Loothing

Fixed

• Council members can see everyone's responses again, not just their own. A guard added in 2.0.41 made the master looter the only client allowed to record a candidate response — every other council member silently dropped peers' votes until their own click went through. Council members now correctly mirror the response state every other client is broadcasting, both before voting officially starts on their copy of the item and after the master looter has called the result. The master looter's authoritative tally is unchanged.
• Vote awards, vote requests, and session-start broadcasts now reach the raid even when the addon channel is congested. Master-looter broadcasts that gate session state on every receiver — VOTE_REQUEST, VOTE_AWARD, VOTE_RESULTS, VOTE_SKIP, VOTE_CANCEL, SESSION_START, SESSION_END, SESSION_INIT, MLDB_BROADCAST, COUNCIL_ROSTER, OBSERVER_ROSTER, ITEM_ADD, ITEM_REMOVE, STOP_HANDLE_LOOT — used to send at default priority. They competed with version-handshake chatter and sync traffic; in a busy raid, an award could sit behind 200 lower-priority messages and never propagate. They now send at the highest priority alongside player responses and vote commits.
• Other addons' channel traffic no longer wedges Loothing's critical sends. When BigWigs, WeakAuras, DBM, or any other addon sent on the WoW addon channel, Loothing was subtracting from its own message-token bucket — eventually driving the bucket so deep into negative that nothing could send for tens of seconds at a time. The bucket is now reserved for tracking Loothing's own send rate; other addons' traffic is tracked separately and informational only. WoW's per-message channel throttle remains the authoritative backstop.
• Roster updates during raids no longer flood the queue. Every time a raider resurrected, vehicled in/out, or otherwise triggered GROUP_ROSTER_UPDATE while the master looter was handling loot, the addon sent four separate broadcasts (session start, MLDB, council roster, observer roster) to onboard any new joiners. In a 19-raid with frequent roster shifts that's a 60+ message amplification on the queue. These now collapse into a single SESSION_INIT broadcast that carries every field; receivers fan out to the same handlers and see no behavioral change.

Hardened

• /lt diag now shows queue depth broken down by priority. The communication panel adds a "Queue (A/N/B)" row showing ALERT / NORMAL / BULK message counts separately, and a "Other-Addon Traffic" row showing the recent rate of non-Loothing addon sends visible on the channel. When queue depth grows, you can now see whether it's session-critical traffic backing up or whether the channel is being saturated by something outside Loothing.
• Reverted the v2.0.42 ALERT-throttle change that turned out to make queue depth worse, not better. Constraining ALERT to the message-token bucket combined with the bucket-poisoning bug above to wedge critical session traffic for tens of seconds. ALERT now bypasses both budgets like it did in 2.0.41 and earlier; backpressure comes from WoW's AddonMessageThrottle response on a real send, which pauses the queue for a defined cooldown — that's the correct feedback loop.