v1.00.84 Crash to Desktop

[TR Shrum]

I reinstalled per the above instructions to a new Folder on my E Drive.  No switches, default settings and got the CTD at the same location as the installed game on C Drive.  I tried reloading the my saved game using HtH on both the C and E install, but they too CTD at the same location.
 
I tried a Guadalcanal Scenario as the Allies and so far I've had no CTD.
 
My system info is on a previous post on this thread.

[awadley]

I removed all switches and started a new senario one as allies vs. computer, no problems, played six turns.  Removed Save folder and restored from backup the original Save files from before the 1.00.84 upgrade, started with the save and updated the old save file, but (per jwilkerson) did not update the data base and got the same CTD.  It must be the beta update that has something to due with the 1.00.84 update that the game engine does not like.

[jwilkerson]

If anyone has a save that we can load that will then crash repeatably we are very interested in such a save. That's what we need to fix it. We do not have such a save at this point. We will keep trying to produce one also.

The process would be.

01 - Download 1079 (initial release) and install into new folder.

02 - Down 1084 (patch 1) and install into same folder.

03 - Leave all settings untouched.

04 - Use no switch settings.

05 - Start game against AI - but violate item 03 on one point - switch to non-historical start (so you can save before running the turn).

06 - Save

07 - Run turn.

08 - Repeat above process until we get a CTD.

09 - Upload save to this thread.

[Panjack]

It took three tries before I got a crash following the most recent "protocol."

Used the same game I had previously (newly) installed above. And same laptop running Vista Business. No switch settings. Scn 1 against AI, but non-historical start. Saved game before running turn 1.

Additionally, as it took a couple of tries to get the crash, I also turned off the in-game showing of combat and reports (to save hitting esc a hundred times). I hope that doesn't matter.

The first two times--when it didn't crash--I got the report "No.700 Sqn FAA S-5 planes diverting to Singapore" embedded in the CombatEvents. This was the point at which it crashed before but in these first two runs it continued on after this point with no obvious problems.

But the third time I got a crash. And, yet again, the last line of the CombatEvent report was "No.700 Sqn FAA S-5 planes diverting to Singapore." So it appears the crash occurred at the same place in the game. But this time it appears that the aoperationsreport was saved after the CombatEvent report. (Perhaps this was due to the combat reports being not displayed in game). This is the contents of the last couple of lines of this txt report:

Japanese Aircraft sighted over 2nd/45th PS Inf Battalion
Japanese Torpedo Bomber sighted over Laoag
Japanese Recon sighted over 2nd Burma Brigade

THEN the crash occurred.

A copy of the game save made before turn 1 was run (for the game that crashed) is attached in a zip.

[jwilkerson]

Very cool - I hope this helps us - it should. Were you in "full screen" mode or "windowed" mode?

[Panjack]

full screen mode.

[Crimguy]

I had the same CTD again, after uninstalling and reinstalling fresh. Didn't touch much except turning off combat animations and reports, to speed things up. Didn't get enough time to see what caused it - will the data have been written to the operationsreport?

Historical 1st Turn
Dec 7th Surprise
Player Defined upgrades ON

ran the turn in windowed mode.

Forgot to add in the other post - I'm running Vista Home Premium, most recent patches applied.

[jwilkerson]

They key now is to try to get a crash in windowed mode. One of three programmers working on this can get a crash in full screen mode but none can get it to crash in windowed mode. Unfortunately we can only debug in windowed mode. Since the issue is not related to something simple like missing ship graphics we will have to debug to find the issue - so we have to get it to crash in windowed mode.

[jwilkerson]

Crimguy ... if you can generate a save that crashes in windowed mode we are good.


Follow steps 05 through 09 which I list out above.

[awadley]

Joe,
     Could the CTD have anything to do with the transfer of No. 700 Sqn FAA S-1 and No. 700 Sqn FAA S-5 when the Repluse and Prince of Wales are attacked and the scout planes fly to Singapore.  As I saw the Repluse make Singapore and the Pw sink.  The Sqn on Pw was lost per the Destroyed air group listing, but a single plane element is ashore at Singapore, while the other plane is still on the ship.  The listing shows 2 destroyed.  Just thought that this might cause a mismatch with air groups.
 
     Have not been able to get the CTD following your above steps after 6 turns.

[jwilkerson]

Well it could have to do with that - but between the transfer and the display of the report there are oh say 10,000 lines of code. Which line should we change???[:D] Also there are asynchronous processes running in parallel created by 1000s of lines of other code.

Which line of code should we change? What should we change it to? [:)]

Without a debuggable save this is a tiny needle in a large haystack ... [:(]

[Crimguy]

OK - How strict must I be in not touching anything? I'd like to speed up messages a bit etc . . .

ORIGINAL: jwilkerson

Crimguy ... if you can generate a save that crashes in windowed mode we are good.


Follow steps 05 through 09 which I list out above.

[jwilkerson]

Go ahead and speed up messages.

[Panjack]

I reran my saved turn that previously led to a crash. It didn't crash the first two times I re-ran it. It crashed only the third time.

I'm only running the first turn as the crash might be something that happens only in the first turn.

So it might be that any given save that led to a crash previously only will lead to a crash in a later run if some particular sequence of dice rolls happens (again) during the turn.

[jwilkerson]

Yeah but the whole point of starting from a save - instead of just starting from the begining of the game - is that the primary seed for the game is determined at the game start. This primary seed is saved in the save file. Whereas if I just start a new game over and over - I get a new primary seed each time. So saving the game should eliminate variations in the die rolls - they should be IDENTICAL. Of course this is "should be" - if there is a bug - if one of the programmers did not use the correct random number generator (there are several for different purposes) then this could be the issue. But in theory, all combat related die rolls should be the same in the same sequence for the entire game regardless of machine - as long as the code being run is identical.

We have been able to reproduce the crash in full screen mode on our side - I can't but one of the three programmers can - the other two can't. But so far none of us can reproduce in windowed mode which is what we need for debugging. We haven't given up. [:)]

[jwilkerson]

Eck ... now it seems we just got to where we can reproduce in full screen and windowed mode - but still not in the debugger ... still crankin ...

But I think you guys have done what you can do. Since we can now reproduce in both full and windowed mode I think we can go internal now.

Thanks for those who have helped out!!! Hopefully we will have this resolved soon! I hate crash bugs [:@] - especially if I can't reproduce them!!! [:D]

[Crimguy]

That's good because I just got a CTD on the first try after you gave me instructions. I'm a CTD madman on this build.

[jwilkerson]

James says he has found it ... [&o][&o][&o] ...

We will try to wrap this an the "nada-update" for the ships up into a hotFix over the next day or so.

[Panjack]

[witpqs]

ORIGINAL: jwilkerson

James says he has found it ... [&o][&o][&o] ...

We will try to wrap this an the "nada-update" for the ships up into a hotFix over the next day or so.

JWE has one to release with some USN ship arriving at a port up a river it can't navigate in AE - maybe that can make it in?