[xii69]

What was wrong

Non-player living entities still skip pushing each other every other tick (old CraftBukkit perf). The gate used ticksLived % 2, which is tied to when the mob spawned. Two mobs can both skip or both push on the same tick, and it drifts under activation-range ticking. The TODO literally asked if we had a better check now.

What this changes

Keep the half-rate skip (intentional on this fork — we’re not doing full SPIGOT-1994 removal), but schedule it on MinecraftServer.currentTick so all non-players share one clock. Players never skip.

Why this is needed

Spawn-parity skipping makes mob-mob collision uneven and harder to reason about with EAR. A global tick schedule keeps the optimization without the weird pairing.

Testing

LivingEntityCollisionSkipTest 5/5; server suite green; package OK.

[kalibrieren]

In your PRs, please keep the diff minimal.
That means, e.g. don't create new classes for checks (like here) if you're only using it once - in that case just inline it - we have the // WindSpigot comments.

Also pay attention to code quality & clarity. Overall just avoid useless shit that AI adds that a human wouldn't.

[windcolor-dev]

Isn't it better to base it on this.ticksLived % 2?

That way there's a bit of randomness. With the change you're proposing, every collision would run on the same ticks instead of being spread out

[kalibrieren]

Isn't it better to base it on this.ticksLived % 2?

That way there's a bit of randomness. With the change you're proposing, every collision would run on the same ticks instead of being spread out

True, would likely be better - especially to keep the Vanilla behavior.