Soni L.

it'd be nice to have this as you can no longer rely on `extern crate` in edition 2018.

What about inline CTCP and escape codes? Also I don't think anyone implements ERRMSG, so you shouldn't. Also any way to expose CTCPs to the user? (e.g. for [this](https://github.com/SoniEx2/CTCP-S))

[ANSI Color Codes](https://en.wikipedia.org/wiki/ANSI_escape_code#Colors) [mIRC/IRCII Color Codes](http://www.mirc.com/colors.html) The bot you're talking about uses mIRC color codes, not ANSI color codes.

you can step through instructions.

I'm not sure how feasible it is to have luac -l -l as part of the debugger. but it'd be kinda nice imo.

I don't care if it gets deleted if you use the wrong key, that's fine. But if you delete it when JS is off...

This is sound. It's just kinda (read: very) weird. The mutability would be held by the "original `Arc`". We manage the "original `Arc`" by encapsulating an `Arc::new_cyclic` inside an `!Clone`...

The `clone_inner` is `-> Arc`, not `-> IntrusiveQCellOwner`. The advantage is that this avoids an unnecessary allocation that's only used for the `QCellOwnerID`. Also note that `Arc::as_ptr` is `usize`-aligned.

The code as above *is* going to work. And the data inside the `Arc` is a `QCell` - which has a `QCellOwnerID` in it so it's never zero-sized anyway. In...

The original `IntrusiveQCellOwner` fully supports being used with "unrelated" `QCell`s. Yes, our use-case is purely to carry multiple VTables around, and `clone_inner` is pretty much limited/designed for that use-case, but...