Case of the Dying Mages (and God)
 

Well on my game on Yanging as Helheim, I ran into something that left me stunned. In my recent counteroffensive against Pangaea, my main army suddenly lost 4 commanders (a few svarts and my god). Naturally, I looked over the battle report and I was left even more flabbergasted. I accounted for all 14 of my commanders that battle, and after the slaughter...I counted all 14 there again. All of them perfectly safe and untouched. I'm freaked out because I lost my God and I'm at a loss to explain why. Perhaps someone can help me understand what I'm missing, since I've taken some screenshots to help.

http://i189.photobucket.com/albums/z...alar/part4.jpg
http://i189.photobucket.com/albums/z...alar/part2.jpg
http://i189.photobucket.com/albums/z...alar/part3.jpg
http://i189.photobucket.com/albums/z...alar/part5.jpg
http://i189.photobucket.com/albums/z...alar/part6.jpg

Re: Case of the Dying Mages (and God)
 

sadly, this may be the usual culprit:
The battle replay you got is not the one that actually happened. There is no known rhyme or reason if thats the case - although some have said it occurs more often when the game is run across platforms.

Re: Case of the Dying Mages (and God)
 

If you happen to have access to the turn files from the previous turn (as an example, your game is hosted on the Llamaserver), it wouldn't hurt to have the files sent to Illwinter for further study. Johan (the programmer) hasn't found the problem from the other files sent to him, but, well, what else can you do? :(

The replays aren't saved as video, but are replayed from the same random number in every computer. If anything changes the order (any action that was done on the host isn't done on your computer, due to death, fatigue, attack missing and not getting to deal damage...), the rest of the battle is different.

Re: Case of the Dying Mages (and God)
 

I'm the Pangaea player involved in this and my battle replay didn't show the deaths either...

Re: Case of the Dying Mages (and God)
 

did the battle report list the deaths? ie. friendly commanders down 4?

if that doesn't, then it is a weird new bug.

Re: Case of the Dying Mages (and God)
 

The battle report shows the same numbers as in the first picture.

Re: Case of the Dying Mages (and God)
 

It sounds like the server did things differently than either of your own computers. Was it hosted on a third-party server or one of your computers? Did you two use the same OS to play?

edit:
Could you both post a screenshot from the same exact scene? Perhaps something like pause the battle immediately without moving around, press 'n' 15 times, then take a screenshot. If your replays are identical, then... well, I'm not sure, but I think it might help solve this bug. Or one of these bugs, if there's more than one.

And would you please make a copy of the folder named after the game? That way, the current turn will stay for future viewing while allowing you to continue your game.

Re: Case of the Dying Mages (and God)
 

The current version of this bug seems to have nothing to do with server/client OS or version differences. I've seen it in SP games.

Re: Case of the Dying Mages (and God)
 

I have to second thejeffs comment.
I have also had widely divergent results with gods. Like my version says I take no casualties. And yet the map says I'm dead.

Re: Case of the Dying Mages (and God)
 

I'm also of the opinion that it isn't to do with OS differences, but I can check if it's run on the LlamaServer, since I have Ubuntu (Linux) both on my server and at home.

Also if it's on the LlamaServer and you let me know what game, nation and turn number, I can see if it's repeatable and if so can send the files to Illwinter. Hopefully they'll be able to track this one down eventually, as it causes much confusion and frustration.

Re: Case of the Dying Mages (and God)
 

Hmm, does that bug only kill units or can it give them afflictions, too?

Re: Case of the Dying Mages (and God)
 

the bug only involves what you see when viewing the battle - the actual results are reported in the battle report, but a different seed somehow is used for the rendition of the battle that you view. You are instead viewing another, equally possible, outcome; just not the one which actually occurred.

Re: Case of the Dying Mages (and God)
 

Llama, the Doomsday2 game has been having problems with battle replays not being the same as the battle result quite often. The last time was turn 47 I believe, happened on a big battle. (It's now turn 49). I play Hinnom in that game.

