Single Player bugs and crashes v36 plus (SVN) - After the 24th of October 2015

[Thunderbrd]

@SO,
Is your game using SM or Hide and Seek?

When T-brd posted that he had a property control issue I had a game in Class Era. I reverted back to 8998 and went back to a save from that build. I was able to play on. I am Not using any Combat Mod Options nor Rev.

After T-brd posted a fix for the Property Controls I updated to 9009 and played all last night on that game. No Problems.

This morning I started a New Game, again No CombatMod stuff. I have played 80+ turns with again No Problem.

Therefore the problem seems to be with your set up and Not with the SVN.

And I got no assert over Bison Rider either. But I have yet to Update to SVN 9012. When I do I'll post back here if Any problem shows up.

Sorry Boss but I think it's on your end.

JosEPh

H&S can exacerbate the problem if I have accurately identified what it is but it can happen in the core as well. Again... I understand the problem and must fix it asap but I'm stuck at work at the moment. It's not a problem with a specific SVN that will show up automatically or not after that update or revision and reverting back to earlier won't fix it either since it's a problem with a 'bad' or infected gamestate.

I'm working on correcting how the gamestate goes bad and how to heal it when it does and shouldn't take me long from here once I can get to working on it.

 

[Thunderbrd]

stachnie reported in the main C2C: SVN discussion:
After upgrade from some mid-December version to something more recent (Jan. 11) I noticed a strange thing. All existing City Guards turned out to be extremely weak: Str. 3.33 -> Str. 0.01, max HP 1 (I play with Size Matters, Seek&Hide and so on). I think I need to disband and rebuild them now because Crime raised dramatically, but it was enough to build 1-2 newely built City Guards (they are very expensive now!) to make it to drop from 200+ increasing about 7/turn to about 140 (and still decreasing).

I think the drop in strength was caused by the svn change where the crime units moved to new techs, and the game not dealing right with units you own but have not researched yet. I had the same issue with a game in the industrial age where the same thing happened when i got the cvs commit with the new law techs.
I kept the sheriffs that got there strength reduced and once upgraded the upgraded units had the strength they should have.

It's more about strength adjustments taking place on existing units in the SM environment not being able to adapt properly. However, as expected, once upgraded, they were corrected because this is a proxy unit recalculation action. Setting up a purposeful direct unit recalculation is a difficult procedure at best so sometimes these things happen but it won't continue to happen once the existing game outgrows the issue, as you've seen.

 

[JosEPh_II]

Yeah i play with alot of stuff, but i dont play with Rev any longer, just dont like it going into a Rev every other turn later in games. .

But as far as ON MY END, cant be, because i dont change anything in the mod at all . .

EDIT: Yeah didnt matter i UNchecked all of the Combat stuff, and i still get a CTD, so playing with the Combat Mod and Hide and seek doesnt matter. . .

I've played with both the older game and the New game with SVN 9012 and no problems have occurred.

You would think if it was a common problem more ppl would be reporting it don't you?

Didn't see T-brd's reply ,nvrmnd.\

EDIT: Even updating to SVN 9012 both games have played well.

JosEPh

 

[Thunderbrd]

Newer game, CTD on next turn . . .

mini and save in zip . .


EDIT: i got Expression: m_DebugCount >= 0 && m_DebugCount < = 1

that was before the game started, then when i clicked on EOT, i got these 2 messages, and poof CTD (see attached pics)

ok here is the save game and the dmp file .

I have VERY good news on this front!

My suspicions proved correct and I've furthermore figured out how to create a healing fix in the code for the bad gamestate. This will take place with a recalculation after the next commit.

What this means is that if you go ahead and recalculate your games after getting the next commit, your games will not only be repaired but they should also stop ever having further trouble with this issue again.

The problem was a fairly nasty little oversight of a danger when I established Stealth Defense functionality so this has been with us for a week or two but had the potential to avoid being truly problematic until the game had been played far past any Stealth Defense events had taken place.

@SO: The REALLY REALLY good news on this is that although the manifestation of the problem was a little different, I was right about it being the same root issue in the larger game you were playing. aka... the dead has risen and your more advanced game is cured! I've even taken the time to prove it with a run through of the round it was crashing at.

However, the bad news there is that the recalc takes a very very very long time. lol

I've played with both the older game and the New game with SVN 9012 and no problems have occurred.

You would think if it was a common problem more ppl would be reporting it don't you?

Didn't see T-brd's reply ,nvrmnd.\

EDIT: Even updating to SVN 9012 both games have played well.

JosEPh

It was quite capable of not manifesting as a problem, even if some stealth defenses had been staged. However, stealth defense events are much less likely in a non-Hide and Seek game so it was less likely to be a problem in your games. That said, it should be totally safe now either way (famous last words I suppose but if there's any other problems it certainly won't stem from the previous flaw in logic!)



Something else that's good that's come from this is that I've been able to further complete verification that the AI is now using its property control units exactly as intended. Which may very well make the AI much much much more dangerous now. Enjoy!

 

[strategyonly]

I have VERY good news on this front!

@SO: The REALLY REALLY good news on this is that although the manifestation of the problem was a little different, I was right about it being the same root issue in the larger game you were playing. aka... the dead has risen and your more advanced game is cured! I've even taken the time to prove it with a run through of the round it was crashing at.

