Bug and error logs on queued actions

[AtlasMapper]

Hello guys,
During a game I did face a bug, while using the in-game action/task queue feature (shift + action): at some point, a lots of errors and messages did appear, and the units did seem to be lost for a while. I did set another action for my units, and they did go to other tasks, but the error messages did keep displaying in an infinite loop. Here is the content (transcribed) of the error messages:

JavaScriptError: simulation/ai/petra/attackPlan.js line 394
Type Error: this.queue is undefined
m.AttackPlan.prototype.update Preparation@simulation/ai/petra/attackPlan.js 394:4
												~attackManager.js:137:8
                                                ~headquarters.js:2057:3
                                                ~petrabot.js:123:3
                                                ~common-api/baseAI.js:81:2

I hope it will help.

[sanderd17]

When reporting a bug like this, it's best to include the replay in your post, so that we can reproduce the issue and see where it went wrong.

Though this bug doesn't seem to be related to your actions, but to the AI (something went wrong in its attack plan).

You can find the commands.txt under the replays (see http://trac.wildfiregames.com/wiki/GameDataPaths where to find them).

[AtlasMapper]

25 minutes ago, sanderd17 said:

it's best to include the replay in your post

It did occur while testing one of the map I am creating. So if you don't have the map .xml file, it is useless to give you the command.txt because you won't be able to reproduce the error. Then, if the error occurs on a custom, user-specific map, I guess it is worthless to check it. It is, again, probably nothing more than an user-specific problem, don't waste your time with it. I was supposing the error might be related to a code mistake on the specified part, and it might affect other players, but I am probably wrong. I was just trying to help. I won't bother you anymore with my "bugs".

Moreover, the GameDataPaths you mention frequently is not up to date, and there is no command.txt file in the directory mentioned in your wiki link.

[sanderd17]

For which OS are they outdated?

And you're right that it will be hard to reproduce (and impossible if we don't have the exact map as you used to play). It will probably be a valid bug (the AI shouldn't error out on an unusual map layout), but as we don't even have the opportunity to test if our fix works, it will be pretty hard to fix.

[AtlasMapper]

@sanderd17: Linux, no replay data in the mentioned folder.

[sanderd17]

Hmm I can still find my replays under ~/.local/share/0ad/replays/0.0.20/...

What flavour of Linux?

[mimo]

I suppose you are on A19. The path was changed on svn (and unfortunately the wiki was changed without refering that important point).

So you have to look in ~/.config/0ad/logs/sim_log

 

Concerning the AI error, I remember having fixed in svn some errors i had in attackPlan, so it may be that your bug will be already fixed when a20 is released.

[AtlasMapper]

@mimo: Yes A19. I found it, thanks! That would be nice if you update the wiki accordingly.

Oh nice. As the error messages contains "ai/petra", attackPlan etc. I was suspecting a bug in the code that might affect other players. So if you fixed it already, it is all good. I'll tell you if I have the same error with A20 on the same map. Many thanks, again.