v1.0.1

CooldownCollaborator

v1.0.1 (2026-06-16)

Full Changelog Previous Releases

• Update CooldownCollaborator.lua
  Updating for 12.0.7
• Update CooldownCollaborator_Mainline.toc
  Update for 12.0.0.7
• Update README: fix stale party/raid spellID claim, document tonight's features
• Fix Lane View stacking for non-transitive overlap chains
  The original stacking offset just nudged an icon down once per prior
  overlapping icon, sequentially by array index. That happens to work for
  a single cluster of mutually-overlapping icons, but breaks for a chain
  like A-overlaps-B, B-overlaps-C, A-does-not-overlap-C: C would land on
  the same level as B (since it only overlapped B, not A) while still
  being close enough on the x-axis to visually clip into B. Replaced with
  proper greedy bin-packing: each icon picks the lowest vertical level
  where it doesn't x-overlap anything already placed there, checked
  explicitly rather than inferred from array order.
  Not live-tested (requires 3+ simultaneous Battle Rez or Bloodlust
  cooldowns at staggered remaining times) - flagging for verification next
  session.
• Fix Request Rez picking the dead target as their own rez provider
  Live test: targeted Fenicia's corpse and got "Fenicia, please Rebirth
  Fenicia!" - GetCapabilityStatus's ready-provider list has no concept of
  who the target is or whether a candidate is alive, so it happily picked
  the corpse itself. Added FindUnitToken(name) to resolve a capability
  entry's name back to a live unit token, and skip any candidate who is
  the target or who is also currently dead/ghost.
• Add raid-lead Request Rez action
  Target a dead player and click Rez on the Essentials Bar's Battle Rez
  header (or run /cdc rez) to announce who should rez them, naming the
  specific provider whose Battle Rez is actually ready right now rather
  than a generic call-out. Gated to raid leader/assist (or party leader)
  via CC:CanRequestRez() - SendChatMessage to RAID + RAID_WARNING in a
  raid, PARTY in a 5-man (no warning channel there). Button only appears
  on the Essentials Bar when both a Battle Rez provider exists in the
  capability scan and the viewer has permission to send the announcement.
• Add roster scan, Essentials Bar, Lane View, and consumable buff tracking
  Four roadmap items built together since they share one data model:
  Roster.lua: class-based capability scan (CC.roster), populated on group
  join/roster change. Knows who CAN provide Bloodlust/Heroism/Time
  Warp/Fury of the Aspects or Battle Rez before anyone has cast anything,
  which the existing row list cannot show (it only ever displays a spell
  once a cast has been observed). Class-only heuristic for v1, not
  talent-precise - a future pass could refine with NotifyInspect.
  Essentials.lua: slim always-visible frame showing only the roster-scanned
  capabilities, with ready/cooldown status per provider. Pure renderer over
  GetCapabilityStatus(), no tracking logic of its own. Toggle via /cdc
  essentials.
  Lanes.lua: alternate renderer over the same capability data, icons
  sliding along a bar (0% = just used, 100% = ready) instead of a text row.
  Written from scratch for the multiplayer case (several players sharing
  one lane) rather than ported from cdtl3-dev/Lanes.lua, which is
  single-player and tied to its own AceDB config schema - only the visual
  language and basic stacking-offset idea carried over. Toggle via /cdc
  lanes.
  Consumables.lua: Cauldron/Feast tracking. Structurally different from a
  personal CD - it's a shared buff applied to everyone nearby, not
  something only the caster has, so no cross-client sync is needed at all.
  Detection constraint: aura .name/.spellId are SECRET even on your own
  buffs (confirmed by tonight's Eclipse/DoT crashes), so unknown buffs
  cannot be auto-discovered - only checked by an ID you already know via
  the new /cdc consumable command, polled via
  the safe GetPlayerAuraBySpellID(knownID) lookup with rising-edge
  detection so an already-active buff does not reset its timer every poll.
  Also fixed Options.lua's Default Spells tab filter, which only excluded
  .custom entries and would have wrongly listed new .consumable entries
  alongside the built-in whitelist.
• Fix bogus remaining-time display caused by cross-client clock mismatch
  Live test confirmed sync delivery works (result=0), but Fenicia's Barkskin
  displayed as 36839:39 remaining on Nelnamara's client. Root cause: GetTime()
  is seconds since each client's own process started, not a shared clock.
  Sending a raw GetTime() timestamp as usedAt and comparing it against the
  receiving client's own GetTime() is meaningless - the two numbers come
  from unrelated clocks.
  Changed the wire format to send elapsed seconds since cast (secondsAgo)
  instead of an absolute timestamp. The receiver re-anchors it to their own
  GetTime() on arrival, so every client's state table always uses
  locally-consistent timestamps regardless of how many relay hops a sync
  message took.
