whole zone trains

[So_1337]

I would ask that you please look at the rogue AA ability Escape as well. It's meant to work in the same way as Feign Death, by clearing the player off of the hatelist of enemies. However, I was just able to verify through testing that it has the same issue as Feign Death does; the client maintains a hatelist of any mobs that attacked it, and overflow melee will lead to that player attacking an enemy that should've been long forgotten.

Here are the steps that I just took to reproduce this. I encourage anyone else to test it as well:

1) Huscok, a rogue on our server, and I as my SK went to SoldungB.
2) I aggro'd Lord Nagafen and started tanking him. Huscok engaged as well.
3) We backed off around 85%, and Huscok engaged Escape, well out of Nagafen's aggro range. I feigned death. Lord Nagafen's hatelist showed Huscok clear from it as well as myself. Hatelists for Huscok and myself showed Lord Nagafen on it still.
4) Still FD, I moved myself to the zone-in and summoned the rogue. I instructed him to start killing kobolds. I went back to Nagafen and waited, still feign death.
5) I confirmed that Nagafen's hatelist was still clear and that Huscok's still included Nagafen. Nagafen was still at around 90% health, and I predicted that we would see Huscok summoned to him shortly. About 12 kobolds later, that came true. One of his final blows on a kobold landed on Nagafen, and it happened just as before.

I can confidently say that through the observations of Starsis, we have a definite way to recreate this. With Huscok's help, we know now that it also affects other classes with means of clearing aggro.

I hypothesize that bards are afflicted with this same exact issue with their Fading Memories ability. I will conduct further testing.

Thank you very, very much for looking into this, Derision. I know a few people who have given up their monks and bards because they could not consistently clear aggro when it certainly should've been done. I'll keep trying to contribute what I can find.

[KLS]

I'm looking into this further.

[So_1337]

Another thought -- The example I've used from the start of this overflow debacle is, let's pretend you have four attacks per round. The third kills the mob, so the fourth goes to the next thing on your hate list. However, how is it that a mob so far away can be hit with a melee attack? I'm guessing that distance is only checked at the beginning of an attack round, not for each individual attack.

This is more for my own understanding than anything, but I was wanting to confirm my understanding.

[KLS]

Distance is checked once, the problem is that it's overflowing for some reason.

[Derision]

Originally Posted by So_1337

Another thought -- The example I've used from the start of this overflow debacle is, let's pretend you have four attacks per round. The third kills the mob, so the fourth goes to the next thing on your hate list. However, how is it that a mob so far away can be hit with a melee attack? I'm guessing that distance is only checked at the beginning of an attack round, not for each individual attack.

This is more for my own understanding than anything, but I was wanting to confirm my understanding.

I instrumented the code with loads of printfs to confirm the overflow to the next mob on the hate list and even saw it in the client chat window. Although I didn't specifically confirm it, I believe, like you, that the distance to the target is only checked at the beginning of the attack round, and not if the target is changed between attacks (in Mob::RemoveFromHateList ) during that round.

[KLS]

I have some changes to both the auto attack and hate list code that I think will correct it. Tell me if it doesn't and I'll dig deeper.

[Derision]

Originally Posted by KLS

I have some changes to both the auto attack and hate list code that I think will correct it. Tell me if it doesn't and I'll dig deeper.

Seems to work well on initial testing. (couldn't reproduce the overflow that was easy to do previously).