Re: Case of the Dying Mages (and God)
 

I've played many Dom_3 games which uses only a single OS, for both SP and MP(hotseat) games... and I can confirm I've experienced this issue as well.
This issue is extremely rare, but does exist. It's possible for this issue to work in favor of the player as well.

This issue has been around for a very long time... unfortunately unless we can find the exact steps to reproduce the issue it will never be fixed.

If you can find the exact steps for reproducing this issue... it would be the most major bug which needs fixing.
It would be Bold Red with exlamations. :)

Re: Case of the Dying Mages (and God)
 

In a blitz that I've been hosting recently there was a report of a replay problem, too. It involves an Ettin Hero dying in a battle to maggots, where he is wielding an Elixir of Life.

Zeldor on Linux 64bit Kubuntu and me on Linux 32bit Gentoo see the Ettin route before he is using the elixir and escaping safely. Hadrian_II, who is on Linux as well, I believe, is seeing the same as what the server, which is on Linux 32bit (Ubuntu), wrote in the log files: That the Elixir of Life is used the turn before the Ettin routes, and then he dies to maggots after routing. The unit is missing in the new turn, so it must have really died, as indicated by these lines from the logs as well:

Attached are the game files. The battle can be inspected for Pangaea, in Patala (province number 32).

Could people that do see the Ettin use the Elixir of Life before routing write what OS they have?

Re: Case of the Dying Mages (and God)
 

I confirm. In this game almost all of my battles have different outcome than the replay shows. Usually it's something of little meaning like wrong aflictions. Big battles differ very much. Some of them, like the example above (I play Lanka there), differ completely giving the win to the wrong side.
I could find more examples if needed.

Re: Case of the Dying Mages (and God)
 

In a game last year I had a lot of battles that were wrong. In most of my other games I can't remember having any incorrect battle replays. Perhaps once a game is corrupted it continues to happen on a regular basis.

Re: Case of the Dying Mages (and God)
 

Actually, that might be the genesis of it. Perhaps the game states have gotten corrupted even before the battle.

I don't know how the game state is maintained. Is that the fatherland file?

Game state being the list of all territories, units etc.
Interesting question.

If list of all units is distributed to all users, it is subject to attack - and corruption.

If only the list of known objects is distributed to each user (units owned, seen, interacted with) then adds moves and changes may not be happening consistently across playerlists (reconciling actions between the player's list and the overall manifest.

If the complete manifest is given to each player, you could CRC it and see if the manifests were the same for the server and each player.

If the people from illwinter are reading this thread, I'd like to make a suggestion:

Some of the results I've seen from this bug cannot be the result of just a different number seed.

For example. You send 4 SC's and 200 troops into a province, with a single scout. My results say I won without a single casualty.

But the server results were I lost - thats just simply not possible as a result of a battle bug.

This seems more of a problem with the results of the battle being misapplied.

Re: Case of the Dying Mages (and God)
 

lch, running XP, I see the battle as you and zeldor do - he escapes w/ an unused elixir and 30 hp. I would guess there was a little corruption in the fatherland file; I find it unlikely, though possible, that Hadrian saw something different - if he does, then I would look to the 32bit Ubuntu that the server was running.

Re: Case of the Dying Mages (and God)
 

I think each user gets only the units he owns.
Battle replays are saved independently of real units. Meaning real units never show in a battle replay only copies of them.

Did that ever happen?
I had other bugs like a forger getting feebleminded because he was sneaking around an enemy province with an astral mage in it.
As I understand that should only happen to a mind hunter (had those, too).
Maybe there is another bug where the server mixes up unit IDs and applies death/afflictions etc to the wrong units?

Re: Case of the Dying Mages (and God)
 

The game states are the fatherland file, yes. And it's the absolute instance that decides what is, and what isn't. The trn files that are being sent to the players are only snapshots from it.

Isn't. Compare file sizes for the fatherland and the nation's trn files. A player only gets sent as much information as he is required to know.

Also, since there were cheating attempts before, JK is using the data from the fatherland file where possible, instead of blindly accepting user input from the .2h files.