• Stop trusting AreOutgoingAddonChatMessagesRestricted, act on send result
  Live testing showed it returning true far more often than expected - not
  just around death/encounters as documented, blocking every sync send
  before it was even attempted. Both test pairs (Nelnamara/Barkskin and
  Nelnamara/Fenicia Rebirth) queued and never delivered.
  Now attempts the real SendAddonMessage call unconditionally and queues for
  retry only based on its actual return code (nil/0 = success, anything else
  = queue and retry). Verbose logging now also prints the restriction check's
  value alongside the real result so we can see whether they actually agree
  or diverge on the next test.
• Queue and retry sync messages dropped by addon-message restriction
  A Rebirth cast on someone who just died never synced - confirmed by the
  sync skipped: addon messages restricted verbose log from the prior fix.
  AreOutgoingAddonChatMessagesRestricted() returns true briefly around
  death/resurrect, not just boss encounters, and the message was simply
  dropped with no retry path.
  Added CC.pendingSyncs queue: a restricted send gets queued instead of
  dropped, and a 2-second ticker retries the queue as soon as restriction
  clears.
• Fix crash: party/raid spellID is secret, not just nameplate/target
  Live test with a second party member crashed 11x with attempted to index
  a table that cannot be indexed with secret keys the instant their cast
  fired UNIT_SPELLCAST_SUCCEEDED on this client - spellID is secret for
  party/raid tokens too, contrary to the prior assumption (corrected in the
  wow-midnight-api memory file). There is no way to read another player's
  spellID off this event at all.
  Only player gives a safe spellID, so RecordCooldown is now only called
  for unitToken == player. Other party/raid members' cooldowns are only
  ever populated via the addon-message sync path (OnAddonMessage to
  RecordCooldownFromComm), which already only carries data each client
  reported about its own safe spellID. Removed the now-unused IsTrackedUnit
  helper.
• Fix silent sync failure: inEncounter flag could get stuck true forever
  SendCooldownSync gated every send on self.inEncounter, which is only ever
  reset by ENCOUNTER_END firing. If that event fails to fire cleanly (wipe
  edge case, zone transition, disconnect mid-fight) the flag sticks true and
  silently blocks all future sync sends for the rest of the session with no
  error output - exactly matching the reported symptom of each party member
  only ever seeing their own Rebirth cast, never their partners, despite
  both having the addon. Removed that gate; the real-time
  AreOutgoingAddonChatMessagesRestricted() check is the actual authoritative
  signal and does not have this failure mode. Added verbose logging to both
  the send and receive paths so any future sync failure is visible instead
  of silent.
• Fix RegisterUnitEvent overwrite bug - zero cooldowns were ever tracked
  RegisterUnitEvent does not accumulate across separate calls for the same
  event; each call replaces the prior unit filter. Looping it once per unit
  token (player + party1-4 + raid1-40, 45 calls) left the frame listening
  only to the last token (raid40), so nothing fired solo or in small groups
  • confirmed via /cdc verbose producing zero output even for trivial casts.
    Switched to a plain RegisterEvent with manual unitToken filtering via the
    new CC:IsTrackedUnit helper.
• Replace Blizzard Settings panel with standalone options frame
  The Blizzard canvas settings panel is protected UI - it locks the spellbook
  regardless of how it is opened (minimap button, slash command, or Escape
  menu), and the game menu reappears on close, re-locking it again. Rebuilt
  the full options UI as a standalone CCOptionsFrame (movable, Escape-closable
  via UISpecialFrames) that opens independently and never touches protected
  UI, matching the pattern used by WeakAuras/ArkInventory/MSCT. The Blizzard
  AddOns category now just hosts a stub panel with a button that opens the
  ...

v1.0.0

CooldownCollaborator

v1.0.0 (2026-06-16)

Full Changelog Previous Releases

• Add CurseForge project ID 1577127
• Remove Wago TOC field; add CurseForge README
• Implement CooldownCollaborator v1.0.0
  Full implementation of group/raid cooldown coordinator for Midnight 12.x.
  Tracks all major defensive and utility CDs for party (5-man) and raid
  (up to 40) using UNIT_SPELLCAST_SUCCEEDED on safe unit tokens.
  Syncs state between CC users via C_ChatInfo.SendAddonMessage after
  encounters. Draggable panel, class-colored rows, countdown timers.
  Data.lua: 28 tracked spells across all 13 classes with base CD durations.
  Events.lua: pre-registers party1-4 and raid1-40 unit event tokens at init.
  Comms.lua: encodes/decodes CD sync messages, guards against encounter lockdown.
  UI.lua: draggable panel with 14-row pool, 0.5s ticker, class color bars.
• Bump Interface to 120007 for patch 12.0.7
• Initial commit: CooldownCollaborator skeleton