Help Guard's Target

[Ο Χάρος]

Hello people,

i've tried to make a code for L2GuardInstance. I wanted add some checks to guard's target list and make them give more hate (focus) on targets that have the proper checks.

Example if a player hit a guard the guard will attack him and by that time a Karma player can move in tower without guard change target and him him.

 

So basically what i did. 

 

In L2GuardInstance.java i added some small code:

public ScheduledFuture<?> getScheduler()
{
	return _schedule;
}

public void setScheduledTask(ScheduledFuture<?> task)
{
       _schedule = task;
}

and then i made a ThreadPoolManager (check every 1250 ms) 

for (PcInstance player : guard.getKnownList().getKnownPlayers().values)
{
     if (player.getKarma > 0)
     {
        guard.setTarget(player);
        guard.dotAttack(player);
     }
{

The threadpool is activated upon @Override addKnonwObject  and stop  at @Override removeKnownObject. (Like bellow)

@Override
public boolean addKnownObject(final L2Object object)
{
      //blablabla

       getActiveChar().setScheduled(ThreadPoolManager.getInstance().scheduledAiFixedRate(new CheckForKarma(getActiveChar()), 1000, 1000);
}

So basically whenever a object enter in knownList, if task == null it will start scheduler. 

 

It didn't work at all.. Any advice or how i should do this? 

Edited June 1, 2017 by Ο Χάρος

[Tryskell]

You should forget the whole idea, Guards already have a AI, and that AI is called every second. L2J is legendary wrong about crafting AI system, no need to spread it more. Moreover you schedule a different task EACH TIME ANYONE is entering inside his range.

 

You only have to tweak AggroList in order first entry becomes second, and the PK becomes the first. I'm lazy to check, but by default the whole system already exists... Guards aggro at sight, probably adding 1 hate. The only problem is the value is too low, so whoever hits him will turn the aggro down. Simply put max value allowed as hate value, if the Ai is correctly done it will choose the new target accordingly his aggroList during the next ATTACK intention.

 

You can also override aggrolist system for guards and add a lock system if you want guards keep aggro no matter what. The lock/release could be done if "most hated" is/isn't a PK. That will avoid things like "confusion" and other operations on aggrolist to work (even if biggest allowed value is probably extremely hard to deny even using aggro skills).

Edited June 1, 2017 by Tryskell

[BeyondScared]

actually its supposed to be like this. guard is not changing aggro unless the target is dead since c4.

[Ο Χάρος]

You should forget the whole idea, Guards already have a AI, and that AI is called every second. L2J is legendary wrong about crafting AI system, no need to spread it more. Moreover you schedule a different task EACH TIME ANYONE is entering inside his range.

 

You only have to tweak AggroList in order first entry becomes second, and the PK becomes the first. I'm lazy to check, but by default the whole system already exists... Guards aggro at sight, probably adding 1 hate. The only problem is the value is too low, so whoever hits him will turn the aggro down. Simply put max value allowed as hate value, if the Ai is correctly done it will choose the new target accordingly his aggroList during the next ATTACK intention.

 

You can also override aggrolist system for guards and add a lock system if you want guards keep aggro no matter what. The lock/release could be done if "most hated" is/isn't a PK.

 

 

actually its supposed to be like this. guard is not changing aggro unless the target is dead since c4.

 

Thanks for respond both.

 

Tryskell i would like to ask since im not familiar with big structure code, where exactly it check the aggro in which file, current i check guardKnownList and l2guardInstance

 

I smell something like L2Attackable as im checking now. but where it has the Next intention and check the aggroList? 

Edited June 1, 2017 by Ο Χάρος

[Tryskell]

Regular guards use AttackableAI (L2AttackableAI on other forks), castle guards use their own system. AI system works based on intentions (ATTACK, CAST, ACTIVE, etc).

 

The generic behavior is handled by AbstractAI and underlayers, but it's overidden on AttackableAI on thinkActive() / thinkAttack().

 

AggroList is stored on Attackable (L2Attackable on other forks) under _aggroList. Useful methods used by AI (at least on aCis) include getMostHated(), addDamageHate(), reduceHate(), stopHating().

 

PS : naming conventions can and will differ.

	protected void thinkAttack()
	{
		final Attackable npc = getActiveChar();
		if (npc.isCastingNow())
			return;
		
		// Pickup most hated character.
		Creature attackTarget = npc.getMostHated();
		
		// If target doesn't exist, is too far or if timeout is expired.
		if (attackTarget == null || _attackTimeout < System.currentTimeMillis() || MathUtil.calculateDistance(npc, attackTarget, true) > 2000)
		{
			// Stop hating this target after the attack timeout or if target is dead
			npc.stopHating(attackTarget);
			setIntention(CtrlIntention.ACTIVE);
			npc.setWalking();
			return;
		}

Edited June 1, 2017 by Tryskell