PPather and fences - did it ever work?

[Hawker]

The babbot team have released a working version of the ppather nav system. I've implemented it today and it works pretty well.

SourceForge.net Repository - [babbot] Index of /trunk/Caronte

One thing I've noticed is that it can navigate buildings, stairs, etc. easily but the fence around the church in Deathknell traps it every time.

Does anyone know if this is a limitation of the reader itself? Or is it just an error in the update for wotlk that I need to track down and fix?

[naa]

As long as i have been using PPather it has always gotten stuck on the fence in AB.

[Hawker]

As long as i have been using PPather it has always gotten stuck on the fence in AB.

Ha! Mine preferred to swim under the BS bridges.

I often went to bed, letting the Undead 1-20 task run and I'd wake up with a toon in the teens. So I assume the stuck detector was a way to deal with this? Or can you enlighten me more?

[Ugge]

It has always been that way. M2 models are misrotated.

I took a quick look just now. I think line 610 in MPQTriangleSupplier.cs:
float dir_y = mi.dir.y - 90;
is supposed to be:
float dir_y = mi.dir.y + 90;

I havent tested it though, it an educated guess. Because thats the value i used in my mpq renderer.

[Hawker]

It has always been that way. M2 models are misrotated.

I took a quick look just now. I think line 610 in MPQTriangleSupplier.cs:
float dir_y = mi.dir.y - 90;
is supposed to be:
float dir_y = mi.dir.y + 90;

I havent tested it though, it an educated guess. Because thats the value i used in my mpq renderer.

Tested it and it did the same fence running exercise. Do you have a URL on how to render the M2 correctly?

[barthen]

The babbot team have released a working version of the ppather nav system.

Mmm, Caronte... Looks familiar:
my crappy ppather patch

GPL rulez

Anyway, if any of the babbot developers is reading this post, I'd like to point that:

• You can safely delete the "Helpers" folder as it isn't used anywhere in the code
  
• There are newer betas of PPather which are faster
  
• I think you're better off throwing this code away. Really, I'm talking from experience. In the open it works fine, but in complicated environments it sucks really hard :yuck:
  

And good luck with babbot, the code looks pretty well done.

[Hawker]

Mmm, Caronte... Looks familiar:
my crappy ppather patch

GPL rulez

Anyway, if any of the babbot developers is reading this post, I'd like to point that:

• You can safely delete the "Helpers" folder as it isn't used anywhere in the code
  
• There are newer betas of PPather which are faster
  
• I think you're better off throwing this code away. Really, I'm talking from experience. In the open it works fine, but in complicated environments it sucks really hard :yuck:
  

And good luck with babbot, the code looks pretty well done.

Well as I have it working, I have to ask, where are these new faster versions please?

Also, as you say GPL rules. Many thanks for your contribution - even if you think its crappy I've enjoyed watching it work.

[barthen]

Well as I have it working, I have to ask, where are these new faster versions please?

Also, as you say GPL rules. Many thanks for your contribution - even if you think its crappy I've enjoyed watching it work.

You can download them here:
Files Section - Pather

[Hawker]

You can download them here:
Files Section - Pather

Instructions

1. Download or update Glider to 1.8.0

Looks like your "crappy" patch is the bleeding edge release!

Thanks anyway.

[Ugge]

Hawker: Did you delete the bin-files in between tests?
Info is available here: WoW.Dev Wiki:Community Portal - WoW.Dev Wiki

[Hawker]

Hawker: Did you delete the bin-files in between tests?
Info is available here: WoW.Dev Wiki:Community Portal - WoW.Dev Wiki

I deleted the pathinfo folder and still the toon ran precisely at the fence between Shallow Grave and Shadow Priest Sarvis in the Deathknell church.

I've something like 5 different ppather code bases and in all of them its + 90. If you are right, this is an old bug and there may well be more like it.

Thanks anyway - at least it does generate a path and I can force it to use certain spots.

            if (!Equals(null, path) &&
                    path.locations.Count > 0)
            {
                int a = 0;

                foreach (Pather.Graph.Location loc in path.locations)
                {

                    if (a % 3 == 0 &&
                        a < path.locations.Count)
                    {
                        ClickToMove(path.locations[a]);
                        Thread.Sleep(1300);
                        // crude stuck management
                        Pather.Graph.Location mySpot = new Pather.Graph.Location(Me.Location);
                        if (mySpot.GetDistanceTo(path.locations[a]) > 5)
                        {
                            ClickToMove(path.locations[a - 2]);
                            Thread.Sleep(505);
                            ClickToMove(path.locations[a - 1]);
                            Thread.Sleep(505);
                            ClickToMove(path.locations[a]);
                            Thread.Sleep(505);
                        }
                    }
                    ++a;
                }
            }

Note: the modulus 3 is an attempt to reduce the number of clicks. I can't get ppather to deliver less waypoints yet.

[risyer]

Pather doesn't load most of the M2 models post 3.0.9 since blizzard fixed the ADTs and corrected the file names. In 3.0.9 they where all ending with .mdx. Now most of the ADTs are fixed and the model file names have the correct .m2 ending.

change:

// change .mdx to .m2
string file=path.Substring(0, path.Length-4)+".m2";

to (also import System.IO):

// Change .mdx to .m2 if needed
string file = path;
if (Path.GetExtension(path).Equals(".mdx"))
{
  file = Path.ChangeExtension(path, ".m2");
}

I have no idea if it's +90 or -90. In my code base it's -90. Can someone confirm +90 is correct?

[Ugge]

risyer: nice find. that might be whats causing hawkers problems too. (maybe even a combination)

The rotation being wrong is indeed an old bug. Afaik its been there since pontus originally made pather. Im gonna try to confirm it tomorrow.

[risyer]

The rotation being wrong is indeed an old bug. Afaik its been there since pontus originally made pather. Im gonna try to confirm it tomorrow.

Hmm. But then the Wow.Dev Wiki would be false. See ADT - WoW.Dev Wiki. The WMOs would be rotated wrong also. Plus the svn - Revision 15: /MPQNav/MPQNav/MPQ implementation would be faulty. I can't imagine they all use false rotation. So I guess -90 is correct. But maybe I find some time to visually confirm it.

[Hawker]

With +90 I saw no improvement.

With risyer's change, the bot actually came close to making it from Mordo to Shadow Priest Sarvis but sometimes gets caught in the gate to Deathknell and it cannot get past the steps into the church.

I think its an interesting tool and useful for navigating to and from mobs on patrol. But to use in cities, you need to specify waypoints it must use. For example, doorways need 3 spots to ensure the bot will go in front of building and walk straight through. I'd see this as something that a bot user should do - we need only make sure the bot has a button to record waypoints.

I've done a quick code comparison with Barthens version and the "official" versions from the ppather site. His are marked with copyright of Pontus Borg. The official ones are marked with GPL. If you delete the copyright notices, the line counts are the same.

In the absence of an open source navmesh, this is a nice tool. If there are more bugs that need fixing, please let me know and I'll fix it and host the project on sourceforge.net