Dominik Charousset

We natively support JSON since it's the most common format. It's a good question where we should draw the line. There are other formats like YAML and INI/TOML that probably...

That document is for schema definition. For an XML API, I believe we wouldn't need that unless we want to also implement validators, no?

`actor_pool` is already deprecated, so it's not worth investing cycles there. I'd rather disable/remove that test suite. The other tests, we should definitely look into. All of them involve robot,...

> Are we axing `actor_pool` for v2? Hm, maybe. Let's see this suite fail again. The robot texts fail quite frequently, so we should prioritize those anyways.

After looking at the sources again, I think I've confused [`delay`](https://www.actor-framework.org//static/doxygen/1.0.0/classcaf_1_1flow_1_1coordinator.html#a465e8ef2d22f06701590183aece33214) and [`schedule`](https://www.actor-framework.org//static/doxygen/1.0.0/classcaf_1_1async_1_1execution__context.html#a5a1c643d1bf2bca1d6ffcc0735193211) in chat. You said you're using `schedule`, right? That will always go through the mailbox. Even when...

> Can you help us understand the `defaults::max_inline_actions_per_run` value? What counts as a "run"? Scheduled actors implement the interface `caf::flow::coordinator`. This is what flow operators run on top of. It...

If you do something like `iota(1).for_each(...)`, yes, you'll end up with a pattern that runs the max. inline actions then pushes an action to the mailbox, rinse and repeat. I'm...

Agree, if it's just a single Clang version that's broken I'd say it's not worth troubleshooting. Especially since Clang 17 only has "initial" support for C++23 anyways.

I think semantics matter here. A generator is not a container. When producing an `observable` from a container, users should reasonably expect a cold observable. Even if generators use (forward)...

Thanks for reporting!