This case is supposed to not send anything at all, only rescheduling the event to send a NACK when we are willing to do that. The early out was missing, causing a (pure) ACK to be generated and a failing assertion in the code updating the state of last ACK sent because it was supposed to be unreachable.
In a release build it should not cause more damage than a sometimes sending ACK where sending nothing was intended; in a debug build it would instead crash.
(Why did I not spot this before merging #2033 ... 😢)
This case is supposed to not send anything at all, only rescheduling the event to send a NACK when we are willing to do that. The early out was missing, causing a (pure) ACK to be generated and a failing assertion in the code updating the state of last ACK sent because it was supposed to be unreachable.
In a release build it should not cause more damage than a sometimes sending ACK where sending nothing was intended; in a debug build it would instead crash.
(Why did I not spot this before merging #2033 ... 😢)