wrong battle report

[hunt11]

In the game MadCraddle i had just gotten the new turn and was checking what had happened, in the overview map and the battle report it had said that i had lost the battle, but in the actual battle replay, I had won the battle. I have the turn file if anybody wants it.

[konming]

It happens a lot, seems no one has yet to figure out why.

[Endoperez]

The fights are calculated once while the new turn is hosted, and the results of that fight are used for determining which units died, who controls the province etc. The fights you see are replays of that fight, using the same random seed (i.e. all the random rolls will come up in the same order as in the original fight). If, for any reason, there's a calculation in a wrong place, the replay doesn't turn out as it should. There have been many problems that caused this, and they've been fixed when they've been identified.

The only example I can remember right now is that in DomII, the comparison used to determine who wins a Magic Duel (random roll A + astral level A vs random roll B + astral level B) was handled differently in Linux and Windows distributions (Windows: first A, then B; Linux: first B, then A). When the random roll was high enough to change the result, the battle went differently from that point onwards. It didn't matter as long as the hosting and playing happened on the same operating system, which caused even more confusion.

As you can see, it is difficult to find the exact bug that causes the symptom in your case.

[konming]

The magic duel problem must still exists in Dom3. One recent turn replayed completely different from the battle result. And there was an early magic duel.

[cthulhu]

Battle Summary / Strategic Results:
Friendly Commanders: 1, 0 killed
Friendly Regulars: 5, 4 killed
Enemy Commanders: 3, 0 killed
Enemy Regulars: 117, 59 killed
I lost the battle.

What I see in the replay:
None of my non-PD units died or ran away, none got new afflictions, only 3 even got hit. Almost all of my PD units (31 PD) get killed or ran away. Most of the enemy army died, including his only mage.

Spells used during the battle:
Bless, Holy Avenger, Sermon of Courage, Flying Shards, and Twist Fate. The only spells that actually had any effect upon the outcome were the bless spells.

Army Compositions:
My army consisted of 31 PD for MA Ashdod, backed up by 5 Sheshai Anakites and a S2H2 Adon. My bless is E9,F4,W4,N4.
The enemy army consisted of 16 Red Guard (boosted by Strength Of Spring), 8 Imperial Horseman, 8 Horsemen, 26 woodsmen blowpipes, about 55 other infantry (mix of ministry guardmen, ministry footmen, footmen, woodsmen, militia, and light infantry), 1 Minister of Magic (E1), 1 General, and 1 Eunuch.

I don't think the difference here is a result of one spell having a different random outcome. The only things that could have produced the alternate outcome would be if my bless spells hadn't worked (they were scripted) or if... I can't think of any other possibilities. If their cavalry had managed to isolate my Adon then I might have lost, but the casualty rate among my Shesai Anakites wouldn't have been so high.

[Agrajag]

I don't think the difference here is a result of one spell having a different random outcome.

It's not the single outcome that changes that makes the difference, it's in how the game handles random rolls that causes a snowball type change.
Imagine that all random results are rolled in advance into an enormous queue, and then each time you need a random number, you get the first number in the queue.
If everything goes alright, then replaying the battle with the same queue will give you the same replay every time, because every result it predetermined; You will get the same "random" number for the same roll every time.
But if because of some small change, one less random number is needed than is supposed to be (for example, originally the mage casted a spell that also needs an MR check, but this time he didn't), then every single random outcome that follows is changed, because the random rolls will use a different number from the one they used originally.

[chrispedersen]

I'm just surprised.

Why not write the actual events to each players turn?
Aka...
unit 1 moves to square 2. Does 12 pts damage to unit 934.
unit 2 casts Blinding light.
unit 3 casts foul vapors...

Then let the replay just step through what actually happened?

[cthulhu]

Possibly to minimize bandwidth or disk space needed?
But it certainly be a good idea to do so optionally, at least for debugging purposes.

Anyway, even with the entire battle redone, every roll redone, even enemy tactics chosen differently, I'd expect the outcome listed in the battle report to be fairly unlikely.

[Hadrian_II]

Originally Posted by chrispedersen

I'm just surprised.

Why not write the actual events to each players turn?
Aka...
unit 1 moves to square 2. Does 12 pts damage to unit 934.
unit 2 casts Blinding light.
unit 3 casts foul vapors...

Then let the replay just step through what actually happened?

Because this would make the .trn files huge.

Now its just battle with these units, start random generator with number x.

Your system would add a huge amount of information to the turn files. as every action of every unit every turn had to be saved.

[cthulhu]

Huge .trn files would be fine, at least for debuging purposes, as long as they were under 1 GB.