Very serious combat bug?

[vfb]

I think the debug log is not being read correctly.

This line:

hitunit 221 16083 dmg-34 spec96 ba2

is negative 34 points of damage. That makes sense for an average head hit:

Flail 3 + Str 11 = 14 (+DRN)
Prot 8+4 + Shield 40 = 52 (+DRN)

dmg ~= -38 (or -49 with no Str added in yet)

The log normally has dmg8, etc, except for a big hit from the King himself, then you see for example dmg22. Like this:

hitloc King of Flames strikes Flagellant wl2 diff-5 -> 3
hitunit 16083 221 dmg22 spec65 ba3
damage 53 on Flagellant, spec0x41 ba3

(I don't understand why it's dmg22 then "damage 53". Strength is 31 though, but that should already be in dmg, I think.)

Then, in both instances, 2 hits from 2 different flagellants get a "dmg" total of (-71), doing:

damage 36 on King of Flames, spec0x60 ba2 (strength 7 flagellant)


damage 32 on King of Flames, spec0x60 ba2 (strength 11 flagellant)


Note that also in the log, a piker hits the King for 1 hit of negative damage, and it has no effect:
16087 striking with weapon Pike. att15 def23
shieldprot for King of Flames = 40
hitloc Pikeneer strikes King of Flames wl6 diff-1 -> 4
hitunit 16087 16083 dmg-35 spec2097155 ba4
... (no damage taken)

So, when there is large negative "dmg", there is a bug converting it to "damage", and it becomes large and positive. I think.

[Forrest]

In a Axis & allies game one night I could not miss with 1 in 6 chance antiaircraft rolls. I lost because it was the only die roll I could make that night.

In a D&D game a friend called his 20's on my dice for 2 hours. His stupid little fighter was a god in battle that night. We spent the night joking about taking him to Las Vegas.

There are times when random just does not work.

[Beorne]

Ballbarian said:
How does "unlikely" suddenly equate to "impossible"?
Get over the loss of your king and find some Markata (monkeys) to take out that overpowered flagellant.

I will wonder you, but the Manchaka King of flame owner is my only remaining enemy (I'm Ryle'h, we are at war) and I will take full advantage of his SC loss. I don't like to take big advantage from bugs.

PS: where is the cake icon to plead for bug corrections?

[vfb]

Do you still think it's huge positive damage from a DRN roll?

Each flagellant did negative 71 "dmg". You just need average DRN rolls to get this. It's not a bug in the RNG, it looks like a calculation bug.

The negative 71 "dmg" turns into:

Strength 7 Flagellant: 36 damage
Strength 11 Flagellant: 32 damage

[Beorne]

I'm sorry for my tone that could have appeared a little harsh. I can hardly bear the dev assertion "We don't like to correct bugs we prefer to introduce more color" but I can't bear the player answer "Well we are happy with this".
I'm a programmer too and I don't like correcting bugs too. But I have to.

[Beorne]

johan osterman said:

It is not alltogether clear that what you have reported is a bug. And the AI ignoring spell order is a a unhappy compromise, when in dom1, I don't recall if it was the beta or the finished game, the AI never ignored scripted orders and there were complaints about that too. The AI ignoring scripting is a feature that was added because It was a common strategy to deplete casters gem resources by making small needling attacks before big battles, for example you could cast a cheap summon before a big battle and htne your opponent would be unable to cast his big battle spells when the big battle hit. The solution was to make the AI calculate when a spell was not desirable to cast. It is sometimes annoying but the way it was earlier worked worse.

I'm not saying this mechanism is wrong, nor I are saying that you must improve AI, I know it is almost impossible cause the large numers of variables.
But is at least three years people ask to implement a simple spell blacklist.

Where is the cake?

[Edi]

I'd have been inclined to dismiss this as a fluke incident, since such things do tend to happen after all when the game is played long enough, but there are couple of reasons I don't, especially after the debug log analysis by people who understand it better than I do..

I'd be interested in what system you are playing on, Beorne. Is it Windows? Because if it is, there may be two different issues here.

