Dominik Charousset

I'd like to agree, but... Let's say they manage to ship in C++20. We need to give our user base a sufficient grace period when it comes to compiler compatibility....

2022 and no networking in sight. C++23 may ship networking or it may not. Probably not. ASIO isn't just sockets and networking, it's a whole concurrency model and it seems...

The natural customization point in CAF is `actor_system_config`. Wouldn't it solve the issue if the logger would ask the system config for log level names that aren't hard-coded? Users then...

I'd like to add a bit more context to this for future reference. > CAF defines the integer representation for the log levels with a step size of 3. This...

I think rather than adding a narrow feature like `delayed_request`, we should rather implement a more generic `run_delayed` and `run_scheduled` function pair: ```c++ void send_request_to_server(std::chrono::seconds retry_timeout) { request(target, caf::infinite, msg).then...

The only hard limit in CAF is that messages use a 32-bit size field on the wire and thus may not exceed ~2GB. But since your individual messages are smaller,...

> The timeout in the call from the parent to the child seems to hold up the main process from exiting. (...) Is this expected? To give a bit of...

> Is there special behaviour required here? No, at least it should report an error in this case but not producing a segfault.

> Does supporting this type of activation make sense? What kind of support would you been looking for? What service are you providing on that socket? - Are you implementing...

> max-threads = 64 @uentity Thanks for sharing! I suppose this isn't running on a Desktop, though. Even "untuned", >10ms for local message passing shouldn't happen. I think the scheduler...