New Version - November 13th (11/13)

[cloudberr]

Hiya,

I'm attaching a minidump from yesterday evening and another created just now, logs as well (logging enabled as per the bug reports topic).

I'm using the full version (with EUI and CBP, C4D, CST and more luxuries). Just checked if it works with the LUA folder cleared and not using any other mods. Still the same - around turn 50 it gives me CTD.

Gazebo, I could also run the debugger if I knew where to find it (hope it's not too dumb a question).

EDIT: I'm on W7 64-bit, if it helps.

EDIT2: Added another minidump along with a save (happened to save literally a second before it CTD'd).

 

[Gazebo]

CTD with 11/13 release on turn 51 - Epic speed. Crash happened right after barbarian archer moved within range of city and worker, which would support the "safe range" bug. I have only experienced a CTD one other time with previous versions.

I know where the CTD is, but new minidumps would be great, thanks!
G

 

[Edaka]

Sorry that I was unable to post these earlier: minidump and lua log attached. Full CPP package w/ EUI and More Luxuries (or is it included in the CPP? I don't know, still installing manually like a hipster I am). Standard map size and speed, "continents" map script, enabled RAs and disabled tech trading. Otherwise all set by default. I hope that helps. Let me know if any additional information might be of use.

 

[Aristos]

Just tried an autoplay after validating steam files, crash at turn 3.

 

[shaglio]

Had anybody had a CTD while using just CP alone? I've only seen reports so far from people using CBP.

 

[Aristos]

Had anybody had a CTD while using just CP alone? I've only seen reports so far from people using CBP.

That is a good question and I just answered it: did an AIautoplay with only CP enabled, nothing else, this was the 11-15beta CP, crash at turn 16.

 

[Gazebo]

That is a good question and I just answered it: did an AIautoplay with only CP enabled, nothing else, this was the 11-15beta CP, crash at turn 16.

Do you have a net_message_debug log? If so, mind posting it?

G

 

[Aristos]

Do you have a net_message_debug log? If so, mind posting it?

G

A 4-liner, didn't have logging enabled, only wanted to test the poster's question... should I run another one?

EDIT: just ran another one, crash at turn 11, log requested attached.

 

[Gazebo]

A 4-liner, didn't have logging enabled, only wanted to test the poster's question... should I run another one?

If you don't mind, that'd be great.


G

 

[Aristos]

If you don't mind, that'd be great.


G

just did. see above.

 

[Aristos]

do you see anything worthy in there? I just noticed the only one with a ranged unit is Boudicca, and the game crashes at the start of her turn 11.

 

[Gazebo]

do you see anything worthy in there? I just noticed the only one with a ranged unit is Boudicca, and the game crashes at the start of her turn 11.

I'm trying to narrow down suspects here. Obviously a ranged unit issue, and it doesn't seem to have anything (necessarily) to do with barbs or the observer slot. So there's something going on with the core m_icurrentturnunits being accessed by the player, or vice-versa. My assumption is that a bad value is slipping into m_icurrentturnunits such that it is throwing an out-of-bounds error (and thus causing the issue). What I don't understand, however, is why this issue is only occurring in a select number of cases. I've had the AI run 10 full simulated games since Saturday, and none of them crashed. I've run 30 games for 50 turns before stopping them, and none have crashed. That's why I am stumped - theoretically I should have had a crash by now, so something's going on with a certain set of values or defines on a user-by-user basis.

The drudgery continues...
G

 

[Aristos]

I'm trying to narrow down suspects here. Obviously a ranged unit issue, and it doesn't seem to have anything (necessarily) to do with barbs or the observer slot. So there's something going on with the core m_icurrentturnunits being accessed by the player, or vice-versa. My assumption is that a bad value is slipping into m_icurrentturnunits such that it is throwing an out-of-bounds error (and thus causing the issue). What I don't understand, however, is why this issue is only occurring in a select number of cases. I've had the AI run 10 full simulated games since Saturday, and none of them crashed. I've run 30 games for 50 turns before stopping them, and none have crashed. That's why I am stumped - theoretically I should have had a crash by now, so something's going on with a certain set of values or defines on a user-by-user basis.

