Modify region system to prioritize entities about apollo

[Major-]

Also mods/admins.

[lare96]

Question about this, it seems to be using gathering the Players from the entire World rather than in the immediate region? Is this intentional?

    for (Player other : player.getWorld().getPlayerRepository()) {

(in PlayerSynchronizationTask, same thing applies to NpcSynchronizationTask)

[Major-]

Intentional in the sense that apollo has always done it like this, before the region system existed. Now it is present it leaves room for that optimisation.

[lare96]

I wasn't sure whether a PriorityQueue or a SortedSet would be faster so I made a benchmark (loosely) simulating their usages and ran it 5 times to see if results would be consistent

Warming up...
Completed warmup.

Took the PriorityQueue 12435ms for 2_000_000 iterations.
Took the SortedSet 11894ms for 2_000_000 iterations.

Took the PriorityQueue 13304ms for 2_000_000 iterations.
Took the SortedSet 12773ms for 2_000_000 iterations.

Took the PriorityQueue 13394ms for 2_000_000 iterations.
Took the SortedSet 12809ms for 2_000_000 iterations.

Took the PriorityQueue 14255ms for 2_000_000 iterations.
Took the SortedSet 13235ms for 2_000_000 iterations.

Took the PriorityQueue 15055ms for 2_000_000 iterations.
Took the SortedSet 13825ms for 2_000_000 iterations.

Not sure if I did the benchmark correctly or not, but the PriorityQueue seems to consistently take longer than the SortedSet, so I guess the set would be the best collection to use

[Major-]

I want to implement this, but first I want to change the player updating implementation to iterate over all players in nearby regions instead of all of the players. This is obviously going to be a performance cost, so limiting it to the players in view first is probably desirable.

[lare96]

@Major- you planning on doing this? cause I was gonna start tonight

[Major-]

first I want to change the player updating implementation to iterate over all players in nearby regions instead of all of the players.

[lare96]

Will have a framework of this done in a min, want to see what you think

[rmcmk]

Progress?

Ping @garyttierney @Major- @lare96

[garyttierney]

@ryleykimmel hadn't done any more than a PoC. Will push a branch on my fork.

[lare96]

@ryleykimmel I never had a chance to test mine, and it was kinda messy iirc

[Major-]

Right now I don't want this in master as it's basically unnecessary and also (see my comment above)

[garyttierney]

@Major-

My code only checks neighboring regions for players while prioritizing which ones are updated. However, unsure of the implementation details of how players should be prioritized.

[Major-]

• Entities you're fighting
• Entities you're interacting with
• Admins
• Mods
• Friends
• Everyone else, by distance (not sure if this one is too performance-heavy, or if we care enough)

[rmcmk]

This can probably be worked on now as region updating has been completed.

[Major-]

Someone is welcome to work on this if they really want to, but there are definitely more important things to work on (sort by high priority to see what). I'd also like comprehensive before/after benchmarks for this, because apollo's updating performance is top-notch and I don't want it to take too much of a hit.