Man, the grammar in the article is pretty bad...
Anyway, I don't think I'll use it until it'll have tree-shaking, but kudos for the effort. Creating a (decent) modern bundler isn't an easy task at all.
Ah, and another thing: referencing the project's root with "~". Really not a fan of it. I won't probably never load a module from the home directory, but this is messing with a widely estabilished convention.
I really don't get this approach to CSS. How is writing "c4 co2 mx1" any good if not for saving a couple of bytes - that could be shaved off by a minifier and a good bundling system anyway?
And if you want to "atomize" CSS like that, why don't you use inline styles and call it a day?
I feel like all the possible semantic approach is lost in the process.
I'm dodging the problem because that's what you probably should do. Deep recursion brings a lot of problems in JavaScript - first of all, the stack's depth.
But you don't seem to even need that stack, since you're ready to use setTimeout, discarding the closure's scope: that's actually a clear sign that you should use another approach, probably iterative. That would give you a good performance boost too.
That's why using setTimeout is generally frowned upon: it's a hack to get over a language limitation that could be solved with other, more efficient techniques.
So I don't recommend using setTimeout, because I don't recommend using recursion in the first place. Tail call optimization, where available, could solve the depth problem but it won't yield a performance as good as an iterative approach.
Nice, didn't know that. Kind of reminds me of Lwip (https://github.com/EyalAr/lwip) but it's entirely in JavaScript. I wonder if it keeps up with the speed.