Toy_VM
Loading

The main data structures used at runtime tend to have a top-down dependency graph, as shown below - Toy_Scope isn't shown, as it acts more as a wrapper around Toy_Table.

graph TB
    Toy_Bucket@{ shape: card }
    Toy_Value@{ shape: card }
    Toy_String@{ shape: card }
    Toy_Array@{ shape: card }
    Toy_Stack@{ shape: card }
    Toy_Table@{ shape: card }
    Toy_VM@{ shape: card }

    Toy_Bucket ---> Toy_String
    Toy_Value ---> Toy_String
    Toy_Value ---> Toy_Array
    Toy_Value ---> Toy_Stack
    Toy_Value ---> Toy_Table
    Toy_Bucket ---> Toy_VM
    Toy_String ---> Toy_VM
    Toy_Array ---> Toy_VM
    Toy_Stack ---> Toy_VM
    Toy_Table ---> Toy_VM
    Toy_Value ---> Toy_VM
Loading

Coding Habits

Here's a few coding habits that I use to keep the source code consistent. While I'd prefer contributors to follow these, breaking these guidelines are ok if you need to.

Testing The Language

When adding a new piece of code, it must be thoroughly tested via a test case. If it has multiple features, they should be tested individually, and in combination with each other. Any kind of corner case which can cause an issue on any supported platform must be resolved (I'm happy to help with this, if needed).

Once a feature has been tested on its own, it can be added to or expanded in the integration tests.

This is probably the most important habit listed here. While I'm not too fussy as to how the tests are written, they do need to prove that the code works flawlessly. Toy is intended to be used by others (potentially many others), so please write simple and straight forward tests to ensure correctness.

Tabs, 4 Characters Wide

I use tabs over spaces, with a width of 4. I don't have a linter, please don't make me use one. For those who care, here's my .vimrc:

" Load the defaults
runtime defaults.vim

" my custom stuff
set tabstop=4
set shiftwidth=4

set autoindent
set smartindent

Error Messages

Fatal errors in the source code have this general format:

fprintf(stderr, TOY_CC_ERROR "ERROR: [Info]\n" TOY_CC_RESET);
exit(-1);

The use of fprintf() will ensure the error is written to the console, and allows extra information to be printed - just replace [Info] with the relevant output. These kinds of fatal errors are intended to catch issues with the language itself, rather than errors in the Toy scripts.

In the test cases, the exit(-1) is instead replaced with return -1 to allow main() to clean up that test case, and run others if needed.

Naming Things

To ensure integrating Toy as an external library is easy, any functions that are intended for use should be:

  • Marked with the macro TOY_API
  • Prepended with Toy_
  • Be named with lowerCamelCase
  • Begin their name with a verb

Likewise, publicly visible structures should be:

  • Prepended with Toy_
  • Be named with UpperCamelCase

If you use a macro, ensure it is all uppercase, and prepended with TOY_. If there's a function that is exposed, but is intended to be used within a macro, prepending its name with Toy_private_ is a good idea.

e.g.

#include "toy_common.h"

typedef struct Toy_Thing {
	int member;
} Toy_Thing;

TOY_API void Toy_useThing(void);

#define TOY_USE_OTHER_THING() Toy_private_useOtherThing()
TOY_API void Toy_private_useOtherThing(void);

Data Type Sizes

Within toy_common.h, the macro TOY_BITNESS is defined as either 32 or 64, depending on if Toy is compiled for 32 or 64 bit platforms (if it doesn't recognise a platform, it is set to -1).

Most, if not all the structures within Toy where memory layout is important have some comments indicating their sizes in memory based on bitness. These comments are only a guide, and should NOT be taken as gospel, but they're useful for thinking about how the data is laid out in memory. Spaces are used to line them up.

//Copied from toy_stack.h
typedef struct Toy_Stack { //32 | 64 BITNESS
	unsigned int capacity; //4  | 4
	unsigned int count;    //4  | 4
	char data[];           //-  | -
} Toy_Stack;               //8  | 8

Single and Double Pointers

Double pointers are used quite often, especially when a function performs some kind of operation on a structure itself, rather than just its data. I refer to these as "handles", and are usually named as such.

Directories

The directories in the repository's root have certain intended uses. If you find a folder which is empty except for a file named .gitkeep, leave it be, as that file is simply used to retain that directory within git for later use.

Directory Intended Use
.github Meta information used by GitHub, such as CI workflows and issue templates.
.notes General storage for any kind of scratch notes or reminders, or random bits of source code that aren't needed. Rough ideas and plans are usually found here, but may be outdated.
lib The source directory for various libraries external to Toy, but which can be made available by the host and accessed with the import keyword.
repl The source directory for the default utility provided with the language. It can be used from the command line, and supports several configuration options.
scripts Storage for various example and WIP scripts written in Toy that can be loaded and executed by the repl.
source The source directory for the core of the Toy Programming Language.
tests The source directory for the standard tests. Within, cases/ is used for test cases, benchmarks/ for benchmarking, etc.
tools The source directory for various standalone tools.