Experimental language and compiler.

"luv" is the name of the language, "60" is a reference to the original motivation for this project which was "why isn't compiling 60fps".

On Windows x64 (requires python in PATH):

> m d test

On macOS aarch64 or Linux x64 (requires python3 in PATH):

$ ./m d test

1. Fast compilation: Stay in the >= 1M LoC/s range. Currently on a M3 MacBook Air, a silly benchmark is ~2.2M LoC/s to -O1 and >= 10M LoC for --syntax-only (TODO: link, see hyperfine 'out/mr/luvc out/mr/dumbbench.luv')
2. Fast iteration: Integration with a determinism-based ("saved inputs") engine, incremental code update, debugger integration.
3. Low-level feel: Not trying to be much fancier than C. Try to help avoid common problems, but you can definitely still pointer-bash and corrupt memory when you want to. No intricate object model or GC, it's all just bytes.
4. Feels like Python, ergonomics-wise. Built-in list, dict, str, with arena allocation built-ins.

Add:

set runtimepath^=.../luv60/misc/vim

to your .vimrc to get indenting and syntax highlighting.

Note: copied from old compiler prototype implementation, not much of this is implemented in this version yet!

(syntax highlighting is tagged as "python" on GitHub, so not 100% accurate):

• Gather some elements in a dictionary (aka map, unordered_set, dict) mapping strings to strings
• Iterate over all of them, building up a list (aka vector) of strings compile-time string interpolation with fmt.
• Sort the slice, and then print it out.

def int main():
    {str}str mymap = {"zip": "zap", "flim": "flam", "oink": "zoink"}
    []str res
    for a, b in mymap:
        res.append(fmt("key: {} -> value: {}", a, b))
    res.sort()
    print "\n".join(res)
    return 0

Output:

key: flim -> value: flam
key: oink -> value: zoink
key: zip -> value: zap

• Some auto-like type deduction
• Terse list comprehensions
• Nested functions with access to parent variables (no heap allocations though, you can't return helper)

def int main():
    cutoff = 20

    def bool helper(int x):
        return x < cutoff

    all = [i for i in range(100)]
    filt = [i for i in all if i % 2 == 1 and helper(i)]
    print filt
    return 0

Output:

[1, 3, 5, 7, 9, 11, 13, 15, 17, 19]

• structs are just POD
• but, pseudo-methods can be added using on to dispatch on the static type
• various ways of initializing structs

struct MyStuff:
    int a
    bool b

on MyStuff def int get_full_value(self):
    ret = self.a
    if self.b:
        ret = ret * 16
    return ret

def int main():
    MyStuff mystuff1
    mystuff2 = MyStuff()
    mystuff3 = MyStuff(0, false)
    mystuff4 = MyStuff(a=10)
    mystuff5 = MyStuff(a=10, b=true)

    print mystuff1.get_full_value()
    print mystuff2.get_full_value()
    print mystuff3.get_full_value()
    print mystuff4.get_full_value()
    print mystuff5.get_full_value()

    return 0

Output:

0
0
0
10
160