Coroutines

[Simn]

The time has come.

[Simn]

There are still various issues to sort out but the whole thing is going to be in experimental state for a while anyway, plus I kind of promised that we would get coroutines in 2025, so let's just do it!

[back2dos]

Congrats to finally getting this done ;)

I have heaps of questions, but I'll just stick to those I can articulate properly:

1. Are there some "docs" about this? I see some tests and inline docs and suppose a lot can be figured out from there. But undoubtedly a lot of thoughts went into this and it would be great to understand some of them. Perhaps just slap the chat history into an LLM and ask it to summarize :D
2. What's the idea behind the hxcoro haxelib? Are implementations of Scheduler intended to happen in user land? Or is it simply to be able to iterate the runtime(s) faster than the compiler?
3. Totally random minor nitpick: Instead of Int64, might it not be better to use the Int53 proposed elsewhere?

[Simn]

Thank you for your interest in our product!

1. Most of the public types and functions should at least have some documentation by now, but there's definitely more work to do in that area. I've started some general documentation over at https://github.com/HaxeFoundation/hxcoro/wiki/Hello-Coro which will see updates in the coming days.
2. We follow Kotlin in this regard by having only the parts that the compiler actually needs in std, while most of the functionality is in the hxcoro lib. So while Haxe roughly knows what a scheduler is (because continuations need that), how exactly it is implemented is not its concern. This allows us to update most things related to coroutines without updating the compiler.
3. We went with Int64 because by now it has good support on most targets, so it seemed like the obvious choice. This could be changed later if that becomes necessary.