Sync Issues

[Mike McCreery]

I have been having problems with the combat reports in the game on replay not synching up with what actually happens in combat. Sometimes it is minor sometimes it is major. A good example is that 3 or 4 times in this game I have been told that I won a base only to find that the odds came out differently and the base did not go over to my side and/or troops that were reported destroyed in the combat replay suddenly are not and when I check the text based combat report version in game it is different.

This morning things came to a head. For some reason I screwed something up and the game couldnt find the PW file. I was frustrated so I uninstalled the game and re-installed and patched to version 1.7.11.23x10 Aug 23, 2014.

When I load the scenario it says it was saved under 1.7.11.23V a different version and asks me if I want to continue.

I loaded and ran the scenario and the replay is football fields away from what I saw when I played it last night. In fact I played the last few days and I dont recognized the replays as being even similar.

NJP appologized for having a single destroyer get in the middle of everything on the last replay and when I ran it that destroyer was in 3 running combat battles up to Rabaul and prevented a bombardment task force from hitting the base. This morning the combat replay I ran had the BB task force take the destroyer out on the first combat and the battleship bombardment went through.

My thinking is that a different version might be causing an issue here but when we were playing the same version stuff was coming up as evidenced above on base attacks.

Could someone please help me to understand what is going on here and allow me to figure out what is really going on with the re-plays. At this point I cannot proceed until I figure out what really happened in that 3 day running naval battle.

*frustrated*

To illustrate. This is the first task force combat of the combat replay. You will see that it shows that destroyer sunk. Whether the destroyer sank or not has pivotal implications for the entire naval combat replay.

[Mike McCreery]

When I load the turn file from the next turn you can see that the destroyer was not sunk in the first encounter according to the combat replay.

[Mike McCreery]

Here is the previous day, the one I am most interested in. The first battle shows the sinking of 2 light cruisers which did not happen.

[Mike McCreery]

In game in the real combat report the situation is quite different...

[Bullwinkle58]

When you are both playing the exact same version there can be occasional synch bugs. Goes back to original WITP. There have been various theories proposed for why they occur, but I don't think any confirmation on any. In my game with Lokasenna we've had maybe three turns where I saw something significant enough to ask for the CR text file. In my other game we had a flurry of synch bugs in the first two weeks of turns, then it settled out and has been fine.

However, if you are running a different version than your opponent all bets are off. You will likely have myriad synch bugs. Don't do that.

And when you upgrade to a new version you must do it in an exact sequence. This is a sticky on the forum.

"1) Japanese player resolves latest turn using current version.
2) Japanese player saves game.
3) Japanese player exits game.
4) Japanese player sends combat replay to Allied player.
5) Japanese player upgrades to new version.
6) Japanese player issues orders and advances turn.
7) Japanese player sends turn to Allied player.

8) Allied player plays combat replay under old version.
9) Allied player upgrades to new version.
10) Allied player issues orders and advances turn.
11) Allied player send turn file back to Japanese player.

The key is that orders must be entered using the same version and then the turn resolved with that version i.e.:
-J Orders v1079, A Orders v1079, J Resolve v1079, A Replay v1079;
-Both update to v1084, J Orders v1084, A Orders v1084, etc."

[Mike McCreery]

ORIGINAL: Bullwinkle58

When you are both playing the exact same version there can be occasional synch bugs. Goes back to original WITP. There have been various theories proposed for why they occur, but I don't think any confirmation on any. In my game with Lokasenna we've had maybe three turns where I saw something significant enough to ask for the CR text file. In my other game we had a flurry of synch bugs in the first two weeks of turns, then it settled out and has been fine.

However, if you are running a different version than your opponent all bets are off. You will likely have myriad synch bugs. Don't do that.

And when you upgrade to a new version you must do it in an exact sequence. This is a sticky on the forum.

"1) Japanese player resolves latest turn using current version.
2) Japanese player saves game.
3) Japanese player exits game.
4) Japanese player sends combat replay to Allied player.
5) Japanese player upgrades to new version.
6) Japanese player issues orders and advances turn.
7) Japanese player sends turn to Allied player.

