mirror of
https://github.com/Bubberstation/Bubberstation.git
synced 2026-01-02 04:52:10 +00:00
1f51bd797e
* Firedoor optimizations Ok so firedoors were eating up a lot of cpu. They were doing this for one main reason. Clearing a fire alarm in an area involved doing a for(var/obj/machinery/light/light in src) loop. For those not in the know, in src in the context of areas involves filtering ALL the objects in an area. This is REALLY nasty. This fix reduced the total cpu of firedoor closing by a factor of 100. I am not kidding Edit: I was kidding, ghil fixed the oversight that caused this in the first place already My change is still good tho, it wasn't good even if only called occasionally I replaced it with a signal that lights register for on area change. I had to do some finicking to make the existing area sensitivity that machines have behave with this, but it's not too messy. I did the same signal treamtment to firealarms to clean them up a bit. I also did some futzing around in firedoor code to make changing behavior inside it easier. * Changes how fire alarms work slightly Rather then fully resetting alarms, and relying on the door itself to close back up, they will force all doors open for 3 seconds (down from a cooldown of 5 because 5 felt way too long) and then drop again if the firedoor has any well, actual problems. This meant adding two new vars, active and ignore_alarms. It also meant sightly changing the meaning of alarm_type, from the alarm that's activating us right now, to our current alarm. I think this is generally positive, since it makes the variable a bit easier to reason about. Oh and I reworked an existing cooldown to make it fit better into this mold. Ah and we can no longer drop atmos problems. This was an issue before, it was possible for a firelock to be in a problem state, but be unable to reflect that because of something that blocked the event reception, but no longer * Pain * Ensures area sensitivity will only be added if we do not already have the trait from innate state