minecraftToBukkit() Poor Performance

Expected behavior

getActivePotionEffects should be fast

Observed/Actual behavior

its slow, possibly because of papers registry implementation, seems to add alot of overhead

Steps/models to reproduce

Call getActivePotionEffects often, i.e in an anti-cheat

Not allowed to link to spark, here is screenshots for example (test it yourself)

Plugin and Datapack List

N/A

Paper version

This server is running Paper version 1.21.4-DEV-main@f0388e2 (1970-01-01T00:00:00Z) (Implementing API version 1.21.4-R0.1-SNAPSHOT)
You are running the latest version
Previous version: 1.21.4-DEV-1a7288a (MC: 1.21.4)

Other

Maybe we do a cache? Kinda like blockdata conversion

[Warriorrrr]

Provide a link to the spark profile or the actual profile file itself. Not sure why you wouldn't be allowed to link to a spark, but if we can test it and see it ourselves, then surely you can test it yourself too.

Provide a link to the spark profile or the actual profile file itself

Sorry, the owner doesn't let me for "security" reasons (pretty dumb)

heres some java code you can use to replicate the issue (just an example of reading player potion effects in a hot method)

 private static final Map<UUID,AntiCheatData> DATA_MAP = new ConcurrentHashMap<>();

 @EventHandler
    public void onPlayerMove(PlayerMoveEvent event) {
        Player player = event.getPlayer();
        AntiCheatData data = DATA_MAP.computeIfAbsent(player.getUniqueId(),k -> new AntiCheatData());
        Collection<PotionEffect> potionEffects = player.getActivePotionEffects();
        data.activeEffects = potionEffects;
    }
    
    class AntiCheatData{
        private Collection<PotionEffect> activeEffects;
        
        public Collection<PotionEffect> getActiveEffects(){
            return activeEffects;
        }
    }

[kennytv]

We don't accept performance reports only containing spark screenshots, especially if the screenshot doesn't show it being slow. I will update a few lines of key handling somewhere in there that look like odd code in general, but this is still an invalid issue

We don't accept performance reports only containing spark screenshots, especially if the screenshot doesn't show it being slow. I will update a few lines of key handling somewhere in there that look like odd code in general, but this is still an invalid issue

I provided code to replicate this issue, simply call getActivePotionEffects, the spark report is only for illustrative purposes.

[electronicboy]

minimal repo code does not include the entire context, we're not chasing bunnys here

minimal repo code does not include the entire context, we're not chasing bunnys here

what context do you need? the method org.bukkit.Player.getActivePotionEffects() is slower than what I believe is reasonable, due in part to papers registry changes, minecraft holder to bukkit method in particular needs some kind of cache or different approach