8) Allied player plays combat replay under old version.
9) Allied player upgrades to new version.
10) Allied player issues orders and advances turn.
11) Allied player send turn file back to Japanese player.

The key is that orders must be entered using the same version and then the turn resolved with that version i.e.:
-J Orders v1079, A Orders v1079, J Resolve v1079, A Replay v1079;
-Both update to v1084, J Orders v1084, A Orders v1084, etc."

Good thing it is so simple... LOL!!!

Ok, the problem we have now is that I am in a more advanced version and just sent NJP that turn file.

Can he upgrade and run it at all or have I wasted all morning?

Also, since I am at the current version, how do I change back to an older version so he can advance first?

Am I stuck in the matrix?

[Bullwinkle58]

It doesn't have to be that hard. The sequence in the sticky is how you do it if you want to avoid any synch bugs. But you can manhandle into parallel if you're willing to have a turn of chaos.

The Japan player's replay and CR are always game "truth." So just get into parallel on game versions over one turn (in any player order) and instead of looking at or relying on the replay movie ask him to send the CR text file as an attachment to the turn file. Read that and take it as true and do the Allied inputs. After one turn the versions are in parallel and you can go on as before in the normal way. IOW, synch bug events are only seen by the Allied player and are phantoms. They aren't baked into the game history for downstream purposes.

If the Allied player is "ahead" of the Japan player version-wise then yeah, it's a pain to go backwards in versions for just one turn. But above, when I say "chaos", it might not be that bad. Even bad synch bug turns usually only affect one or at most two combat events in my experience. It's possible you could run a non-parallel turn and have no bugs at all. You never know.

[witpqs]

AFAIK, it also seems that "a" sync bug occurrence is only for 1 turn. Why? Because the sync occurs in the combat replay file. Each combat replay file is a dead end, not read into the game or referenced when the next turn is done.

You can continue to experience sync bugs (plural) turn after turn, but that would be sync bugs happening turn after turn for some reason.

But once you get through the upgrade process and get both of you using the same game version for entering orders and resolving/replaying orders, then the upgrade process cannot haunt you any further.

[Mike McCreery]

Thanks for the input.

I am willing to live through a turn or two of confusion.

However, when you need to know whether naval combat has penetrating hits, etc. it is impossible from just the end of turn combat file.

Over the course of the day(s) I think that I have a pretty good handle on what exactly happened.

[Lokasenna]

For those without programming knowledge, this appears to be a bug with the random number generator "seed". A random number generator, in programming terms, is not a dice roller. It is an enormous list of numbers, in a random order. Basically a pre-generated table of die rolls. When generating a random number, the computer uses a "seed" - which "line" of the table to begin on for all the rolls.

I've noticed that sync bugs happen at a certain point in the turn. If you compare the combat reports for the turns they occur, you'll notice that they always begin with the same results, but at some point they begin to differ. So at some point, the "random numbers" begin to be different on your machine than they were on your opponent's. As mentioned, nobody really knows why this occurs. Myself, I suspect it is something to do with the air superiority calculations that happen in the start of every combat replay (specifically, as it applies to carriers), as all of my sync bugs have been on turns with carrier strike (or defense from strike) action to some degree.

Just know that whatever the Japanese player sees is always accurate, so if they send the combat report .txt file, that is what actually happened. And that the turn file they sent you to do the orders is the same.


However, if you are running different versions... the reason the IJ player needs to run the turn under the old version, before upgrading, is that the orders were presumably entered under that version. Often times there are changes between versions that affect what would happen in that replay. So you could end up with wildly different results if you run it under a different version. From my somewhat limited programming knowledge, this would also seem to apply if you loaded a wpae001.pws (the replay) file under a different version than it was resolved under. Your version would be performing slightly different operations at certain points in the replay which could cause it to be different.

*shrug*

It's a complex piece of the software.