[Fixed] Out of Synch

[ultifd]

Fixed #1245

[james876]

What is your operating system to use?

[titi]

yes, something must be wrong :-/

We should really stop adding any more features until the problems are found. I think some more tests with older revisions will be needed, and we will need proper builds to do this. I would suggest to test revision 1221 first. Its the last revision where I was not involved with my multiplier changes  
As far as I remember I already had an out of sync in this revision!

Maybe softcoder can provide proper binaries for this revision so we can verify the problem existance?

[titi]

We tested 1221 with linux32/linux64. No problems so far, but we just played one long game.
So we need some more tests ( including windows ) can someone provide windows binaries for revision 1221?

[titi]

svn revision 1233: There are still problems :/. Our setup was the following:

server: titi linux32
client1: nosogo linux32
client2: ultifd win32

We got an out of sync after a while. Interisting in this case were the following points:
- both clients got the out of sync message in the same moment
- server just did display "disconnected"  ( is this normal )
- nosogo and me used the same platform! ( but different builds! Nosogo used ubuntu 10.04 and me ubuntu 8.04 )

see james876 bug too maybe its related:
https://forum.megaglest.org/index.php?topic=6156.0

[titi]

We really need a lot more tests her for svn revision 1221!!!
Ultifd/tomreyn/notsogod if you have time, please test! ( and others too! )
Is the svn revision 1221 still stable? ( its not the head! )

please report everything ! Bad and good!

[ultifd]

OK tomreyn, nosogo, and I did several tests
We found that

• rev 1221 had no problems...(tested for one game)
• rev 1223 had no problems...(tested for one game)
• then rev 1226 with nosogo, out of synch

So after that, we decided that since before, when we tested rev 1223, we only had two people, it could be different with 3...so we tried the same setup with 1223, but then we lost...
Anyways, it got out of synch on 1226. So it should be something on 1222 to 1226 that is related or affects the synching...
Here are the debug logs:
Mines: http://www.filefront.com/17538694/debug.7z
Nosogo: http://filebin.ca/xvwm/debug.oldlog.7z
tomreyn: http://freeit.org/glest/logs/megaglest_btrunk+r1226_d20101121-014253UTC_ptomreyntrunk_toutofsync.tar.gz

[tomreyn]

I second ultifd's report, and I'll try to visualize our results a little bit different so they may be more easy to understand:

Revision	| Host	| Client(s)	| Test result
----	| -------------------	| ----------------------------------------	| ----------------------------------------------------------------------------------
1221	| tomreyn (Linux 64)	| ultifd (Windows 32)	| no issues
1223	| ultifd (Windows 32)	| tomreyn (Linux 64)	| no issues
1226	| ultifd (Windows 32)	| tomreyn (Linux 64), nosogo (Linux xx(?))	| out of synch on both clients at same time, almost exactly 20 minutes into the game

[ultifd]

Well, we did a test today, so far it seems rev 1223 is OK. I guess tomorrow we'll be testing rev 1224.

[titi]

for 1224 a linux user must be involved! 1224 only had changes for linux.

[james876]

I want to know about the svn revision 1233, I raised the question why has not it solved? Thank you!

[ultifd]

...We are still testing. Please don't ask these type of questions unless it is a stable build...

[james876]

Yes, because svn1233 is an unstable version, so there's Bug should be resolved in the collection of your Bug, and then work together to solve these bug? It takes long to complete?

[ultifd]

We have been working on this problem for a long time...you just don't know...and for now we're just trying to see what caused the problem, by testing revision by revision. Also that the devs and testers have been busy these days.

[titi]

@kobe @james876 @tom123 and friends:
We don't have time at the moment! It will be fixed if we have time and find the bug!

If you cannot wait look at the code and tell me whats wrong! It's all open source!!  

[james876]

ok, I'll wait until you have time, I hope this will not be too long to wait.
thank you, titi!

[ultifd]

Titi is going to try to change something tomorrow...it seems like it is really related to the multiplier...

[titi]

I changed/fixed some things and we found a bug in a cache.
We played one good game ( only 2 players vs 2 cpu players ) and it was fine.

Now we have to tet more to be shure it works ....

[tomreyn]

This seems to have helped. We still observe out of synch issues but only on some larger games, and those have been occurring in the past, too (and are more difficult to fix due to not being easily reproduced).

[ultifd]

Fixed after a lot of testing and discussion... Thanks @testers and developers

Fixed #1245