Another bot problem "sigh"

[Nightrider84]

So bots atleast on the UF client and the RoF,RoF2 client's can't keep up with the clients. They rubber band pretty badly, and even with sow you can just run ahead and turn around and just see them rubbing banding like crazy.

[Kingly_Krab]

I think you're underestimating the difficulty of making bots accurately trail clients, especially with the newest movement speed changes. This isn't a new problem, and likely something that cannot be easily resolved.

[Nightrider84]

No I realize bots are a pain, I figured id just post it here wasn't sure if anybody was aware of it. =)

[Kingly_Krab]

Yeah, not an issue, bots are definitely a pain in the ass. They're a huge improvement over what they were though.

[rencro]

This is a new issue for sure (newish anyways) as my build from Feb 2015 bots follow fine.

Curious, mercs follow fine on current build. Anyways, not sure if the rubber banding is from posupdates not processing correctly but I spent most of the night playing with this. Its beyond my limited abilities.

In Bot::AI_Process I changed speed from "int speed = follow->GetRunspeed();" to "int speed = follow->GetRunspeed() * 3 / 2;" after converting speed to ints in bot::CalculateNewPosition2, and they stay a bit closer but you still lose em on a constant run across a zone due to rubber banding. I changed it to * 3 but thats just silly fast, almost better off making a call to "CalculateNewPosition" instead of "CalculateNewPosition2", which causes the bots to just warp onto you as you move, but that suxs too.

[Furniture]

After trying out bots for the first time in years it seems that they always rubber band when catching up to you and then appear at their normal positions. This wasnt the case in the past so im not sure what happened. Anyone know how to fix this?

[Uleat]

I've had some contact with people that have reported this..with pets as well.


In your case, are we talking z-axis bouncing or actual x-y regression?

There does appear to be some issue with packet updates..but, it's probably not packet loss nor actual movement calculation based.

[Furniture]

I'm talking about x=y axis. The bots follow, but when the player stops, the bots appear to run past where they should stop and then appear back at their intended stop points.

[Uleat]

I'll take a look at the pathing code when I get a chance.

I know that there are specific checks for the last segment of a path and what we do there may be causing that issue (or not...)

[Furniture]

I just want to add that on my build of the server which is only a few weeks old, this is also the issue with pets and all fleeing npcs. Not necessarily rubber banding like that but just general warping and not being at correct position.

[Shin Noir]

Originally Posted by Furniture

I'm talking about x=y axis. The bots follow, but when the player stops, the bots appear to run past where they should stop and then appear back at their intended stop points.

This symptom is probably described here, right?
https://www.youtube.com/watch?v=UvEpZxS2AUc

I don't think it's bot exclusive.

I notice there's an update loop KLS added and made process loop only happen half as much (likely to save packet spam), may be related. But just wanted to know it's been observed.

[Furniture]

Yes, that video shows the issue. I've heard from a player on my server that all recent servers have this issue so i'm glad its not just me.

[Furniture]

I'm guessing the issue may be with this commit: https://github.com/EQEmu/Server/comm...64e442076c22f7

I'm going to mess with it and see if I can fix it.

[Furniture]

undoing that commit seems to do the trick, i havent been able to test with bots yet though but as far as I can see NPCs arent warping to me when I use #aggrozone.

[Shin Noir]

Originally Posted by Furniture

I'm guessing the issue may be with this commit: https://github.com/EQEmu/Server/comm...64e442076c22f7

I'm going to mess with it and see if I can fix it.

Yeah, that's the code I theorized was reason, thanks for finding it.