v 25, Bugs/Crashes reporting thread

[JosEPh_II]

OK, the turn is now fine for me...but only with deactived viewport!

Edit:

Wow, the turn times are really really long without the viewport! Thank god for the viewports...

I have yet to use the viewports and turns are taking 3+ minutes to process in the Classical Era on Normal speed Huge Island map with only 6 AI. Barb World is On.

I have never had turns take this long even in the Industrial era. Something is taking way to long to process. Just to get the Running man to show up can take a full 30 seconds. Then he runs for up to on the last couple of turns for over 3 minutes. This is only a recent thing SVN v3333 did not have this problem.

JosEPh

 

[Zain]

Financial trouble decays slowly. You don't just have to have a positive income, your science slider must be over 60% and your "bank balance" must be over a threshold. I can't remember what the thresholds are at the moment. I vaguely remember it being 3000 in the ancient or classical eras.

I've just entered the middle ages. I believe I've had "Financial trouble" all game, despite being the wealthiest nation for about 80% of the time. The science slider has never been moved below 100%, and gold income has practically never dipped below zero except during Anarchy (civic changes, is this what's screwing it?).
Bank balance has never been negative and has been at least over 30k for hundreds of turns.

Thanks for your time

 

[ls612]

@JosEPh: please post a save I can run through the profiler. However, given that my turntimes are actually down since rev 3333 and you are having issues with the cursor, I suspect that the turn will profile fine for me and that you may have a hardware problem.

 

[JosEPh_II]

@ls612,

Here you go. Last update from SVN was 3463.

JosEPh

Edit: System Specs
Dual Core IBM E7400 Wolfdale chipset 2.8Ghz
8GB DDR2 1066Mhz Ram
R6450 MSI HD Radeon vid card with 2GB DDR3, Memory Bus 128 bits
Win XP Pro x64

Checked task manager during EoT and Civ exe was using 1,46x,xxx kb closest other program was 64,000kb.

It's in the game and not in my hardware. Something in the game changed after SVN build 3450 or 3453. And in my case it's the AI turn processing during EoT. Once I'm out of EoT it runs normal.

 

[calvitix]

The turn ended fine for me. Which suggests it also is a memory issue.

Edit: Yeah, the minidump confirms it as a MAF. Something strange is going on though, as that should not be MAFing on a 40x40 viewport, I ran it fine on a 75x50 viewport.

Strange, I've a 4Gb System wit Win7 64bit, Civ only use 900Mo at that point.

It seems that viewport can be the problem in that case : by removing the option, I've been able to continue.

I'll activate the option again, and go one to see if the problem often occurs.


p.s. : How can I check myself with the minidump if it is a MAF ?

 

[ls612]

Strange, I've a 4Gb System wit Win7 64bit, Civ only use 900Mo at that point.

It seems that viewport can be the problem in that case : by removing the option, I've been able to continue.

I'll activate the option again, and go one to see if the problem often occurs.


p.s. : How can I check myself with the minidump if it is a MAF ?

Do you have VC2008 Express (the thing that allows you to compile the DLL)?

 

[ls612]

@ls612,

Here you go. Last update from SVN was 3463.

JosEPh

Edit: System Specs
Dual Core IBM E7400 Wolfdale chipset 2.8Ghz
8GB DDR2 1066Mhz Ram
R6450 MSI HD Radeon vid card with 2GB DDR3, Memory Bus 128 bits
Win XP Pro x64

Checked task manager during EoT and Civ exe was using 1,46x,xxx kb closest other program was 64,000kb.

It's in the game and not in my hardware. Something in the game changed after SVN build 3450 or 3453. And in my case it's the AI turn processing during EoT. Once I'm out of EoT it runs normal.

Nope, it's not just you. I got a horrible end-turn as well (on an i7!), so I'll try and debug it, I suspect it may be related to Koshling's maintenance changes, but the profiler will tell me for sure.

 

[JosEPh_II]

I tried another 2 turns with SVN version 3468 and after 10 mins with "press <ENTER> to end turn" still showing on the screen I had to reboot the computer to get out of the game. No other actions/commands did anything.

In the olden days of COBOL punch card programming that would be classified as an "endless loop" bug. And a ton of wasted printer paper too!

JosEPh

 

[ls612]

I tried another 2 turns with SVN version 3468 and after 10 mins with "press <ENTER> to end turn" still showing on the screen I had to reboot the computer to get out of the game. No other actions/commands did anything.

In the olden days of COBOL punch card programming that would be classified as an "endless loop" bug. And a ton of wasted printer paper too!

JosEPh

Yeah, I noticed that too. It wasn't a full-on WfOC crash for me, but the end-tunr took far longer than it should have. I'm trying to get the profiler to actually give me output, so that I can look and see what is going on.

 

[AIAndy]

Yeah, I noticed that too. It wasn't a full-on WfOC crash for me, but the end-tunr took far longer than it should have. I'm trying to get the profiler to actually give me output, so that I can look and see what is going on.

