Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Already on GitHub? Sign in to your account

[PetAI] Pet LOS should not apply to command attack #5613

Open
MrSmite opened this Issue Mar 9, 2012 · 29 comments

Comments

Projects
None yet
Contributor

MrSmite commented Mar 9, 2012

Rev: 9f93681
OS: XP Sp3
DB: 46 + updates

Details

When a player clicks 'attack', pet LOS should not apply, pets should attack the target. This allows the player to target a creature in an instance (while behind a wall, etc.) and use the pet to pull by clicking "attack" and then "follow" after initial aggro.

Pet LOS should only apply for the AI automatic target selection to prevent pets from picking targets on different Z axis or behind walls.

Patch

Removed by MrSmite - issue still needs review and other things in the core fixed first (see discussion below)

Want to back this issue? Post a bounty on it! We accept bounties via Bountysource.

Toshik commented Mar 9, 2012

It will add exploit like attack through the walls.

Contributor

MrSmite commented Mar 9, 2012

That's not an exploit. If you have an NPC targeted, clicking attack should send the pet in.

Try it... Enter a dungeon and click "stay", then run up ahead and select an NPC your pet can't see. Click "attack". Your pet will attack the target. This fix has nothing to do with the automatic AI selection.

I've played a Hunter on retail for several years. You've always been able to send your pet in to attack something as long as you could target it.

We used to use it to pull the trash in BWL by hiding behind walls / doors and targeting a mob in another room and sending the pet in. Once aggro initiated we'd pull it back to the group.

Toshik commented Mar 9, 2012

Try another case:
target unavailable NPC by command (/target <npc_name>) and then send pet command to attack. With your patch pet will attack unavailable mob and bring it to you through the wall.
It is definitely an exploit...

Contributor

MrSmite commented Mar 9, 2012

Sorry, it is not an exploit.

Go on retail and try it. Target a mob in Westfall, then run behind a building. Click "attack" and your pet will attack as it should. Running through walls is a different problem entirely.

Contributor

MrSmite commented Mar 9, 2012

Look, here's an example where a Hunter uses Eagle Eye to target a mob out of LOS and then sends his pet to attack:

http://www.youtube.com/watch?v=z59v6jDgx1Q

As I said, this is how it's always been on retail. Pet pulling used to be one of the things Hunters were valued for in instances.

Toshik commented Mar 9, 2012

We need mmaps to have such behavior.
I mean cases when hunter actually can't reach mob at all, e.g. it is behind the wall in some instance with closed doors. But hunter may obtain target and send pet to attack mob through the wall...

Contributor

MrSmite commented Mar 9, 2012

No, mmaps has nothing to do with LOS.

That's supposed to be what dynamic vmaps are for which is already implemented. If it exposes something else that's broken then that needs to be fixed, leaving this broken should be considered a "hack" by TC standards since it isn't Blizz like.

Toshik commented Mar 9, 2012

Mmaps are needed to prevent pets/mobs from going through the walls/obstacles. If path from pet to target could not be calculated - so pet should not atack.
Now it is impossible.
So switching off LOS for pet attacks will cause "trinity-specific" exploit described before.
I understand that it is blizzlike, but attacks/movement through the obstacles are much more unblizzlike :)

Contributor

MrSmite commented Mar 9, 2012

The LOS can be turned off in the CONF already but it turns it off for all aspects of the PetAI which is not correct.

Leaving something in that is not blizz-like is supposedly against TC standards. Like I said, if it exposes another problem then a proper fix should be implemented for that problem.

Contributor

MrSmite commented Mar 9, 2012

Look, if you're worried about the /target command then the proper fix would be to make it so /target doesn't select NPCs that are not in LOS.

Also, mmaps is not needed for this functionality. The dynamic vmaps is supposed to handle this type of problem, it isn't a pathing issue it is a targeting issue.

Toshik commented Mar 9, 2012

Maybe I am wrong, but /target command is client-side command, isn't it?

Generally I' not against your patch, but regarding current situation i think it should be configurable, to prevent mentioned behavior.

Contributor

Star-lion commented Mar 9, 2012

it IS mmaps that matters in this case not dynamic vmaps. dynamic los is TOTALLY different from automated pathing movements.

