[Barrelfish-users] RCK message buffers

[Simon Peter]

> But I don't understand what's happening here. The notification of the
> message being sent is lost when the buffer is overflowed? As far as I
> know there is only one flyby message at a time. So then how the
> continuation closure can be executed in the notification gets lost?

If I understand your test case correctly, then that's not the case. You 
seem to be sending many messages, as fast as possible, on both ends of 
the connection. Maybe I'm mistaken, but it doesn't seem to be a 
ping-pong experiment where you have only one message in flight at any time.

> In any case, it's very good that the payload is not lost and always gets
> delivered. In our communication model that should work.

Please don't adjust your model around this bug. It really just needs 
fixing in the implementation.

Simon