Impressive stuff. Unfortunately, Lua. From a user point of view (well, at least this user), TypeScript would be a better fit.

What are your misgivings about Lua?

Not OP, but having written some Lua, I can make some guesses:

- Relatively obscure language, so potential contributor base is limited from the start

- Lua shares PHP's "One datatype to rule them all" design, which works but feels ugly. In PHP it's "arrays", in Lua it's "tables", but either way you have all the attendant problems and weird edge cases to learn

- Expanding on above point, Lua's APIs for working with tables are, uh, idiosyncratic. Slicing a "list", last I looked, was an unintuitive monstrosity like `new_table = { table.unpack(old_table, start_index, length) }`

I could keep going, but I have other things to do during my brief time in this universe.

To any Lua aficionados out there, my apologies if I've misrepresented it. Corrections to my misunderstandings will be appreciated.

> Relatively obscure language

Lua is among the most used languages in existence.

It's probably the most used language by under-21-year-olds, and almost certainly so for under 16. Roblox is absolutely enormous.

It's the usual choice of embedded scripting language, so a great number of programmers who don't use it as a daily driver, nonetheless learn it to modify this or that program. Being a very simple, minimalist language, this is easy to do.

It's true that it's missing some affordances which you'll find in larger (and frequently less efficient) languages. You'll end up doing more iteration and setting of metatables. That bothers some people more than others. These are the tradeoffs one must accept, to get a 70KiB binary which fits in an L1 cache while being multiples faster than e.g. Python or Ruby.

Lua is too bare bones. It's a carefully considered tradeoff for the project goals of being a very small & easily embeddable language for C projects, but it means you need to implement too much load-bearing functionality yourself. Each lua project ends up being an ad hoc framework composed of C bindings and standard lib extensions, but it doesn't have an ecosystem that makes this straightforward or consistent.

A lot of people like lua from using it on very small projects, or like, spiritual reasons related to its implementation simplicity. But having used it quite a lot professionally, in practice it ends up being a slog to work with.

All that said, lua is closely tied to C, and embedded in a C project is both its intended use and the place where it fits best. So while I don't really like lua and would almost never choose it myself, this is one of the rare exceptions where I think it's a good choice.

A lot of people want redbean to use TS or python instead, and imo either would make it much bigger, more complex, for relatively little benefit. It would definitely benefit from a more full-featured language, but I think a weirder one that is still intended for embedding in C would be best. Something out there like janet or fuck it, tcl. But lua has a lot of allure for a lot of programmers and I think gives the project a feeling of "old school cool" while still being pretty accessible. So is probably the best choice in the end.

None in particular, except that all the code I am currently writing is in TypeScript. Especially if it comes together with HTML files.

>> I would use redbean, from the same author, and use HTML.

Wait... Could we build Solvespace web version and bundle it into that and have a completely portable CAD program that runs locally but in the browser?