Another option to find out what it is doing most of the time is just issueing a break from Visual Studio at a random time during the end turn. If you do that some times you will know what it is actually doing. It only works for massive problems of course. For others there is the profiler.

 

[ls612]

Another option to find out what it is doing most of the time is just issueing a break from Visual Studio at a random time during the end turn. If you do that some times you will know what it is actually doing. It only works for massive problems of course. For others there is the profiler.

Unfortunately that didn't help. I tried that three times at different points in the end-turn execution, and it gave me three mostly unrelated breakpoints, one in canTrainInternal, one in getTeamWonders, and one in isBuildingClassMaxedOut. The profiler is refusing to work for me, and I can't debug it properly in VC2008 because BtS refuses to let me leave it and actually use VC2008 with my cursor.

 

[Thunderbrd]

Unfortunately that didn't help. I tried that three times at different points in the end-turn execution, and it gave me three mostly unrelated breakpoints, one in canTrainInternal, one in getTeamWonders, and one in isBuildingClassMaxedOut. The profiler is refusing to work for me, and I can't debug it properly in VC2008 because BtS refuses to let me leave it and actually use VC2008 with my cursor.

I have that problem too. Really makes debugging a pain in the arse!

 

[calvitix]

Do you have VC2008 Express (the thing that allows you to compile the DLL)?

no, I tried to read the dump with dumpchk (from windows SDK debug tools).

I get some informations (8 Streams, with hooked DLLs,...)

 

[ls612]

no, I tried to read the dump with dumpchk (from windows SDK debug tools).

I get some informations (8 Streams, with hooked DLLs,...)

Well, to read the dump properly you need the PDB, which is on the SVN, one directory up from the trunk, in the PDB & DLL archives folder. Put the PDB and the DLL from the save revision as the dump, and the dump itself in the same folder with VC2008 installed and double click the dump. If it is a MAF the exception will either be read/write access violation or a bad alloc.

 

[calvitix]

Well, to read the dump properly you need the PDB, which is on the SVN, one directory up from the trunk, in the PDB & DLL archives folder. Put the PDB and the DLL from the save revision as the dump, and the dump itself in the same folder with VC2008 installed and double click the dump. If it is a MAF the exception will either be read/write access violation or a bad alloc.

ok, thansk, I'll try.

by th way, I have installed VC2008 : there are some hardcoded english text strings in C++ files, I'll try to change the code, so that we can translate them.

 

[AIAndy]

Unfortunately that didn't help. I tried that three times at different points in the end-turn execution, and it gave me three mostly unrelated breakpoints, one in canTrainInternal, one in getTeamWonders, and one in isBuildingClassMaxedOut. The profiler is refusing to work for me, and I can't debug it properly in VC2008 because BtS refuses to let me leave it and actually use VC2008 with my cursor.

Are you running Civ as windowed version (not full screen)?

No common part of the call stack of the different breakpoints?

 

[Thunderbrd]

Are you running Civ as windowed version (not full screen)?

That could well be my problem there. Next time I want to try that I'll try running it in windowed mode. Thanks for the tip!

 

[calvitix]

Some informations about the problem I had with my game (repeatable CTD at turn 808). see save here

if the minidump error is an access violation, it's not a standard MAF :

after debugging, I discover that an AI has completed the Great Wall at that turn :

when calling the AddGreatWall method to build it, I get assertion fails (getviewpotX, getviewportY,...), and after that the access violation.

I manage to pass that turn by disabling the viewport for that turn, and reactivate it after.

But the Great Wall isn't at the right place :

as I use a small viewport (default 40x40), I only have a part of the AI territory in the current viewport

@Koshling : when you'll be back, could you have a look at that bug ?

 

[ls612]

Some informations about the problem I had with my game (repeatable CTD at turn 808). see save here

if the minidump error is an access violation, it's not a standard MAF :

after debugging, I discover that an AI has completed the Great Wall at that turn :

when calling the AddGreatWall method to build it, I get assertion fails (getviewpotX, getviewportY,...), and after that the access violation.

I manage to pass that turn by disabling the viewport for that turn, and reactivate it after.

But the Great Wall isn't at the right place :

as I use a small viewport (default 40x40), I only have a part of the AI territory in the current viewport

@Koshling : when you'll be back, could you have a look at that bug ?

Oh, good catch. I'll PM Koshling about it, even though he's on vacation he may be able to tell me (or you for that matter) what is wrong. Also, how good are you with c++? We could always use more SDK coders, and if you are good with c++ it would be nice if you were willing to help us more with that.

 

[Questdog]

Rev 3470:

My game went from having 0% inflation before update to having 79% inflation and worsening after update. My game is in Ancient Era and I switched to my first inflationary civic (15%) not too many turns ago. Went from having ~+125 gold per turn at 100% research to ~-375. Research dropped to 0% and it's still at ~-175 gold per turn.