Yes, but the symptoms in this case are not things the AI would ever do (disbanding Southwestern Front is nowhere in the code, moving tons of support units back from combat units is not something the AI would normally do, and moving HQs 10 hexes to the rear would only happen if the AI was trying to do a major retreat, in which case why didn't the units move as well?). I left Gary a message about the prior bug fix in case it rings any bells, but it's a long shot. Thanks for mentioning it though, as you never know.
Then why did all of the above happen over atleast 4 turns?
I dont understand why with at least 3 turns of data we are all still in the dark?
Your getting very direct support from the producer of the game, a alpha/beta tester and one of the programmers..
The bug(?) is nasty, but they'll squash it eventually.
Chill out, they are doing the best they can
AKA Cannonfodder
"It happened, therefore it can happen again: this is the core of what we have to say. It can happen, and it can happen everywhere.”
¯ Primo Levi, writer, holocaust survivor
The reason we are in the dark is because we have not been able to repeat the bug from any save. If it's not repeatable, then it can be near impossible to figure out. We are taking more saves at more points during the turn so in case it happens again we will know when it is happening. Hopefully it will then be recreatable, but if not, at least we know when it's happening. We don't know that yet from what we've seen. A server game is always harder to figure out because we don't have every save going back through time, unlike a PBEM game.
All understanding comes after the fact.
-- Soren Kierkegaard
If the game is somehow corrupted, is it even possible to fix that? I'm not sure this is a coding issue at all since nobody seems able to duplicate this behavior. Gremlins took a liking to this game in particular.
If the game is somehow corrupted, is it even possible to fix that? I'm not sure this is a coding issue at all since nobody seems able to duplicate this behavior. Gremlins took a liking to this game in particular.
Thats what I thought... The problems reported seem only related to THIS particular game. Thats why I said bug(?).
AKA Cannonfodder
"It happened, therefore it can happen again: this is the core of what we have to say. It can happen, and it can happen everywhere.”
¯ Primo Levi, writer, holocaust survivor
Well, having Pelton go back and replay his turn (with us downloading each save as it is made), has allowed the game to get to Michael T's turn again without the support units leaving the combat units (which they had done the first time we reached this point). So this bug did not repeat. We will continue on and see if Michael sees any other probelms that have come up.
All understanding comes after the fact.
-- Soren Kierkegaard
Is it only the Front HQ's that are disappearing? or are army HQ's vanishing as well. Also has Pelton been checking to see if anything similar is happening to him on his turn?
Manstein63
'There is not, nor aught there be, nothing so exalted on the face of god's great earth, as that prince of foods. THE MUFFIN!!!'
In the Michael T vs Pelton AAR game there were several bugs that came up during the past 5-6 turns. They were:
1) Some Soviet HQs unexplainably moved 5-10 hexes to the rear. This was the first oddity that happened in the game, and it also happened again on the last turn played when we were closely following each save made, but it could not be repeated.
2) Soviet sapper units attached to combat units en masse moved to be attached to HQs. This happened at least twice, once either with the first HQ move, or a turn later, and once a couple of turns after this.
3) The Soviet Southwestern Front and several air command HQs disbanded. This happened once, sometime after the first Soviet HQ move, IIRC.
4) Many Soviet air groups moved back to the national reserve. IIRC this happened twice, once while we were watching the saves, but it could not be repeated.
All four of these strange bugs could not be explained by any normal code, whether for the human or AI players. Because this was a server game, it was very difficult to collect saves at every step. For the final two turns Michael and Pelton were contacting me at each step so I could download the latest saves. Unfortunately on the two occasions where we thought we might be able to recreate the bug from saves (items 1 and 4), we didn’t have the right saves to pin it down to either a logistics phase or during player movement and the bugs did not recreate when we just ended the turn.
In the last case which was our attempt to repeat item number 1, we had the save at the end of the Soviet turn but did not have a save at the start of the German turn because Pelton had loaded in the game and made a couple of moves before saving. Had we had a save right at the beginning of the turn we could have said for sure if it was happening without German player input. Unfortunately, the process of downloading, saving, and contacting us before going on is very tedious. It’s a lot to expect that it will go off flawlessly. The interesting item was that in the few moves Pelton made during his last turn, he had attacked a partisan unit, and the combat report listed the 1st Rifle Division as the attacker. This unit existed on the Soviet side. Normally due to a display glitch you don’t get an attacker listed when attacking partisan units. The fact that it listed the 1st Rifle Division as the attacker (when by the moves it was clearly an attack by a security division) is indicative of some other kind of corruption. Compounding the difficulty of repeating this particular bug, Pelton had build some Fortified units on top of and next to where the partisan unit had been before it was attacked. He couldn’t say exactly what he did during this move, and I tried many combinations but could not get the problem to reappear (the problem of the Soviet HQ’s moving to the rear).
