japl
/
japl
mirror of https://github.com/japl-lang/japl.git
1
1
Fork 0
Code Issues Projects Releases Wiki Activity
japl/README.md

9.3 KiB
Raw Permalink Blame History

JAPL - Just Another Programming Language

JAPL is an interpreted, dynamically-typed, garbage-collected, and minimalistic programming language with C- and Java-like syntax.

J.. what?

You may wonder what's the meaning of JAPL: well, it turns out to be an acronym for Just Another Programming Language, but beware! Despite the name, the pronunciation is the same as "JPL".

Disclaimer

This project is currently a WIP (Work in Progress) and is not optimized nor complete. The design of the language may change at any moment and all the source inside this repo is alpha code quality, for now.

For other useful information, check the LICENSE file in this repo.

JAPL is licensed under the Apache 2.0 license.

Project roadmap

In no particular order, here's a list that is constantly updated and that helps us to keep track of what's done in JAPL:

  • Parsing/Lexing
  • Type system (string, integer, float)
  • Builtin collections (arraylist, mapping, tuple, set)
  • Builtin functions (print, etc)
  • An import system
  • Control flow (if/else)
  • Loops (for/while)
  • Comparisons operators (>, <, >=, <=, !=, ==)
  • Casting with the as operator
  • Modulo division (%) and exponentiation (**)
  • Bitwise operators (AND, OR, XOR, NOT)
  • Simple optimizations (constant string interning, singletons caching)
  • Garbage collector (Mark & Sweep)
  • Operations on strings (addition, multiplication -> TODO both sides)
  • Functions
  • Closures
  • Functions default and keyword arguments
  • An OOP system (class-based)
  • Global and local variables
  • Explicit scopes using bracket
  • Native asynchronous (await/async fun) support
  • Bytecode optimizations such as constant folding and stack caching
  • Lambda functions as inline expressions
  • Arbitrary-precision arithmetic
  • Multi-line comments /* like this */ (can be nested)
  • Generators
  • A standard library
  • String slicing, with start:end syntax as well
  • Multithreading and multiprocessing support with a global VM Lock like CPython
  • Exceptions
  • Logical operators (not, or, and, is)
  • Basic arithmetic (+, -, /, *)
  • Optional JIT Compilation
  • Static type checker (compiler module)
  • Runtime Type Checking
  • 0-argument functions without ()

Contributing

If you want to contribute, feel free to open a PR!

To get started, you might want to have a look at the currently open issues and start from there

Community

Our first goal is to create a welcoming and helpful community, so if you are so inclined, you might want to join our Discord server and our forum! We can't wait to welcome you into our community :D

A special thanks

JAPL is born thanks to the amazing work of Bob Nystrom that wrote a book available completely for free at this link, where he describes the implementation of a simple language called Lox

JAPL - Installing

JAPL is currently in its early stages and is therefore in a state of high mutability, so this installation guide might not be always up to date

Requirements

To compile JAPL, you need the following:

  • Nim >= 1.2 installed on your system
  • Git (to clone the repository)
  • Python >= 3.6 (Build script)

Cloning the repo

Once you've installed all the required tooling, you can clone the repo with the following command

git clone https://github.com/japl-lang/japl

Running the build script

NOTE: as of now, you will need version 0.1.0 of the nimble package jale installed:

git clone https://github.com/japl-lang/jale --branch 0.1.0
cd jale
nimble install

As a next step, you need to run JABT (YES, Just Another Build Tool). This will generate the required configuration files, compile the JAPL runtime and run tests (unless --skip-tests is used). There are some settings that can be tweaked with command-line options (or environment variables), for more information, run python3 build.py --help.

To compile the JAPL runtime, you'll first need to move into the project's directory you cloned before, so run cd japl, then python3 build.py ./src and wait for it to complete. You should now find an executable named japl (or japl.exe on windows) inside the src folder.

