This implements a solution that preemptively exits the tick loop if we have already marked the connection as disconnecting. This avoids changing the result of Connection#isConnected in order to avoid other possibly unintentional changes. Fundamentally it should be investigated if closing the connection async is really still needed.
This also additionally removes the login disconnecting logic for server stopping, as this was also a possible issue in the config stage but also shouldn't be an issue as connections are closed on server stop very early.
Additionally, do not check for isConnecting() on VERIFYING, as that seemed to be an old diff originally trying to resolve this code, however isConnected is not updated at this point so it pretty much was useless.
We are already using the dropped stack to determine the type, we
might as well also use it for the count, given that plugins can already
mutate the type, might as well let them mess with the amount.
Fixes#12745
That cooked is mildly cooked and should be fixed differently, e.g. a custom loot context param, but reverting the recent change in this line works for now
When converting entity/poi data (SimpleRegionStorage),
we need to insert the new data version so that force upgrading
will write the data back with the correct version.
The vast majority of the time, there are no tasks scheduled to
the EntityScheduler. We can avoid iterating the entire entity list
by tracking which schedulers have any tasks scheduled.
If the player dies, then they are removed from the world and
as a result are not present in the world entity map.
To guarantee that the player entity scheduler is ticked, we can
tick all schedulers for players in the server player list, and
then skip all players we find in the world entity map.
This problem is not present on Folia since Folia must guarantee
that the player remains in the world.
The ServerLevel#getAllEntities function only returns entities which
are accessible. FeatureHooks#getAllEntities will return all
entities, whether or not they are accessible.
Use the new hook in the EntityCommand, which allows server admins
to inspect entities in unloaded chunks.
Use the hook as well for ticking the EntityScheduler. This fixes
an issue whether unloaded entities did not have their scheduler ticked.
Send more keepalives, record all transactions within the last minute.
We send more keepalives so that the latency calculation is more
accurate. Since we send more keepalives, we track all pending
keepalives in case multiple end up in flight.
Additionally, replace the latency calculation with a true
average over the last 5 seconds of keepalive transactions.
We add the change to guarantee that the player position moves
in-sync with the vehicle's. As a result, it should be teleporting
the player using the new position of the vehicle rather than the
old.
If you are creating a click callback using e.g. ChronoUnit.FOREVER.getDuration() this code will throw an ArithmeticException because toNanos overflows. The only way toNanos throws this exception is if the nanos overflow, so we can just safely cap it here as the max value for a long.
Note: This cannot happen on the Fabric/NeoForge versions since
async ticket level processing is not allowed, but can happen on
Paper. This change is made here so that Paper can
remain in sync.
Ticket level decreases may be handled asynchronously when the
off-thread invokes processTicketUpdates() when the main thread
is running ChunkHolderManager#tick(). This is because the ticket
update is queued during tick(), but not executed (invoking
processTicketUpdates) until after releasing the ticket lock.
This creates a small window for an off-thread to invoke
processTicketUpdates() and steal the update.
When the update is stolen, the full chunk status update (if any)
will be eventually queued to execute via the chunk task queue.
If the chunk queue is processed during the server tick at any
point other than the ChunkHolderManager tick, then any ticket
level decrease will violate an important invariant in the
Moonrise chunk system: ticket level decreases only occur during
ChunkHolderManager tick. This invariant exists to make interfacing
with the chunk system easier, especially working with off-thread
contexts.
This change is specifically made to work towards fixing
https://github.com/PaperMC/Folia/issues/363
