add tests demonstrating use of co_await, co_yield, etc., which is now available via clang/llvm

[unicomp21]

No description provided.

[unicomp21]

Hopefully we can deprecate Asyncify.

[unicomp21]

https://blog.panicsoftware.com/your-first-coroutine/

[kripken]

The new c++20 support for coroutines is orthogonal to Asyncify - we need a way to implement coroutines in wasm, and currently LLVM doesn't have one. Asyncify might be a way to implement them, and we might import it to LLVM at some point, or maybe some other approach will be proposed, but I haven't heard of one.

[kripken]

As for getting c++20 coroutines working with Asyncify, that should be pretty easy to do, but I don't know if anyone is working on that atm. The first step might be to get the C coroutine API working, then use that for C++.

[Akaricchi]

I'm very much interested in getting any kind of coroutine support working with emscripten; unfortunately I don't have nearly sufficient understanding of the internals to implement it myself. It doesn't help that the fastcomp asyncify and empterpreter implementations are both broken in different ways. I figured these are not maintained anymore, so I'm not sure if its worth reporting the bugs.

[kripken]

@Akaricchi

Yes, fastcomp-asyncify and emterpreter are probably not worth filing bugs on, as we're focused on the new Asyncify.

If you're interested to work on coroutines there, I can help explain any parts of emscripten and Asyncify that you have questions about! :)

[unicomp21]

I'll see if I can test a simple test going ...

[unicomp21]

My humble apologies for being so arrogant, didn't fully understand Asyncify. Wow, this rocks!

Is there a future within webassembly where we can track cache-line faults and schedule coroutines accordingly?