The seeds are the same. The starting seed for each battle is being saved and reused upon replay. JK even does checksums for each turn of battle to check if there is an inconsistency. If there isn't, then in the log files for the replay you'll find the message for each turn. If there is an inconsistency, then you won't see those anymore, but instead, you'll see (in my case) for each turn.

Re: Case of the Dying Mages (and God)
 

Now, I have a good hunch where the battle replay inconsistencies might be coming from. I have logged the replay from the game that I posted on the page before. It's very short, only 7 turns, and not too crowded, which is good. I removed all the graphic display logging information from the log of the battle replay, and diffed the files against each other. The relevant sections from the log files are attached to this post. The diff can be inspected in a highlighted format at http://paste.pocoo.org/show/111416/ (removed lines from the original battle generation are in red, new lines from the battle replay are in green)

The battle rapidly develops very differently, as can be seen from the amounting changes between the log files. What stands out as one of the first real changes is which, as I interpret it, shows the available spells that a mage may cast, and evaluates their usefulness. In the battle replay, there are two more choices. A little later you see the same: The evaluation of spells from then on differs a little, because, it seems, there are more spells available. And this evaluation may be using the RNG, so that it then corrupts the battle replays because it uses more random numbers, thus giving different random numbers for target and hit and damage determination.

What I'm getting at, I think the problem is that during hosting, the game may be using the old spellbooks. And during replay, the game may be using the new spellbooks with the spells that have been researched during the last turn. I can't proof it, but that's what seems the most likely to me.

Another possibility is that the battle background is different. The "OptimizeWorld" lines differ even before anything regarding the spells pops up in the diff. May be worth looking into, too.

Re: Case of the Dying Mages (and God)
 

Yeah that would make sense about the Research levels being different and thus changing the random numbers generated. Seems like a good place to start anyway. Good work lch.

Re: Case of the Dying Mages (and God)
 

P.S.: Even more of a hint that there's some kind of overflow regarding the spell books in the log diff:

The game evaluates the spell named "end", which is no real spell at all, but just a marker in the data to indicate the end of the spells, and even the "spell" after that, which just consist of empty, zeroed out data.

So even if this is not the culprit, there's definitely something fishy here.

Re: Case of the Dying Mages (and God)
 

Great debugging work lch

Re: Case of the Dying Mages (and God)
 

lch:

It was all on PAtala's side, right? Well, it must be, as Pangaean SC had only H1, so not much to evaluate there.

I don't think that even server log of battle result that you showed in first post is good. Even if it used Elixir of Life before routing, there is just about 30-40 damage dealt after that. And he'd have 118HP. So he should be still alive. And yeah, he got 2 deaths in that battle, according to HoF.

Re: Case of the Dying Mages (and God)
 

Zeldor, if the battle was evaluated correctly is secondary, if the Elixir of Life has been correctly applied is of minor interest. The biggest concern is, can the battle vcr restored and viewed like it has happened before. This is of a LOT of interest to people. It's no trouble that they have lost, but they want to be able to see how it happened.

Re: Case of the Dying Mages (and God)
 

And if this seeming battle replay bug is fixed, it will be MUCH easier to report any other bugs that might have happened during the battle, since you can see what's going on.

Re: Case of the Dying Mages (and God)
 

P.P.S.: I hang my head in shame. :( I remembered that we've been using mods, namely CBM, and I rechecked, the mod that I was using for the replay differed from the one that I used on the server, although they've both been named CBM 1.41:

server md5sum: e0393301aa565bc98ad5eff476778a37
client md5sum: 79d3146137c36d8b5ddc4016941690a1

I suppose that Zeldor has been using the same outdated mod like I did. When I watch the battle with the right mod, then I can see correctly that the battle is a little different, with the water elementals gaining more ground early on and a maggot infestation on the Ettin Mandragora which makes it use its Elixir of Life and still kills it in the end, with a huge poison bar before routing. This might have been a wrong alarm, I'm sorry. :(

Still, if there are battle replays that are wrong, then I recommend checking the server logs that have been generated during hosting (every server admin should make sure to generate full logs and keep them!) with those that are being generated when viewing the replay, like I did here. That's the best way to find out what's wrong. Are there any other known battle replay inconsistencies?

