Rework Event System

[liam-middlebrook]

Considering reworking the event system to write events to a named pipe and then creating another application for polling these events. This would fix the issue where once the gravestone is displayed the game locks up for a moment while the event is processed.

@eatnumber1 what's your opinion on this? Would it be better to just fix the issue given the current fork+exec structure of the system?

[eatnumber1]

Adding a pipe won't remove the blocking writes to it. You can fix the lock-up issue by using non-blocking sockets, which you'll need to do with or without a named pipe.

FWIW, I think there's value in "not fixing what's not broke". Also properly passing complex state (e.g. errors) between process boundaries is non-trivial.

That said, the one real potential for value here is if you can convince upstream to accept patches for publishing events, they're more likely to accept something that just writes to a file than something that writes to a db or random multicast address.