So in two cases we had pre-saves but were not able to recreate any of the bad bugs. Given the kinds of problems, without a way to recreate the bug, there’s no way to fix it. Since these bugs have not been reported in any other game, it’s very possible this is due to some kind of data corruption that has happened, which we cannot explain. In theory it could be game or server related, although Matrix has said that it’s unlikely to be caused by the server. The problems did start happening during the few weeks when Matrix was having server issues in January, but it’s hard to see how this could have led to the bugs we saw in the game.
If you see one of these bugs in one of your games, please stop immediately and notify us. If your game is not on the server, if you have saves going back several turns then we can look at these saves to try to find a cause. If it is on the server we can only go back one player-turn so best not to save your game, just contact us and stop your move.
I don’t really expect these things to show up again, but there’s no way to know. I appreciate Michael T and Pelton’s help at trying to get to the bottom of these problems, and I’m sorry we weren’t able to resolve the issue and save their game. I had been looking forward to seeing how it came out.
All understanding comes after the fact.
-- Soren Kierkegaard
Perhaps problems could arrive if players are using different versions/updates? Would it be possible to program the game so that files executed under different versions, even partially so, will not process?
Anything's possible, but Gary couldn't find any code that would do some of these things for the AI player (moving HQs back but not units, disbanding certain HQs). If the AI doesn't do it for itself, it's not likely to do them for a human player. The difficulty is that usually we find and fix these kinds of bugs when we have a pre-save that allows us to see the behavior in action. We were never able to duplicate any of these bugs from a pre-save (and getting the pre-saves was a royal pain since they were on the server and only a fwe are kept around at any time).
All understanding comes after the fact.
-- Soren Kierkegaard
It seems I am suffering from this bug too. [:(]
Didnt check SU units, however my Army HQs were displaced in the same matter as Michaels.
Its server game Peltonx - 821Bobo
I have one "possible" explanation. For past couple of turns, I was short of vehicles(mud and snow turns). It was something like 180k needed and I had only 150k. This turn after HQs were displaced it is 130k needed and 150k on hand. It is possible that game engine force moved HQs closer to rails?
Well, well, this is very interesting. @ Bobo I also have a feeling it is truck related because I saw the same thing. Perhaps in bad weather the AI is trying to do something to alleviate the truck shortage problem. Also in my game (against Pelton) in the last turn I had I saw that I had 10,000 trucks damaged/destroyed in one turn. I have no idea how that could have happened.
If this bug can be found and fixed I will be very happy and my mission to steamroll Pelton can resume.
This is actually a good thing. This means that the bug can be traced and squashed...
AKA Cannonfodder
"It happened, therefore it can happen again: this is the core of what we have to say. It can happen, and it can happen everywhere.”
¯ Primo Levi, writer, holocaust survivor
I'm very sorry to hear about that actually. Given it's a server game, I'm not sure what we can do about this. In order to trap a bug we need to pin down exactly when it happened and have pre-saves that allow us to recreate it. With the Pelton vs MichaelT game, we were never able to pin it down and make it repeat. We tried stepping through a turn with us grabbing every save made, but because Pelton had loaded in the turn and made a few moves before we were told to grab the save, we couldn't say for sure that the bug didn't happen because of some action during the German turn that we could not duplicate. I will ask Matrix if there's anyway to have every save stored off instead of always having the latest save overwrite the last one made by the same player. If we could get that capability then at least we might be able to pin down when this problem is happening.
I see that this game is on turn 76. The problems started in the other game around turn 88, IIRC. The vehicles shouldn't have any impact on this, and as I said before Gary is at a loss because there is no code that is designed to move a HQ backwards. What we really need is a PBEM game or AI game that has similar problems where we can get a string of saves.
Since we had trouble getting all the saves the last time, I'm reluctant to try to do the same thing again in this case. Althought it's not good for Pelton, I'm hoping it's something with his setup since he's been the German player in both cases. I realize that's wishful thinking on my part. As I said, I'll ask Matrix if there's something that can be done to save off all the saves automatically and will let you know what I find out.
All understanding comes after the fact.
-- Soren Kierkegaard
Joel is there no way you can convert a server game to a regular PBEM or Hotseat? Simply to provide a test vehicle.
Also I am happy to try again with this as long as Pelton can stick to the protocols needed to track it down. Maybe Matrix could just keep all saves from these two games affected?