Re: Case of the Dying Mages (and God)
 

Shouldn't the spells recently researched be used by the map then, as research is ahead of the battle phase?

Re: Case of the Dying Mages (and God)
 

I tried to reproduce the research problem by just accumulating mages without researching; then grabbing level 4 all in one month, at the same time attacking an indy province. It worked without trouble, there was no replay inconsistency. But, I thought maybe that's because even though I was running as a network game, the client and server are the same system.

Anyway, it sure sounded like you had it figured out.

There's some game in MP (sorry, I'm not in it, and I can't remember which) where they have more replay issues than normal. I wonder if the llamaserver has a slightly different version of whatever mod they are using.

Re: Case of the Dying Mages (and God)
 

People should get the new version of the CBM mod (my modified version of the mod is in the first post of the CBM thread). I wrote about it in the MP forum but I'm not sure if everyone saw it. It's different by, er, two spells... *ahem*

Re: Case of the Dying Mages (and God)
 

Maybe do a second install, and throw in the Cross Path spells mod. In the install you run the server, blank out the mod file, so there are no new spells in the fthrland. Put the real mod into the play install, and then do the same test. Bet you see your mages blow off a bunch of cool spells, that didn't really happen in the server(official) version of the battle.

Re: Case of the Dying Mages (and God)
 

is it called 1.41 on the server, or 1.41b? as when I d/l and enable the modified cbm from the top of that thread, it is 1.41b; but after exiting any llamaserver game, it is 1.41 enabled under preferences, which might indicate a naming mismatch, and hence the viewed discrepancies, as the turn is hosted under "llama special" 1.41, but then viewed on the client under the original 1.41, not the 1.41b.

Re: Case of the Dying Mages (and God)
 

I think that only Preponderance and Madness use 1.41b. It was not released anywhere and we urgently needed it to start a game without that A8 Blindlords. It changes only that afaik. It shouldn't appear in any blitzes etc.

Re: Case of the Dying Mages (and God)
 

right, sorry, I got confused, and then was wondering wth this 1.41b was :p

Re: Case of the Dying Mages (and God)
 

I recently had this problem (twice in a row) in a LAN game. My friend and I both have Macs (Both Leopard, for that matter), no mods whatsoever, all played over LAN. I noticed one battle where my borderline SC died after the replay showed a flawless victory. My friend said he had noticed a similar thing a couple turns earlier (I hadn't been watching all the replays).

Now, I have patch 3.21b, but I am not 100% sure he does, so that may be part of the problem (but wouldn't it say something if the patch numbers didn't match?). I do have a number of mods on my machine, but none of them were in use.

I can give no explanation.

Re: Case of the Dying Mages (and God)
 

Well it looks like it happened again. This time, it was my forces somehow losing to a Carrion Lord thug. My battle report showed about 5 Helhirds dying, yet the Carrion Lord routing, whereas the map says I lost 3 Helhirds and I routed.

Re: Case of the Dying Mages (and God)
 

I saw some odd battle results again, but then I downloaded the newer version of CBM mentioned above and the report matched the battle. I think that was the problem.

Re: Case of the Dying Mages (and God)
 

lch confirmed that it was a CBM issue, no bug in the vanilla game.

Re: Case of the Dying Mages (and God)
 

Hmm, did your units in the battle show anything they shouldn't have had?

Experience, Afflictions, etc?

Re: Case of the Dying Mages (and God)
 

I'd like to call it a llamabeast issue rather than a CBM issue :mean: but yes, we need examples from vanilla game files.

Re: Case of the Dying Mages (and God)
 

.:)

Re: Case of the Dying Mages (and God)
 

This problem has started to plague me in almost every single one of my MP games. Is the solution to just redownload CBM 1.41? (the games started before 1.5)