Desync error since update 7.009

[Zebreu]

Hey,

I've got a desync with a friend since we updated the game. It happened twice after a variable length of time (around an hour into the game). We haven't seen any kind of event that would trigger it either.
The first save was when the error happened and the other one is the save we started from.

Thanks for looking into it!

[keith.lamothe]

Thanks for letting us know, sorry to hear about the desyncs.

So they're not really predictable at all?  How long between each one?  Are we talking minutes, hours, days?

Thanks

[Zebreu]

Well, the first time we played, it happened after a bit more than an hour. Then, we started again, and it took a little less than an hour.
Also, just thought of mentioning that, we had that bug where we got free hacking unlocks, because we wanted to see what the response was below 0 hacking progress, than loaded a previous save, and we still had the ships.

[TechSY730]

Well, the first time we played, it happened after a bit more than an hour. Then, we started again, and it took a little less than an hour.
Also, just thought of mentioning that, we had that bug where we got free hacking unlocks, because we wanted to see what the response was below 0 hacking progress, than loaded a previous save, and we still had the ships.

Hmm, potential save game state "leak"?

Isn't that usually a sure fire sign of something that is very likely to cause a desync?

[Zebreu]

Well, we wanted to continue the game, and we played many many hours after we got that bug (and after the bug was resolved, I read the wiki).

[keith.lamothe]

Ah, a failure of clearing the save state definitely can cause desyncs, though I'm not sure that's what's happening in 7.009.  To clarify: were all these desyncs in 7.009?

[Zebreu]

Yes.

[keith.lamothe]

Just looked at the saves to confirm that the state "leaking" across save/load no longer happens, and that does seem resolved.

Trying to isolate the cause of the desync, which is notoriously difficult... how much did you play these games in 7.007 (7.008 only being the "current" version for a pretty brief time)?  In other words, how confident are you that the desync issue was introduced in 7.008 or 7.009?

[keith.lamothe]

Btw, do you have any error log files in the RuntimeData directory in your AIWar directory?  On either player's machine?  Specifically things like:

ArcenDebugLog.txt
UnhandledErrors.txt
ErrorsReportedByEngine.txt
(anything with the word "desync' in it)

Thanks

[Zebreu]

Just looked at the saves, and we played 7 hours at 7.001, then 3 hours at 7.005, and 3 hours at 7.007. And then that last hour or so at 7.009.

[keith.lamothe]

Hmm, ok; thanks.  Does seem like it happened after 7.007, or at least after 7.005.  Looking through the code diffs; not a whole lot changed but I did do some architectural rearrangement of hacking logic to make adding new types faster.

[Zebreu]

And here's the Desync log (and others). You'll notice we also were the one with that odd desync with carriers from some time ago, haha.

[keith.lamothe]

Another thing: if you can make sure that each time you play together you've both just started the program (and haven't started or loaded any other game since that run of AIWar.exe) that will eliminate desyncs caused by state "leaking" across save/load.  If you continue getting desyncs while observing that precaution let me know as that narrows down the sorts of things that could be going on (at least for one cause of the problem).

[Zebreu]

I am fairly sure that for those two desyncs we just got, we indeed started the program and loaded one save only, we generally just do that.

(Edited: But yeah, except for that one time where there was this leaking)

[keith.lamothe]

I am fairly sure that for those two desyncs we just got, we indeed started the program and loaded one save only, we generally just do that.

(Edited: But yeah, except for that one time where there was this leaking)

Ok.

Is the problem ongoing as you continue trying to play that game?  I'll keep hunting but it is helpful to know the frequency of occurrence to actually expect (for attempts to reproduce the problem).