Bug report: Multiplayer lag crash (1.8.6)

Having trouble with the game? Report issues and get help here. Read this first!

Moderators: Forum Moderators, Developers

Forum rules
Before reporting issues in this section, you must read the following topic:
Post Reply
User avatar
CannonFodder
Posts: 27
Joined: November 29th, 2011, 9:40 pm

Bug report: Multiplayer lag crash (1.8.6)

Post by CannonFodder » November 29th, 2011, 10:04 pm

Hi!

This blasted game crashed on me in the middle of the most intense bedtime action-filled story one could ever imagine!!! GRRR! (Nothing serious - this is an awesome game, thanks for all the fuss making it! ;) )

So the details:

Wesnoth 1.8.6, Linux (Debian Squeeze), compiled from source.
Multiplayer game: Undead (1st player, red) vs. Drake (2nd player, blue), on Cynsaun Battlefield. Timed game, otherwise all vanilla defaults.
I experienced the crash when (due to a lag, see below) the timer ran out while I queued an attack which could not complete. When the timer hit zero, it sailed off to WindowMaker.

On the console I got the following lines:
20111129 22:14:17 error display: could not open image 'units/undead/soulless-drake-defend.png'
Segmentation fault

The particular file really does not exist, but does such really end up in a segfault? (I just tried deleting a random unit's images - it just results in not displaying the particular unit. The actual crash might have happened several minutes later, I can not recall. There were a "few" converted Drakes on the battlefield).

My network access is through a crappy USB stick modem, no wonder for the lags. It most commonly happened that way that I set an attack, then sit back and wait, the turn clock mercilessly ticking my time. I think that was the first time it happened that my attack was interrupted with this lag. I didn't try to reproduce - net lags aren't that predictable, and I would need someone to go in to test with me (I mean some other player across the wire). It might (or not) also matter that it was not me who set up the game.

Thanks for investigating these! (Sorry if it was already reported / worked on - I just didn't find the ways to do this properly now. I read the FAQ, hope my report contains all you need. Tomorrow roughly this time I can go up to help with tracking this down if necessary - I don't think my net access will get any better until then :D ).

AI
Developer
Posts: 2394
Joined: January 31st, 2008, 8:38 pm

Re: Bug report: Multiplayer lag crash (1.8.6)

Post by AI » November 30th, 2011, 1:40 am

This is probably indeed due to your client trying to perform the action without being allowed to. That shouldn't result in a segfault though.

User avatar
CannonFodder
Posts: 27
Joined: November 29th, 2011, 9:40 pm

Re: Bug report: Multiplayer lag crash (1.8.6)

Post by CannonFodder » November 30th, 2011, 4:38 pm

AI wrote:This is probably indeed due to your client trying to perform the action without being allowed to. That shouldn't result in a segfault though.
Could indeed be! If I guess it right, when I play multiplayer, my client responds to my actions, and communicate it with the server afterwards when it gets to access it. Now with an attack someone has to throw the dice, so the client then just has to wait it all up what the server decides. It just can't continue responding me, since either unit could die in the duel what would change the battlefield dramatically. Now what the client sees when the clock struck it? There is clearly an attack queued in, it is waiting for it to be performed, but the server probably never even got the request to roll the chances. So clock is up, the client (I think on it's own without any server response) tosses my turn, but what? An attack was still queued in which did not complete! So effectively the turn got corrupted. Blast. Or maybe the server indeed did respond, but the clock was already up at me, so the attack from my side became an illegal action.

Just speculations, I never looked in the code, maybe it's all wrong. I am here though, my Internet connection with all it's glorious uber-crappiness is rolling all the bytes in and out, so if anyone is up and willing to test can join in for a bit of Drake-smashing or whatever on the menu :)

By the way I am a computer programmer (embedded, but does not matter much, C is all the same), so if needed, I could ungzip the source tree and make changes in here and there to aid debugging (Guess cmake will do it's job just as well as usual makefiles and won't go in a hour of compiling once it is already compiled once. Huh, I guess I will do that hour of compile now to be ready if anyone falls in with such a request!). Anyways probably first it is better to see if it occurs deterministically just by playing. So anyone in for sorting this out?

EDIT: I found the docs on how to properly get around with bug reports (:phew: - read the docs first, eh? :D). I checked in the gna bug tracker, this bug does not seem to be discovered yet (At least none of the multiplayer related issues seems like matching what I found here). I think I am going to register in there to get this up properly. Here it is: https://gna.org/bugs/index.php?19083

Post Reply