If you're running under windows, you might encounter some issues when using forward-slashes as opposed to back-slashes in paths, so you should replace ./src with .\src

If you're running under linux, you can also call the build script with ./build.py (assuming python is installed in the directory indicated by the shebang at the top of the file)

Advanced builds

If you need more customizability or want to enable debugging for JAPL, there's a few things you can do.

Nim compiler options

The build tool calls the system's nim compiler to build JAPL. If you want to customize the options passed to the compiler, you can pass a comma separated list of key:value pairs (spaces not allowed) via the --flags option. For example, doing python3 build.py src --flags d:release,threads:on will call nim compile src/japl -d:release --threads:on.

Known issues

Right now JAPL is in its very early stages and we've encountered a series of issues related to nim's garbage collection implementations. Some of them seem to clash with JAPL's own memory management and cause random NilAccessDefects because the GC frees stuff that JAPL needs. If the test suite shows weird crashes try changing (via --flags) the gc option to boehm (particularly recommended since it seems to cause very little interference with JAPL), or regions to see if this mitigates the problem; this is a temporary solution until the JAPL VM becomes fully independent from nim's runtime memory management.

JAPL Debugging options

JAPL has some (still very beta) internal tooling to debug various parts of its ecosystem (compiler, runtime, GC, etc). There are also some compile-time constants (such as the heap grow factor for the garbage collector) that can be set via the --options parameter in the same fashion as the nim's compiler options. The available options are:

  • debug_vm -> Debugs the runtime, instruction by instruction, showing the effects of the bytecode on the VM's stack and scopes in real time (beware of bugs!)
  • debug_gc -> Debugs the garbage collector (once we have one)
  • debug_alloc -> Debugs memory allocation/deallocation
  • debug_compiler -> Debugs the compiler, showing each byte that is spit into the bytecode
  • skip_stdlib_init -> Skips the initialization of the standard library (useful to reduce the amount of unneeded output in debug logs)
  • array_grow_factor -> Sets the multiplicative factor by which JAPL's dynamic array implementation will increase its size when it becomes full
  • map_load_factor -> A real value between 0 and 1 that indicates the max. % of full buckets in JAPL's hashmap implementation that are needed to trigger a resize
  • frames_max - The max. number of call frames allowed, used to limit recursion depth

Each of these options is independent of the others and can be enabled/disabled at will. Except for array_grow_factor, map_load_factor and frames_max (which take integers and a real value respectively), all other options require boolean parameters; to enable an option, pass option_name:true to --options while to disable it, replace true with false.

Note that the build tool will generate a file named config.nim inside the src directory and will use that for subsequent builds, so if you want to override it you'll have to enable --override-config (via CLI, env. variables or build profiles). Passing it without any other option will fallback to (somewhat) sensible defaults

P.S.: For now the test suite assumes that all debugging options are turned off, so for development/debug builds we recommend skipping the test suite by passing --skip-tests to the build script. This will be fixed soon (the test suite will ignore debugging output)

Installing on Linux

If you're on linux and can't stand the fact of having to call japl with ./src/japl, we have some good news! The build script can optionally move the compiled binary in the first writeable entry inside your PATH so that you can just type jpl inside your terminal to open the REPL. To avoid issues, this option is disabled by default and must be turned on by passing --install, but note that if in any directory listed in PATH there is either a file or a folder named jpl the build script will complain about it and refuse to overwrite the already existing data unless --ignore-binary is passed!

Environment variables

On both Windows and Linux, the build script supports reading parameters from environment variables if they are not specified via the command line. All options follow the same naming scheme: JAPL_OPTION_NAME=value and will only be applied if no explicit override for them is passed when running the script

Build profiles

JABT comes with a handy --profile option that allows to pass a JSON file with all the build arguments and flags. The JAPL toolsuite comes with some build profiles by default, they are found inside the resources/profiles directory and are particularly useful to replicate common build options. Note that profiles have priority over command-line options and environment variables!