Skip to content

Latest commit

 

History

History
163 lines (130 loc) · 4.6 KB

README.md

File metadata and controls

163 lines (130 loc) · 4.6 KB

CI

About Gb

Gb is a project I created to learn more about the inner workings of programming languages. While it is not fully functional, it has achieved this goal and (usually) works within its extremely constrained scope.

Why Gb?

Gb is intended to be a lightweight scripting language that combines the type system of Rust with the ease of use of Python. Combine a familiar syntax with an automatic command line argument parsing (not implemented), Gb should be perfect for small CLI automations in your daily workflow.

Does it work?

Yes! The body of the interpreter is finished and the language can be used for simple calculations. The standard library is extremely sparse and error handling and messages can often be improved. Also, there is currently no repl nor anyway to use it without writing at least a touch of Rust (unless you want to write your entire project in "file.gb", in which case, cargo r is all you need!).

It's clear that this project is still in its infancy, but it should see more and more progress as time goes on.

Features

So, What makes Gb unique? Here are a few present and planned features on the docket for Gb.

  • Top level code with auto-executing main function (if present)
  • Strong, inferred typing (not implemented)
  • Automatic parsing of command line arguments (not implemented)
  • First class support for interpreted, JIT, and compiled modes (1/3 complete)
  • Algebraic Data Types (not implemented)

Code Sample

Gb is a very familiar language in its very early stages. While it looks largely unassuming now, it should gain additional character as development progresses.

fn main() {
    let x = 0;
    while x < 10 {
        x = x + 1;
    }
    std.print(x);

    let foo = fn() { std.print("Hello, foo!"); };
    
    x = 0;
    while x < 10 {
        foo();
        x = x + 1;
    }
}
Outputs:
> 10
> Hello, foo!
> Hello, foo!
> Hello, foo!
> Hello, foo!
> Hello, foo!
> Hello, foo!
> Hello, foo!
> Hello, foo!
> Hello, foo!
> Hello, foo!

Recursion!

fn main() {
    std.print(fib(10));
}

fn fib(n) {
    if n == 0 {
        return 0;
    } else if n == 1 {
        return 1;
    }

    return fib(n - 1) + fib(n - 2);
}
Outputs:
> 55

Optional Main Function

Gb will automatically execute any top-level function named main, but this is not required. The following is just as valid as the previous examples.

std.print("Hello, World!");
Outputs:
> Hello, World!

Working with multiple files

Gb automatically attempts to load a file when namespace lookup fails. Folders are currently not supported. In the future, you will be able to alias paths to types.

// foo.gb
fn foo() {
    std.print("Hello from foo!");
}
// file.gb
fn main() {
    foo.foo();
}
Outputs:
> Hello from foo!

Why not use a parser generator?

Since Gb was primarily a learning exercise, I wanted to try and do as much as possible of it from scratch. Armed with only my programming languages textbook and some serious willpower, I created Gb by myself as an extracurricular project. There are a few branches where I have experimented with using parser generators (like Pest), but ultimately, these efforts weren't as fun for me as it was to build it from the ground up.

The Process

Gb has been built and rebuilt a few different times. The current iteration is based on Thorsten Ball's book Writing an Interpreter in Go, which ended up being quite a convenient resource because the syntax for Monkey was nearly identical to what I had planned for Gb. But the initial versions were built entirely on my own.

In specific, this branch represents the work I was able to put in with a parser generator. As mentioned above, I ended up being pretty unhappy with this approach. I especially disliked how difficult it was to hook into the Pest error system to emit useful error messages from Gb.

It's also worth checking out this branch, which was the version I had before I began to use the Writing an Interpreter in Go version. I learned the most from this version and ultimately decided to rewrite it again only because I wanted to switch to a fancier Pratt-parser rather than the grammar-based parser of this version.

The future of Gb

I occasionally resume work on Gb. My plan is to one day make it a sleek scripting language with support for interop with Rust. As of now, development of the project has taken the back seat to my studies in college.