However, the bad news there is that the recalc takes a very very very long time. lol

Danke, i had one take around 15 minutes awhile back, so nooooo problem, lol just glad to hear it,k because that was the best game i have had in 5 years. . .

 

[Thunderbrd]

Danke, i had one take around 15 minutes awhile back, so nooooo problem, lol just glad to hear it,k because that was the best game i have had in 5 years. . .

I'm tickled pink to have been able to solve this for ya. It did look like a great game at the peak of the story of the globe there since you were taking on the 10 ton giant there. I'd have been personally upset to have such a game die at that point too.

Unfortunately, it'll still suffer a little from some property control unit overbuild (trained before the problem was spotted and solved) but now that they'll be headed to where they are most useful, it might actually be more beneficial to the AI than penalizing. Would've been better if they'd have been using the latest AI arrangements all the way through but that's what version improvements are all about right?

It's rather interesting how it was this and the other crash scenario that have not only lead to the fix to both but to the spotting of the need for major improvements in other areas as well. So it's been a good experience for the mod despite it having been a frustration for you I'm sure.

 

[civsmitty]

Nice.. and i can confirm i just loaded and got pass the CTD with my saves and strategyonly's Egyptian type save game too ,,so far so good.. Great Job

Did this bug originate from an much earlier SVN revision ? if so i would like to know cause i got a game on Gigantic Map and i am into turn about 6500 of 14000 and that game was still working on revision 8824

 

[Thunderbrd]

Well... it was a much earlier version but not TOO long ago. Just a few weeks back. When Stealth Defense was added. But either way you'd want to recalc when you update and play a game that's from that far back as it is.

 

[Dancing Hoskuld]

@TB I need to understand the "invisibility" settings for mines both land and sea. Or I can just put them up with a cost of -1 for now and let you do it.

I have been looking at Orion Veteran's Mine Warfare mod old Python version. The Python AI has a few loops it does not need which explains why it was running slow.

 

[Thunderbrd]

Are the mines improvements or units?

 

[Dancing Hoskuld]

Units the AI code does the placing, stepping on/attack function. There are individual mines, mine fields and nuclear! There is a "mission" for the units that arm them which means place them on the map. Once armed you can disarm yours and move them. If you have minesweepers you may be able to remove other nations mines. Otherwise step on them and go boom!

 

[Thunderbrd]

Interesting. With Hide and Seek they can be made to have quality differentiation for their invisibility values. The units will need to be made to support both non H&S and H&S games but that should be quite fun to have to play on. Let's discuss this in another thread and bring up the list of them and we can look at what invisibility types to use - maybe even a unique one just for them

 

[Snofru1]

I am observing a very strange error which unfortunately is not replicable:

I have reached late Transhuman Age now, playing with current SVN. When I play for quite a long time (many turns) without loading the game again then I receive a unit for free - in every city, every turn! Every city on earth gets a Scout Droid, every city on moon or mars an EMP SAM Infantry. When I quit Civ 4 and restart the game then it finishes, no more free units.

This happened now twice, each time after pretty long playing sessions.

 

[Thunderbrd]

I am observing a very strange error which unfortunately is not replicable:

I have reached late Transhuman Age now, playing with current SVN. When I play for quite a long time (many turns) without loading the game again then I receive a unit for free - in every city, every turn! Every city on earth gets a Scout Droid, every city on moon or mars an EMP SAM Infantry. When I quit Civ 4 and restart the game then it finishes, no more free units.

This happened now twice, each time after pretty long playing sessions.

I wouldn't have a theory on this. Maybe some python somewhere having some odd effects? Baffling.

 

[tmv]

I got once a Scout Droid in every city when a war started.

 

[NevFin]

Ctd - svn 9014

Says minidump file is invalid.

Here is the save game before CTD. Ending the turn CTDs.

 

[strategyonly]

@ Sparth

PINK confed units? I believe they are missing the 128

 

[Thunderbrd]

Ctd - svn 9014

Says minidump file is invalid.

Here is the save game before CTD. Ending the turn CTDs.

Thanks! Looking into it now.

 

[Thunderbrd]

Ctd - svn 9014

Says minidump file is invalid.

Here is the save game before CTD. Ending the turn CTDs.

Looks like I didn't catch all possible situations where Stealth Defense could cause a bad gamestate issue. I suspect withdrawal in such a situation was still quite capable of causing a problem.

I've braced the code against the problem further still and I think it should be a good solution now. But the cool thing is, while you're waiting for the code to compile then get a simple test before committing, your game is resolvable by doing a recalc with the version you have now.

So simply recalc and keep playing if you haven't got my next update yet but when players can they should, once again, all update to the next revision I post here.

 

[strategyonly]

@ DH

here it is again that darn Holy Mountain python error??

Traceback (most recent call last):

File "CvRandomEventInterface", line 171, in getHelpHolyMountain1

ArgumentError: Python argument types in
CyTranslator.getText(CyTranslator, str, int)
did not match C++ signature:
getText(class CyTranslator {lvalue}, char const *, class boost:ython::tuple {lvalue})
ERR: Python function getHelpHolyMountain1 failed, module CvRandomEventInterface