Attract to Sound ([NEO]Forge/Fabric): Sound & Stealth.

Fix a bug with goal priority

v4.1.0 for Fabric

v4.1.0 for Neoforge

v4.1.0

Fix the problem with see-through blocks. Added Player Profile, adding customization to the player detection range. By default, support the Feline origin from the Origin mod.

v4.1.0

• Fix the problem with see-through blocks.
• Added Player Profile, adding customization to the player detection range. By default, support the Feline origin from the Origin mod.

v4.0.9

• Ported v4.0.9 for Forge 1.19.2

v4.0.9

• Performance fix.

• Auto config.

v4.0.9

• Performance Fix
• Auto config

v4.0.9

• Performance fix
• Auto config

v4.0.9

• Performance fix
• Auto update config

v4.0.8 for Neoforge

[1.20.1-4.0.8] - 2025-08-25

Changed

• Sound switching semantics clarified and updated

  • src/main/java/com/example/soundattract/config/SoundAttractConfig.java
    • soundSwitchRatio now uses multiply semantics with ratio in (0.0–1.0]: switch when newWeight > currentWeight × soundSwitchRatio.
    • Updated comments and bounds; default set to 0.5.

• Novelty bonus applied consistently to effective weight

  • src/main/java/com/example/soundattract/SoundTracker.java
    • The effective weight returned for the selected sound now includes soundNoveltyBonusWeight during soundNoveltyTimeTicks.
    • The async candidateRecordById mapping also stores novelty-adjusted weight so async-picked IDs honor novelty.

• Resolve same-soundId ambiguity

  • When multiple candidates share a soundId, keep the one with higher effective weight (including novelty); tie-breaker by closer distance to the mob.

• Config responsiveness tweaks

  • soundScoringSubmitCooldownTicks default is 1 to make async scoring responsive.

Defaults updated

• soundSwitchRatio default: 0.5
• soundNoveltyBonusWeight default: 9.5

Added

• Off-thread sound scoring integration

  • src/main/java/com/example/soundattract/SoundTracker.java
    • findNearestSound(mob, level, mobPos, mobEyePos, currentTargetSoundId) now builds SoundScoreRequest with:
      • Mob position and time
      • Current target sound ID to enforce switch ratio off-thread
      • List of muffled SoundCandidates (range/weight after block muffling)
      • soundSwitchRatio, novelty bonus and window (from config)
    • Async results drained via WorkerScheduler.drainSoundScoreResults() and cached per-mob.
  • src/main/java/com/example/soundattract/ai/AttractionGoal.java
    • Updated call sites (canContinueToUse(), tick(), findInterestingSoundRecord()) to pass the current target sound ID into SoundTracker.findNearestSound(...).

• Dynamic configuration keys (accessed via SoundAttractConfig.COMMON.<key>.get()):

  • soundScoringSubmitCooldownTicks (default 5)
  • asyncResultTtlTicks (default 10)
  • raycastCacheTtlTicks (default 200)
  • raycastCacheMaxEntries (default 5000)

Changed

• Worker scheduler rejection policy

  • src/main/java/com/example/soundattract/worker/WorkerScheduler.java
    • Replaced CallerRunsPolicy with a non-blocking DiscardOldestPolicy-based handler to avoid main-thread stalls when saturated.
    • Added debug logging when tasks are dropped (gated by debugLogging).

• Per-mob async submission control in SoundTracker

  • Added per-mob maps LAST_SUBMIT_GAME_TIME and LAST_CANDIDATE_HASH.
  • Submissions are skipped if within cooldown, async result is still fresh, and candidate set (incl. current target) is unchanged.
  • Async result TTL and submission cooldown are now read from config.

• Raycast cache eviction and gating

  • SoundTracker.applyBlockMuffling(...) cache now stores RaycastEntry(result, gameTime).
  • TTL-based expiry and size-based eviction implemented; values read from raycastCacheTtlTicks and raycastCacheMaxEntries.
  • Honors enableRaycastCache to bypass both cache reads and writes when disabled.

Notes

• All configuration values are looked up dynamically via SoundAttractConfig.COMMON.<key>.get() to reflect live changes.

• Defaults for the new keys match previous hardcoded behavior for compatibility.

• Debug logs are gated by SoundAttractConfig.COMMON.debugLogging.get().

• Added a worker scheduler and refactored mob grouping logic to run heavy math off-thread using snapshots, while applying results on the main thread.

Added

• Worker pool and DTOs
  • src/main/java/com/example/soundattract/worker/WorkerScheduler.java
    • Bounded daemon thread pool (safe defaults; optional config via SoundAttractConfig.COMMON.workerThreads and COMMON.workerTaskBudgetMs if present).
    • Immutable snapshots: MobSnapshot, ConfigSnapshot.
    • Result queue + GroupComputeResult for leader mapping, edge mobs, deserters.

Changed

• Group computation off-thread and result application on tick

  • src/main/java/com/example/soundattract/ai/MobGroupManager.java
    • New submitGroupComputeSnapshot(ServerLevel) to build snapshot (main thread) and submit worker task.
    • New applyGroupResult(ServerLevel, GroupComputeResult) to rebuild uuidToLeader, leaders, lastEdgeMobMap, deserterUuids (main thread).
    • updateGroups(ServerLevel) now uses an early-submit pattern; retains previous synchronous path as fallback.
    • Fixed inner class structure for SoundRelay (kept equals/hashCode inside the class).

• Apply results during server tick

  • src/main/java/com/example/soundattract/SoundAttractionEvents.java
    • In onServerTick(END), drain WorkerScheduler.drainGroupResults() and apply via MobGroupManager.applyGroupResult(...).
    • Preserved existing tick sequence (dynamic cooldown -> sound tracker -> block breaker -> apply group results -> pending goals).

Notes

• All world/entity/goal access remains on the main thread; only pure computations are off-thread.
• Config lookups continue to use dynamic access (SoundAttractConfig.COMMON.<key>.get()). Optional worker tunables are read via reflection with safe defaults.
• Logs are gated behind SoundAttractConfig.COMMON.debugLogging.get().

v4.0.7

• Simple Voice Chat range adjusts dynamically

v4.0.7

• Simple Voice Chat range now adjusts dynamically.

v4.0.7

• Dynamic Simple Voice Chat range.

v4.0.7

• Enhance the simple voice chat integration by dynamically adjusting the range based on the speaker's volume.

v4.0.6

• Support PlasmoVoice mod

v4.0.6.

• Mobs can now break blocks to get to the player or sound source (disabled by default)

• Support PlasmoVoice mod

v4.0.6

• Mobs can break blocks to get to the sound/player location (disable by default)
• Support PlasmoVoice mod.