The first one would be whether or not there is a calculation error in the Dom3 routines when dealing with large numbers (negative or positive). The second, which may or may not be tied to the first, is a problem with Windows itself. Whether the second problem is relevant depends on if Dominions3 uses its own random number generator or if it relies on the default random number generation method of the operating system it is installed on.

This article on how the Windows RNG is flawed is the one where I got that idea. If Dom3 relies on the default OS RNG, then on Windows system this non-random behavior on the RNG would translate into non-random behavior in Dominions and could cause problems. Especially if there is also some sort of calculation bug in Dom3 to boot. The Windows RNG problem could also explain some of the battle review inconsistencies between systems that have different operating systems (e.g. different outcome according to battle review in Linux and Windows).

Of course, all of the above goes with a veritable mountain of salt, since my knowledge of both programming and cryptography is at a very basic and rather non-practical level. But it IS a possible explanation, even if a farfetched one.

In any case, it requires some real looking into, which would be JK's province.

[Endoperez]

vfb said:
The log normally has dmg8, etc, except for a big hit from the King himself, then you see for example dmg22. Like this:

hitloc King of Flames strikes Flagellant wl2 diff-5 -> 3
hitunit 16083 221 dmg22 spec65 ba3
damage 53 on Flagellant, spec0x41 ba3

(I don't understand why it's dmg22 then "damage 53". Strength is 31 though, but that should already be in dmg, I think.)

I think strength is indeed calculated afterwards. I think damage multiplications, e.g. 3x against undead, would appear between the two phases.

Note that also in the log, a piker hits the King for 1 hit of negative damage, and it has no effect:
16087 striking with weapon Pike. att15 def23
shieldprot for King of Flames = 40
hitloc Pikeneer strikes King of Flames wl6 diff-1 -> 4
hitunit 16087 16083 dmg-35 spec2097155 ba4
... (no damage taken)

I think this means that the piker just failed to pass the (RNG+str - 35 (+RNG?) ) roll. That is, he didn't get extremely lucky, so he didn't deal any damage.

Most often, when a number rolls over in negatives, it also works like that in positives. Soul Slay deals base 1000 damage and has no problem killing 10 hp humans, leaving 9990 points of overflow that DOESN'T translate into a negative. So overflow bug isn't probable.

[vfb]

Oh, I see what you are saying ... the "dmg-35" could be prior to the DRN RNG.

But then I can't get the numbers (Wpn-Prot) to add up, no matter what body part I choose, and no matter whether the shield counted or not.

Guess I'd need to look at a bunch more battles to understand what "dmg", "att", and "def" really represent (At least I know "damage" is the actual damage dealt, post DRN, post everything).

[Agrajag]

Edi, I find it extremely unlikely that Dom3 uses any Windows related RNG. The most obvious reason is that most of the time replays work exactly the same on different OSs.
IIRC (this is from a topic that was discussed a very very long time ago) the difference between OSs was due to an oversight, where there was a phrase similar to "if (random() > random())" and on some OSs the left random() was called first and on others the right random() was called first, which obviously completely alters how the battle plays out.
This may be the time to point out that the battles have some random seed associated with them and are played by using that random seed to simulate the battle. (The simulation remains identical through replays and OS changes because the random values are fixed since they rely on the random seed)

As an aside, depending on how Dominions generates random numbers, something extremely unlikely may be a bug (though considering Johan's response, I really doubt it). If Dominions uses a (huge) "table" of "pre-randomized" numbers (basically, a gigantic "sequence" of numbers that have been generated through some awesome RNG [like random.org's]) and asking for a random number merely fetches a number from the table according to some sequence (for example, the random seed for a battle may be a place in the table, and asking for a random number fetches the number in that place, asking for another random number fetches the number in the following place etc.), it is possible that the table does not contain a sequence of numbers that will allow something this improbable, in which case a bug may be involved.

And of course there's also vfb's debug analysis which might also point to a bug.
EDIT: ^- or maybe not, that remains to be seen. (Just see when mine and his posts were made )