Routing bug

alexti · #1 October 28th, 2004, 11:22 AM

Quote:

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.

Chazar · #2 October 28th, 2004, 11:34 AM

@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 · #3 October 28th, 2004, 12:06 PM

Quote:

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.

Chazar · #4 October 28th, 2004, 12:15 PM

Quote:

Cainehill said:
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.

You're quite good when it comes to workarounds, aint you?

alexti · #5 October 28th, 2004, 04:11 PM

Quote:

Chazar said:
@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)...

No particular reason... I was just replying to your and to Wyxard's post, then I've discovered that your questions largely overlap. So I've eliminated duplicates from my replies. After that I've trimmed my replies to about equal size to distribute server load. So, you see every step was logical and that was just misfortune-3 that the result didn't make sense

Quote:

Chazar said:
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

It probably has Version number, but considering that the file format has not been changed, Dom2 proceed. I don't mind it, but it'd be useful to have a warning telling that the turn was hosted by the server running Version such and such.

DeathDaemon · #6 October 28th, 2004, 08:43 PM

OK different scenario, another bug...

Playing ctis, have 8 sauromancers, 3 inf leaders, each sauromancer with 5 or so bodyguards, and an army of infantry (30 or so). Attempt to storm a wizards tower. The tower shoots and kills a sauromancer. Marignon's mages all cast fire shield. A few xbows fire but completely miss. Everything runs... my mages didnt do a thing... comon 1 mage killed, no other injuries, but they all run?

EDIT: this is a SP game so no host/client issues possible

archaeolept · #7 October 28th, 2004, 09:32 PM

most likely you had your army of infantry on "guard commander" accidentally - all the commanders will rout if one dies if all you have are commanders or commanders and guards.

DeathDaemon · #8 October 28th, 2004, 11:18 PM

This might be a possibility if I only have mages spamming skeletons. My troops are given an order to guard commander. Commander has orders to cast, not retreat. Why must I have at least 1 guy on attack/fire?

deccan · #9 October 28th, 2004, 11:29 PM

Quote:

DeathDaemon said:
This might be a possibility if I only have mages spamming skeletons. My troops are given an order to guard commander. Commander has orders to cast, not retreat. Why must I have at least 1 guy on attack/fire?

Troops on guard commander are considered as commanders themselves, hence they all rout when one dies. Just another one of Dom2's quirks.