The drudgery continues...
G

are you sure about that OR test in the IF at the beginning of the offending loop? I know it's basically the opposite of the original vanilla IF condition for testing if we have a qualifying ranged unit, and I look and look at the IF and it looks perfect, but somehow I just distrust it (chaining ORs as opposed to chaining ANDs)...

the drudgery continues...

 

[Deadstarre]

Right on time my friend! Happy Friday with wife out and children in their stuff...!

ohhh what could have been. Happy drudgey monday my friend =)

something's going on with a certain set of values or defines on a user-by-user basis

I have a specific set of game options i always use that were in place for each crash, maybe a lot of other people do the same thing. would it help if crashers put their game info here to possibly find a common theme or would it just bog down the thread with unhelpful stuff?

 

[Aristos]

I have a specific set of game options i always use that were in place for each crash, maybe a lot of other people do the same thing. would it help if crashers put their game info here to possibly find a common theme or would it just bog down the thread with unhelpful stuff?

I don't think he is referring to game settings per se, but other values that are passed to functions and may depend on each machine settings...

 

[Deadstarre]

I don't think he is referring to game settings per se, but other values that are passed to functions and may depend on each machine settings...

Ah i see. best solution then is to mail him my computer! need your addy G =)

 

[Aristos]

I'm trying to narrow down suspects here. Obviously a ranged unit issue, and it doesn't seem to have anything (necessarily) to do with barbs or the observer slot. So there's something going on with the core m_icurrentturnunits being accessed by the player, or vice-versa. My assumption is that a bad value is slipping into m_icurrentturnunits such that it is throwing an out-of-bounds error (and thus causing the issue). What I don't understand, however, is why this issue is only occurring in a select number of cases. I've had the AI run 10 full simulated games since Saturday, and none of them crashed. I've run 30 games for 50 turns before stopping them, and none have crashed. That's why I am stumped - theoretically I should have had a crash by now, so something's going on with a certain set of values or defines on a user-by-user basis.

The drudgery continues...
G

I wonder if the definition of the iterator should be inside the preproc directive block and not outside...? As in:

#if defined(MOD_BALANCE_CORE_MILITARY)
	list<int>::iterator it;

	//for each of our ranged units, see if they are already in or close to a plot that can bombard that can't be attacked.
	for(it = m_CurrentTurnUnits.begin(); it != m_CurrentTurnUnits.end(); it++)

Maybe that would be treated differently by some processors/machines, so that some settings/configs give you the out of bounds exception, while others correctly interpret the directive?

Worth a try. Same for the #else block in this case... just shooting blind here, don't have VS installed, just reading the code...

 

[heinous_hat]

Testing now with just the CP...

I can reproduce the exception quickly and consistently. I've set a breakpoint and am stepping through the loop in question (@line 10372 in CvTacticalAI). Problem is, I'm not familiar enough with the code (meaning not at all ) to know what I should be watching.

I could use some guidance, or maybe a tip on how to log the session to get some more eyeballs on it.

 

[Gazebo]

Testing now with just the CP...

I can reproduce the exception quickly and consistently. I've set a breakpoint and am stepping through the loop in question (@line 10372 in CvTacticalAI). Problem is, I'm not familiar enough with the code (meaning not at all ) to know what I should be watching.

I could use some guidance, or maybe a tip on how to log the session to get some more eyeballs on it.

I've run it over and over and I just can't get it to proc the CTD. What's your specific setup that triggers it?

 

[heinous_hat]

I've run it over and over and I just can't get it to proc the CTD. What's your specific setup that triggers it?

Any game I've tried since 11-13 produces it. For testing, I'm just running the CP (alone) on a standard map at quick pace. This particular session, I'm hitting it in 7 turns.

If you're not seeing it on your end, I sort of doubt the save game is going to be helpful, but I can upload that. I would suspect that will confirm that it's a user end problem.

My trouble with the debug session is that I don't know what to look for. I'm stepping through, watching values for m_CurrentTurnUnits, m_pPlayer, pUnit, the loop iterator etc... but I have no clue as to what would be irregular. I guess I could revert to an older version and observe that to gain some perspective.