LilScript could make this library 5-45% smaller vs current bundlers

[yeargun]

LilScript is a new typed language, compiles into optimized compressed js.
Compiles into different js based on your objective (raw/gzip/brotli)
Compiles class/struct into var/let/const or arrays, if possible, and extra tryhardings all around.

• It already compresses several most popular npm packages.
• It also optimizes runtime performance by a very small margin.
• Type safety without glue fixes like const xd = 5: int;
• Its an ongoing effort, but definitely the future direction imho.
• It supports cross compilation, needs more work.

https://yeargun.github.io/lilscript/

Would love to know your opinion. Open for collaboration, always

[mvaligursky]

Are you suggesting we rewrite the whole engine to a new language?
I'll close this as this is unlikely, but please make some arguments where this might be worth a cost, vs potentially only maybe saving 5% of size.

[yeargun]

I wouldn't rewrite it right now. But in some days, yeah why not.

Bun get rewritten with Rust. From experience I can tell that, rewriting packages with LilScript was pretty straightforward.

With opus-5 / fable it is several shottable.

< 1000$
< 4 weeks

I would bet that its 15%+ size savings for this library

Maybe next year it will be

< 800$
< 2 weeks?

Who knows

[mvaligursky]

20% saving is not a bad win.
But we have a large dependency on JS and TS tools as well. Replacing and maintaining this, along with some new language contributors would need to understand is a MAJOR undertaking. And 20% of 500K or less for a typical project is too small win for all this I think.

[yeargun]

yeah definitely agree. it's an ongoing effort actualy. But LilScript idea is exciting to me.

Slowly I would expect/hope more core js libraries to get rewritten with LilScript or something similar. Tbh, I'm contacting people usualy for visibility and getting feedback. Thanks for your opinions

[yeargun]

@mvaligursky btw would love to get a star maybe if you liked the idea/direction of LilScript

I am getting ghosted from lots of people for some reason.
motionjs,
mobx,
and lots of else npm libraries are already cloned with 100% test pass with significant size wins

https://github.com/yeargun/lilscript

[Maksims]

So the whole community, contributors and engine users who read engine code with docs, would need to learn a new language?

This is very outlandish ask that disregards existing users, that would dramatically damage the whole project.

[Maksims]

How long it took to make LilScript yourself?

[yeargun]

3 days part time, it was able to make "npm i marked" smaller with LilScript

Last 3 days, again part time but sol 4.6 / grok4.6 were performed pretty bad on compiler stuff. so 4 days is fair.

[Maksims]

Such a short-lived solution, without battle tested projects, long-term track record of improvements and support - will be almost impossible to convince anyone really to use.

If you honestly believe that your solution can solve real world challenges with bundle size, then I recommend you to provide alternative builds with your solution where it really matters. By simply making builds for every new version released, and some easy way to access it, so library developers will notice. Asking anyone to switch language or large/core process in workflows - is always a massive task, that can be justified only by solutions that are 5x+ better than a previous one in most of the aspects.

[yeargun]

Thanks for your feedback. I'll update you

[yeargun]

I kept updating LilScript compiler, its much robust right now.

Idc libraries much tbh, a subagent of mine created this
https://yeargun.github.io/lil2-react-markdown/

I want clout imma be honest with you, and I am for some reason finding this lilscript thing valuable still..
I can give consultation for your guys for 1k$ and can rewrite your entire codebase with lilscript

[LeXXik]

No, thank you.