Inconsistent PBEM results (Full Version)

All Forums >> [Current Games From Matrix.] >> [World War II] >> War In The Pacific - Struggle Against Japan 1941 - 1945 >> Tech Support



Message

ZonkerHarris -> Inconsistent PBEM results (12/23/2004 1:52:06 AM)

ADavidB and I have been having some problems with our PBEM game of scenario 15, which we started with version 1.40. After we played throught the first few turns, I had a crash while entering my orders, and had to re-load the turn he sent me. When I did, I got combat results that differ from the results I'd seen (and emailed to ADavidB) the first time I loaded the turn. He has also seen some previous turns' combat replays that did not agree with the situation on the map once he loaded the saved turn. We checked our settings and it appears that we've been using the same ones on both sides, so that shouldn't be causing the problem.

The altered combatreport.txt is attached to this message; I'll attach the original one to the next message in this thread.

(All of the saved turns are available if one of the developers / bug hunters would find them useful.)



ZonkerHarris -> RE: Inconsistent PBEM results (12/23/2004 1:54:07 AM)

And here's the combatreport.txt that I got the first time I ran the turn. (The differences begin at the surface battle between the Nagato TF and the PTs.)



rogueusmc -> RE: Inconsistent PBEM results (12/23/2004 2:13:55 AM)

I'm scanning and can't find any difference?

Belay that...I see it now.

Frag tested reports late last week. A bunch of differnet folks ran the replay and sent him the .txt. He said they were all the same.



ZonkerHarris -> RE: Inconsistent PBEM results (12/23/2004 2:23:00 AM)

It must be a pretty subtle thing, since I was one of the people who sent reports to Frag. But since we only made it to December 12 in this game, and have all the saves available, I'm hoping that this might be enough data to help pin it down.



2ndACR -> RE: Inconsistent PBEM results (12/23/2004 3:35:19 AM)

Me/Ron/Pzb see different things all the time in PBEM. We have the same settings and everything.

Send the combatreport with the turn as the Jap player.



Ron Saueracker -> RE: Inconsistent PBEM results (12/23/2004 3:45:50 AM)

If one player has different art thanthe other player,will something as subtle as this cause the differences? I say subtle because the art changes still have the original bmp #.



Mr.Frag -> RE: Inconsistent PBEM results (12/23/2004 4:08:11 AM)

quote:

If one player has different art thanthe other player,will something as subtle as this cause the differences? I say subtle because the art changes still have the original bmp #.


Don't know ... since every single turn I have gotten back so far matched and I presume some folks have some custom stuff installed, I doubt it comes from that.



Captain Cruft -> RE: Inconsistent PBEM results (12/23/2004 12:55:58 PM)

It's nothing to do with art or other data files. Random number generation and its seeding is the issue.

Every time you re-load a normal save file the seed changes, you can see it by the difference in the data given when hovering the mouse over enemy bases. Now I presume that when you re-load a combat save file the seed is NOT supposed to change (it's stored in the save file somehow?). Is it possible that both cases are handled by the same "load_file()" routine and it sometimes gets confused ???

Or, postulating wildly, since this problem always crops up around "new patch time" perhaps the seed is stored in the WITP.PRF file, which gets over-written by every patch. Therefore, if the Japanese player generates a combat save in version 1.30 and his opponent runs the combat save in 1.40 (with a new "default" seed) we get different results. If this true then the problem could be avoided by both players upgrading in between combat save generations i.e. before the Japanese player runs the turn.

This whole thing could have been avoided by one-time generation of a combat event log rather than having to re-generate the combat results every time by processing the "combat save". Bad design decision IMHO.

Having said all that, in my two PBEMs we so far have seen no inconsistencies ... :)



ADavidB -> RE: Inconsistent PBEM results (12/23/2004 1:18:34 PM)

quote:

ORIGINAL: Captain Cruft

It's nothing to do with art or other data files. Random number generation and its seeding is the issue.

Every time you re-load a normal save file the seed changes, you can see it by the difference in the data given when hovering the mouse over enemy bases. Now I presume that when you re-load a combat save file the seed is NOT supposed to change (it's stored in the save file somehow?). Is it possible that both cases are handled by the same "load_file()" routine and it sometimes gets confused ???

Or, postulating wildly, since this problem always crops up around "new patch time" perhaps the seed is stored in the WITP.PRF file, which gets over-written by every patch. Therefore, if the Japanese player generates a combat save in version 1.30 and his opponent runs the combat save in 1.40 (with a new "default" seed) we get different results. If this true then the problem could be avoided by both players upgrading in between combat save generations i.e. before the Japanese player runs the turn.

This whole thing could have been avoided by one-time generation of a combat event log rather than having to re-generate the combat results every time by processing the "combat save". Bad design decision IMHO.

Having said all that, in my two PBEMs we so far have seen no inconsistencies ... :)


Zonker and I started this gave "fresh" in v1.40 - so we shouldn't have any leftover 1.30 "baggage".

I tested this out on the other PBEM game that I am playing, where I am the Japanese player. In this particular game, we did upgrade to v1.40 from v1.30. We haven't seen any changes to the combat reports. However, this game is scenario 4, which is a lot smaller than scenario 15.

The other difference is that in the scenario 4 game where I re-ran a save and got the same combat results, there wasn't any surface combat. What strikes me about the game between me and Zonker is that the combat report in the re-run game only divereged at the point where the surface combat occured and then every subsequent item in the combat report was different.

Dave Baranyi



Page: [1]

Valid CSS!




Forum Software © ASPPlayground.NET Advanced Edition 2.4.5 ANSI
3.484375