Fix public quest stages failing to spawn, hanging, and showing a 0.00 reset timer

[Chris443C]

Four related fixes to the public quest system, found while bringing a local server up against the published database.

1. PQuestObjective.Reset() — one bad row aborts the whole stage

When a Type 2 spawn's gameobject proto was missing, the method returned instead of skipping the row. Spawns are iterated in list order, so a single unknown entry silently prevented every creature listed after it from spawning.

The visible symptom is a stage that comes up with a partial mob set, no in-game error, and no corpses — including cases where the mob required to complete the stage is the one that never appears. The Type 1 branch immediately above already used continue for the equivalent case; this makes the two consistent.

The log line now names the entry and objective, so the underlying data gap is diagnosable rather than silent.

2. PublicQuest.AdvanceToNextStage() — QUEST_PROTECT_UNIT stages could hang forever

Protect stages were excluded from arming the Failed event. If the protected unit died, the stage could neither complete — credit comes from PQuestCreature.Protected(), which never fires once the unit is dead — nor time out. The quest stayed stuck until a GM ran .pq reset.

The timer can't simply be armed at Stage.Time: PQuestStage.Time is copied from PQuest_Objective.Time in the constructor, which is the same value the protect credit fires on, so arming it there makes failure race completion at the identical instant. PROTECT_FAIL_GRACE turns it into a backstop so a surviving unit always wins.

3. PublicQuest.Failed() — client showed "resetting in 0.00"

Failed() scheduled Reset correctly but never refreshed _stageTimeEnd, which is what SendCurrentStage renders as the countdown. End() already does this. Without it the field still held the already-elapsed stage deadline, so the client displayed 0.00 for the entire TIME_PQ_RESET wait. Behaviour was correct; only the display was wrong.

4. KeepCommunications.SendKeepStatus() — NRE during startup keep locking

innerDoor.GameObject is only created once a door is spawned into its region, so it is null while keeps are locked at startup. innerDoor was already null-checked; this extends the guard to its GameObject.

Testing

Verified on a local server against the published war_world dump:

• Before: PQ 165 (Raven Host Vanguard) stage II spawned 8 of 20 creatures, omitting Father Sigwald — the protected unit — because two Type 2 entries earlier in its list have no gameobject proto. The stage was unwinnable.
• After: the full spawn set appears and the stage completes normally.
• Failure path exercised by letting the protected unit die: the stage now fails and resets with a correct countdown instead of hanging.

Changes are limited to those four sites; no configuration or data files are included.

🤖 Generated with Claude Code