Routing bug

[Chazar]

A single dice roll actually does matter! As does restarting the server after patching!

Upgrading ongoing games seems to be troublesome anyway, as is mixing Versions: Someone here mentioned that it suffices to preserve the dom2.exe between patching, but this didnt seem to work well. Copying the entire directories seemed much better to me...

Although I wonder why it sometimes notices that the client is of the wrong Version (2.08 vs 2.12) and sometimes not (2.12 vs 2.14)?

[Wyxard]

I was still running 2.12 as I didn't recieve an in game message or see one here that it had upgraded to 2.14
(above, when switched to 48 hours, he questions whether we upgraded)
so I guess this could result in the discrepancy, although my question now is, if I had the correct Version, would I have won on the main computer as well? ie did I see a valid film of how it should have basically panned out since my computer had all the data concerning the armies on both sides and somehow with the different Versions, my armies data was sent to the host incorrectly, so it ran the battle with erroneous data on my army, or would the patch have swung the battle completely the other way? and I think former is probably correct, since it is hard to see how I could have won one by so much and lost the other by so much.
so I guess I got screwed by not having the same Version running (since noone said anything)

in the future, games should probably stay the same Version for the whole game, to prevent these problems, since they basically ruin the game for some people and are so erratic

wyxard

[alexti]

Chazar said:
Although I wonder why it sometimes notices that the client is of the wrong Version (2.08 vs 2.12) and sometimes not (2.12 vs 2.14)?

It depends if the algorithm passes through the branch with altered RNG calls or not. For example, if commander had burning pearl, he would have 100% fire resistance and when he was hit by the fire damage spell or weapon, RNG wouldn't be called. After the patch his resistance went down to 50% and RNG had to be called in that case to determine amount of damage. After that all subsequent RNG calls get shifted by one, which will produce different results.

Example:
RNG sequence:
1
6
6
5
1

Actions:
Mage A casts fire flies and it hits enemy commander,
Mage B casts soul slay on enemy's Air Queen with 25MR.

Server:
- commander has 50% fire resistance, the fire damage rolls uses first RNG number, 1 and deals appropriate damage.
- mage B rolls his number for MR check: 6. Roll again: 6. Roll again: 5. Ok, 11+17=28 total. AQ rolls its MR check: 1. 25+1=26. AQ is dead.

Client:
- commander has 100% fire resistance, so no fire damage roll is needed.
- mage B rolls his number for MR check: 1. Ok, 11+1=12 total. AQ rolls its MR check: 17. 25+17=42. No problem for AQ.

[Graeme Dice]

Wyxard said:
but that shouldn't change the real result that much, as I still should have won, although if the code values didn't match, that might create erroneous data match ups and thus result in a bug giving us incorrect results.

Well, you were going up against Pythium. Consider what would happen if a paralyze spell cast against your units succeeded on the host computer, but failed on your computer due to there being an extra call to the random number generator somewhere on your computer.

[Graeme Dice]

Wyxard said:
so I guess this could result in the discrepancy, although my question now is, if I had the correct Version, would I have won on the main computer as well?

No, your replay would have shown you why you lost that many troops.

[alexti]

Wyxard said:
I was still running 2.12 as I didn't recieve an in game message or see one here that it had upgraded to 2.14
(above, when switched to 48 hours, he questions whether we upgraded)
so I guess this could result in the discrepancy, although my question now is, if I had the correct Version, would I have won on the main computer as well? ie did I see a valid film of how it should have basically panned out since my computer had all the data concerning the armies on both sides and somehow with the different Versions, my armies data was sent to the host incorrectly, so it ran the battle with erroneous data on my army, or would the patch have swung the battle completely the other way? and I think former is probably correct, since it is hard to see how I could have won one by so much and lost the other by so much.
so I guess I got screwed by not having the same Version running (since noone said anything)

2.12 and 2.14 have compatible turn files, so the only problem that arise from mixing them is that you won't see battle replays correctly. The results you see are correct according to the rules of the server Version. So if you have burning pearl on your commander, your 2.12 client will show he has 100% fire resistance, but in the battle on the server he will only have 50% resistance according to the server rules. Random number can also fall differently and affect results of the battle.

Btw, if you still have that turn, you can view it with 2.14 client and see what happened in the battle on the server.

Typically, there're few elements which are sensitive to randomness. For example, if you have mage scripted to antimagic, you may lose the battle if enemy flier's reach and kill him before he casts. If the random numbers fall differently and the mage survives, the battle can go completely differently.

[Soapyfrog]

Wyxard said:
I was still running 2.12 as I didn't recieve an in game message or see one here that it had upgraded to 2.14

This is the problem then. The server is almost certainly running 2.14... if you use 2.12 to replay battles on a 2.14 server you will see radiaclly different results than what the server actually processed (which you will see in the battle report).

I took over for Atlantis in that game, and I distinctly remember the server hanging and being restarted perhaps 10 or 12 days ago, which would mean the server is definitely running 2.14.

Upgrading to 2.14 will completely fix your issues.

[Chazar]

@alexti:

I do not understand why you cite me? I think that I am well aware of how things work (at least I agree with your post)...

With the line you have cited from me I wanted to express my misunderstanding why a 2.08 client denies to process a 2.12 .trn file right from the start by giving an error message, while a 2.12 client processes a 2.14 .trn file without an error message (but replay inconsitencies)?

I think it should not be too difficult to do that the .trn carries its Version number along which is checked by the client! It would help players who are unaware that the host was updated like in this case. I also think some simple Version control is a necessity.

[Cainehill]

Wyxard said:
I was still running 2.12 as I didn't recieve an in game message or see one here that it had upgraded to 2.14
(above, when switched to 48 hours, he questions whether we upgraded)

After you were reporting battle replay inconsistencies, I posted that the game was running 2.14 on the Try_Something_New thread. This was on the 18th, enough turns back that you could have, and should have, tried upgrading to 2.14 and see whether or not it made more sense when viewing the battles.

Then again, you might have me killfiled, in which case you wouldn't see my post there (or this one here), and thus didn't see the comment about it being 2.14. After all, since I answered, others didn't.

[Cainehill]

Chazar said:
With the line you have cited from me I wanted to express my misunderstanding why a 2.08 client denies to process a 2.12 .trn file right from the start by giving an error message, while a 2.12 client processes a 2.14 .trn file without an error message (but replay inconsitencies)?

I think it should not be too difficult to do that the .trn carries its Version number along which is checked by the client! It would help players who are unaware that the host was updated like in this case. I also think some simple Version control is a necessity.

Yeh, I agree - right now there is no good in-game way to tell what Version the server is running. IMO, this is information that should be displayed in the Game Settings screen (where it tells the victory conditions, independent strength, etc). Otherwise right now you have to put two clams or two fever fetishes on a unit for a turn and look to see if you got two gems or one as a result, to tell if it's 2.14 or not.