Huge Bug - Devs?

[Jazzepi]

Here's what happened. I stormed Marverni's castle in a PBEM game I'm in. My troops routed his, killed his pretender god, and every last one of his men fled the field. His crushers also disappeared (3 magical troops) which I assume dissolved or something after his mages left/died.

Afterwards, I get a message like it's a normal battle report (not the normal siege victory one which tells you nothing). It tells me the kills/losses on both sides. And it said I lost!! Also, his pretender God is still alive. Just to make sure, I checked him for enchantments or items during the fight, and I didn't see any.

Needless to say, I'm really frustrated by this. I lost 200 troops assaulting the castle. I included several pictures. You can see where I killed the enemy God. You can see a long shot just before that happens to prove I still have commanders on the battlefield and that my armies haven't routed. You can see a picture of the province spy-report after the fight. And you can see a picture of the battle report.

http://i208.photobucket.com/albums/bb173/Jazzepi/goddeath.gif
http://i208.photobucket.com/albums/bb173/Jazzepi/conquestmarv.gif
http://i208.photobucket.com/albums/bb173/Jazzepi/crop.gif
http://i208.photobucket.com/albums/bb173/Jazzepi/homeprovince.gif

I can provide the game save if one of the developers wants to look at this. I have no idea what happened.

Also, the battle didn't take forever. There's no way we hit that super high 30 turn cap. The battle replay ends directly after the last one of his units flees on the right side of the map, not at some arbitrary point during it.

Jazzepi

[llamabeast]

It won't be a bug. You must have a different version of some mod to your host, leading to an incorrect battle replay. We just had a similar screwup (on my part) in the Alpaca PBEM game.

[Edi]

Post the savegame and let us take a look. This could be related to the battle replay inconsistency where the replay can show a completely different outcome than the actual real results due to host and client having different operating systems. In cases like this, a savegame is always required.

[Jazzepi]

Is there a place I can just e-mail it to? I'd rather not throw my saved games around online. It's bad enough posting information about my fights in public.

Jazzepi

[thejeff]

How did it look to Marveni? Did he give orders to break the siege this turn?

I've seen the battle replay inconsistency bugs before, but this looks like you got the battle replay for storming the castle but the actual results from him breaking the siege.
I don't see how that can happen.

[Jazzepi]

I asked Tromper to come post about what he saw. I have no idea.

Jazzepi

[Methel]

Ill try going in with the masterpass and looking at the replays for both nations.

[Methel]

Well, both battles show Marveni loosing, he lost his god and everyone routed.

The actual reports for both nations show it as a normal battle with a win for marverni. His god is still alive.

Cant say I understand how thats possible.

[tromper]

Hmmm, well I suspect I'm seeing the same deal that Jazzepi does, which was the obvious outcome of course. He did take loses, but I doubt they were anywhere near that high (busy today and don't have hours to replay it), I certainly routed and Marsention took a second dump. Have no clue what the problem is....

Just relax, Jazzepi, if we must we can always get everyone to redo the turn. heh

[llamabeast]

Methel, did you look at the turn on the machine that did the hosting?

I would guess the hosting machine has something different (mod, OS or something) to all the machines you're using to view the battle.

The battle views aren't contained in the trn file at all. Your own copy of dominions makes them up as you watch them. Normally the result will be identical to the result the host produced though, because everything about the battle was identical, and the same random numbers are used. If there is anything different about your copies of dominions though, the results can be incredibly different (it's like Sliding Doors, if you've ever seen that - a tiny difference can end up having a huge effect).

[Edi]

Are all the machines patched to the same version of Dom3?

[Jazzepi]

I'm using 3.8 no mods.

Jazzepi

[thejeff]

But the thing is, it's not the same battle. The view is a castle storming, but the report is a normal battle. That can't from the normal battle view replay random number weirdness.

[llamabeast]

Oh. That is really strange.

[Cor]

I know its not helpful, but the turn limit is 75, not 30. You can definatly rule that out.

[Jazzepi]

Cor said:
I know its not helpful, but the turn limit is 75, not 30. You can definatly rule that out.



I appreciate that. I thought it was a lot lower!

Jazzepi

[Edi]

The reason it's a normal battle report is that if the defender wins against the forces stroming the fortress, it is shown as a normal battle report for both sides. If the attacker wins a siege, he gets the "We captured the fortress!" and the defender gets a normal battle report.

Since the outcome was that the defenders won (and we have a wonky replay that visually shows the opposite), it's handled as a case of the defender winning and the battle reports accordingly.

[thejeff]

OK, somehow I'd forgotten that happened.

Then it is just a standard replay weirdness.

[tromper]

So this has happened before, Edi? Any idea what the 'wonky replay' was about? I mean, hey, I don't mind watching another battle next turn as he re-storms the castle, but the only way I can imagine him winning was if he had some funky routing problems. And certainly I'm sure the other players would rather Jazzepi loses as many units as possible as he finishes me off.... *snicker*

[ologm]

Were there any immortal units on the playing field?

[Edi]

As said before, the replay problem can happen when the host and the client machine have different operating systems. The report numbers are right and the replay is gibberish, but as for how and why it happened, I have no idea to the actual cause.

[Methel]

Its the same way when I look at it on the host machine.
No mods, and using ver 3.08 with winxp.

[Endoperez]

Was the turn generated on an older version? That's one known cause of weird battle results.

[Jazzepi]

I guess we should just proceed as normal then, the only problem was the replay?

Do the devs want the turn so that they have more examples of the bug?

This is kind of frustrating, because I have no idea how I lost. =\

Jazzepi

[Saint_Dude]

Edi said:
. . the replay problem can happen when the host and the client machine have different operating systems. The report numbers are right and the replay is gibberish. . .



I run into this a lot since I play on a Mac. From what I see, the battle replays usually make perfect sense, and it is the report numbers that are entirely confusing and gibberish.

In one example I can think of (one that actually went in my favor), my non-cold resistant SC pretender was swarmed by winter wolves. In the replay he was quickly rendered unconscious by the cold and killed. Yet in the stats report he killed a fair number of wolves and then ran away to safety. Considering the rate at which he was rendered unconscious in the replay, I found the replay much more believable (although I was happy to still have my SC pretender stomping around).

If the problem is due to client and host using different operating systems, I wonder why clients using different operating systems (with one using the same operating system as the host) see the same replay?