This seems like a reasonable fix to me. If it can be exploited because of missing mmaps or other weaknesses in the core then so be it. Wouldn't exposing those weaknesses make working on mmaps or fixing the real problem easier?

Contributor

MrSmite commented Mar 9, 2012

@kandera: This isn't about automated pathing.

The problem is that the target should never be selected in the first place which has to do with vmaps and LOS. Using mmaps as the excuse is the same as putting a bandaid on the symptoms rather than repairing the root problem.

Owner

Aokromes commented Mar 10, 2012

Sorry, but retail don't stops from /target or use tab on out of los mobs.

Contributor

MrSmite commented Mar 10, 2012

"Sorry, but retail don't stops from /target or use tab on out of los mobs."

And retail doesn't stop the pet from being commanded to attack a target that's out of LOS as shown in the video I linked.

So what's the problem? This wouldn't be an "exploit", it would be like retail which is supposedly what TC is all about. If you're worried about attacking through walls, etc. then the proper fix would be to fix that, not break something else in the interim.

I've had patches denied because they were considered "hack" fixes and "there must be a better way". Wouldn't the same stand for the attacking through walls? Why allow a "hack" fix of applying an LOS check where it doesn't belong instead of fixing the problem itself?

Owner

Aokromes commented Mar 10, 2012

Well, like you was said, MMAPS is needed to fix this bug on proper way, until them you don't go to see this setting disabled because obvious reasons.

taigarn commented Mar 10, 2012

http://www.warcraftmovies.com/movieview.php?id=164633 minute 7.50 and 10.15, you can notice that pet is send to attack players out of los

Contributor

MrSmite commented Mar 10, 2012

"Well, like you was said, MMAPS is needed to fix this bug on proper way, until them you don't go to see this setting disabled because obvious reasons."

If this setting doesn't fix the bug in a "proper way" then by TC standards it shouldn't be in the core. I guess hacks are OK in some cases but not others. Maybe you should outline when exactly a hack is acceptable because this one breaks an important function of the Hunter class - pet pulling.

Maybe if people were really trying to run a blizz-like server they'd ban people for exploiting instead of putting things in the core that don't belong.

@MrSmite MrSmite closed this Mar 10, 2012

@MrSmite MrSmite reopened this Mar 11, 2012

Contributor

MrSmite commented Mar 11, 2012

Reopened for the future, didn't mean to click close in the first place.

Still bugged. I applied this patch but didn't work.

Contributor

MrSmite commented Apr 25, 2012

@johndays

A lot has changed since this patch so it probably doesn't apply. Also, as you can see from the discussion that there is disagreement as to wether this should be applied at all.

@ MrSmite Thanks for the reply.

Yeah, i know, is some way to allow pets to attack through LoS? I added a if in CheckCast but when they cast, nothing happens and the cd is activated.

Any news ?

Btw. pet follow works properly - when master is not in LoS of pet and he commands pet to follow him, it works.

@MrSmite MrSmite referenced this issue Oct 13, 2012

Closed

Core/Pet: #8043

@Star-lion Star-lion was assigned Nov 25, 2012

This is the problem that pet ignore the door. Pet shouldn't ignore the door!

Member

tkrokli commented Oct 1, 2014

Can anyone confirm if this still happens on new core and with Mmaps + LOS enabled?

Contributor

Trisjdc commented Oct 1, 2014

@tkrokli it's redundant that it still happens, since it solves exploits that are not solved due to GOs not being incorporated in mmaps. This can't be solved until that's done.

Member

jackpoz commented Oct 2, 2014

We need mmaps to have such behavior.

@Toshik @MrSmite can we have the fix now that we have working MMAPs ?

Summing up: ignore LOS but don't ignore MMAPs, if pet can't reach the npc it will not attach it, if pet can reach the npc it will attach it even if beind walls/trees.

We can leave doors out of the discussion of this issue.

@Star-lion Star-lion was unassigned by jackpoz Oct 2, 2014

Contributor

MrSmite commented Oct 4, 2014

I'll see if I can find the original patch and try to rework it for current sources.

@DDuarte DDuarte added the Sub-Exploit label Jul 1, 2016

@Aokromes Aokromes changed the title from [PetAI] Fix - Pet LOS should not apply to command attack to [PetAI] Pet LOS should not apply to command attack May 31, 2017

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment