* the destination prediction is now much more reliable
* the movement generator will now more frequently check for valid positions to match retail behavior
* the catch-up mechanic will now adapt its speed to the target's one when being close enough to hit it
* temporarily disabled backwards movement when being too close to a target to avoid some edge cases that need additional research
* fixed the pathfinding reaching behavior so that they no longer follow you to unreachable obstacles and return home when they can't reach you anymore
closes#345closes#260
* added radius event parameter to limit the search range of the executing creature
* moved the pct check to a macro and corrected the selection logic to support all target types
- PvE combat is now always mutual. UNIT_FLAG_IN_COMBAT is backed by actual references to the units we're in combat with.
- PvP combat is now also tracked, and almost always mutual; spells like Vanish and Feign Death can break this rule. That means we can easily determine a list of players we're fighting.
- By extension, IsInCombatWith now has sensible behavior when invoked on nonplayers.
- Threat and combat systems are no longer the same.
- They still have an enforced relationship (threat implies combat - clearing combat clears threat)...
- ...but we can have combat without threat. A creature (with threat list) isn't considered to be engaged until it has an entry on its threat list...
- ...which means we can now faithfully replicate retail engage behavior. Combat on projectile launch - engagement start on projectile impact. Yay for progress!
- AI method refactor, as already ported in 6113b9d - `JustEngagedWith`, `JustEnteredCombat` and `JustExitedCombat`.
- Vehicle threat is now properly pooled on the main vehicle body (fixes#16542).
- Various edge case bug fixes for threat redirects (Misdirection "cancelling" Vigilance and similar).
- Target re-selection is now significantly faster.
- Fixed a ton of other smaller edge case bugs, probably.
Closes#7951 and #19998.
* Combat/Threat rewrite (PR #19930) prep work. Mostly refactors, and a compatibility layer on ThreatManager/HostileReference that allows scripts to be changed already.
Unlike SMART_ACTION_AUTO_ATTACK, this action is actually stopping the entire attack procedure, rather than just hack-disabling auto-attacks. Keep in mind that this action requires a passive react state to prevent a immediate AttackStart ai reaction
* Core/Spells: Implementation of DoT clipping.
* Core/Utilities: Backport "EnumFlags v3.0" from master.
* Core/Spells: Switch AURA_REMOVE_MODE to a flag set.
The number of edge cases in which weirdness is seen on "effect movements" will be kinda reduced, plus consistency, plus movementInform on custom movement spline initalizations.
Core/Movement: use helpers for validation
Core/Movement: crashfix workaround
- Map::RemoveRespawnTime(SpawnObjectType, LowType, doRespawn) split into Map::Respawn and Map::RemoveRespawnTime, without the extra boolean
- Map::RemoveRespawnTime(RespawnInfo*) merged into Map::DeleteRespawnInfo(RespawnInfo*) and is now private
- Map::DeleteRespawnInfo(void) renamed to Map::UnloadAllRespawnInfos to properly describe what it does
- Map::ProcessRespawns now actually saves the delayed respawn time to DB if the respawn was delayed
- Map::AddRespawnInfo now takes const reference, and returns success as a boolean
- Map::AddRespawnInfo no longer offers an unused "replace" parameter
- Map::DeleteRespawnInfo no longer offers a variety of unused private overloads
- Map::SaveRespawnTime no longer offers a tantalizing writeDB parameter. Parameter is now called "startup" to properly describe what it does.
- Map::SaveRespawnInfoDB now takes RespawnInfo reference instead of all the various fields. Still public because compatibility mode. QQ.
- Map::GetWorldObjectBySpawnId sanitized
- Map::GetXRespawnTime methods sanitized to all go through Map::GetRespawnTime
----------
Core/DB: Unify `creature_respawn` and `gameobject_respawn` into a single `respawn` table
Ovahlord