Single Player bugs and crashes v38 plus (SVN) - After the 20th of February 2018

[Thunderbrd]

@Toffer90 Used to be I could at least attach the debugger successfully to games with PPIO but this one seems to be completely incapable of it for some reason. It seems to think it's running on a different dll or something even though I just compiled a new debug dll just to run for it. I'm not sure it has anything to do with PPIO. It's very late game and on a space map and I'm not sure what other modmods might be at work. I don't know why but I can't get any trace on what the code is doing at all here. Maybe someone else knows more about why this is than I do. I've never seen python adjustments do anything like this. He says this crashes and I haven't the patience to find out if it does, knowing that if it does, it's doing so outside the processing I can see because it simply fails to attach properly (just like if it was trying to attach to a totally different program or something.)

This is a report from discord btw. May just be a problem with the save somehow.

 

[Toffer90]

@Toffer90 This is a report from discord btw. May just be a problem with the save somehow.

I played 12 turns in this save without problems. SVN rev.10564 + PPIO

@Toffer90 Used to be I could at least attach the debugger successfully to games with PPIO but this one seems to be completely incapable of it for some reason. It seems to think it's running on a different dll or something even though I just compiled a new debug dll just to run for it. I'm not sure it has anything to do with PPIO. It's very late game and on a space map and I'm not sure what other modmods might be at work. I don't know why but I can't get any trace on what the code is doing at all here. Maybe someone else knows more about why this is than I do. I've never seen python adjustments do anything like this. He says this crashes and I haven't the patience to find out if it does, knowing that if it does, it's doing so outside the processing I can see because it simply fails to attach properly (just like if it was trying to attach to a totally different program or something.)

Erm, don't know if I can help you with this...
Have you tried debugging a new game with your current setup?
I have a hard time believing PPIO could cause this problem... Perhaps something went wrong during compiling...?

 

[Thunderbrd]

I played 12 turns in this save without problems. SVN rev.10564 + PPIO

This makes me think it was a simple MAF then. He might have hit his computing limits.

Erm, don't know if I can help you with this...
Have you tried debugging a new game with your current setup?
I have a hard time believing PPIO could cause this problem... Perhaps something went wrong during compiling...?

I have a hard time believing it would too. This save is the first time I've encountered such an issue. Nothing seems to be going wrong with compiling. Unless it's a totally new problem I haven't encountered before thanks to some changes I didn't catch. If i have the problem again I'll look into it further.

BTW, in the written lets play thread someone just commented how strange it was that PPIO wasn't core. Man, I so wish it could be for the release... Even if that means we have to do some debugging before then. Take this to mean that I REALLY respect this work of yours! If I'm not mistaken, PPIO doesn't do much to disturb the mod does it? I know there's a few things still needed to rework if I'm following that thread properly.

 

[TaylorItaly]

CTD at EoT and MinDump

 

[strategyonly]

If I'm not mistaken, PPIO doesn't do much to disturb the mod does it? I know there's a few things still needed to rework if I'm following that thread properly.

He does do things that r different and i'd like changed, i have mention a few of them, but i'd like more changes also, not that many just a few, but first the ones i have mentioned . .

And remember ONCE it goes to core, WE r responsible for it and not just Toffer and WE decide what goes in and what goes out. . . . that might be a deciding factor also .. .

 

[raxo2222]

This makes me think it was a simple MAF then. He might have hit his computing limits.

This save takes 1300 MB when loaded.
Someone would have to play it whole day to make it MAF, as graphic paging option (that one in BUG options which got added long ago) is leaky.
This map is 60x147 = 8820 tiles big, so its Large+ sized.
Also hes only one civilization here.
That it was simply nonrepeatable CTD.
He messed with calendar somehow.

I reported, that game takes more and more memory when moving around map but @alberts2 didn't believe it - I even posted one of PIT's scenario saves in Medieval era for him to test.

 

[strategyonly]

havent seen this is a looong time, but now enemy workers r just around enemy civs ALL BY THEMSELVES. again?? i think i have captured like 25-30 of them so far . . .

pic1 enemy workers all by themselves
pic2/3 workers captured all around my civ

dont know for sure if i have a good save there or not ..

 

[raxo2222]

havent seen this is a looong time, but now enemy workers r just around enemy civs ALL BY THEMSELVES. again?? i think i have captured like 25-30 of them so far . . .

pic1 enemy workers all by themselves
pic2/3 workers captured all around my civ

dont know for sure if i have a good save there or not ..

@Thunderbrd removed need of escorts from workers.

I guess its worse on slower speeds when there is more chances of workers getting destroyed.
But overall AI is better off without worker escorts.

 

[Thunderbrd]

havent seen this is a looong time, but now enemy workers r just around enemy civs ALL BY THEMSELVES. again?? i think i have captured like 25-30 of them so far . . .

pic1 enemy workers all by themselves
pic2/3 workers captured all around my civ

dont know for sure if i have a good save there or not ..

As Raxxo explained, that's how I made them get to work again. I might try again to provide some kind of OPTIONAL protection later. The demand was holding things up too much.

 

[Thunderbrd]

This save takes 1300 MB when loaded.
Someone would have to play it whole day to make it MAF, as graphic paging option (that one in BUG options which got added long ago) is leaky.
This map is 60x147 = 8820 tiles big, so its Large+ sized.
Also hes only one civilization here.
That it was simply nonrepeatable CTD.
He messed with calendar somehow.

I reported, that game takes more and more memory when moving around map but @alberts2 didn't believe it - I even posted one of PIT's scenario saves in Medieval era for him to test.

Yeah it didn't seem to be taking as much memory as it would need to for that. I dunno why I couldn't get it to attach properly to this one.

By the way, it's not just the graphic paging option. That panning crash could happen in core Civ if you did it enough, particularly if you were using the 'next city' buttons. It's a problem in the EXE that one.

CTD at EoT and MinDump

I'll take a looskie.

 

[raxo2222]

By the way, it's not just the graphic paging option. That panning crash could happen in core Civ if you did it enough, particularly if you were using the 'next city' buttons. It's a problem in the EXE that one.

Ah so what I noticed was EXE bug vastly outstripping memory saving capability of Graphic Paging option.
It happened when I was clicking on minimap on various places.

So on smaller maps I would actually notice (scenario where I tested it was pit's map - its largest map around here, save was in Medieval era with all land covered by civs) memory usage fluctuations as I click on minimap.

 

[Thunderbrd]

Ah so what I noticed was EXE bug vastly outstripping memory saving capability of Graphic Paging option.
It happened when I was clicking on minimap on various places.

So on smaller maps I would actually notice (scenario where I tested it was pit's map - its largest map around here, save was in Medieval era with all land covered by civs) memory usage fluctuations as I click on minimap.

I'm not sure what exactly you're trying to say but test what happens when you pan around a bunch on Vanilla (or click on a lot of spots on the minimap or go from one city panel to the next in rapid succession.) It can easily be crashed by doing that.

 

[raxo2222]

I'm not sure what exactly you're trying to say but test what happens when you pan around a bunch on Vanilla (or click on a lot of spots on the minimap or go from one city panel to the next in rapid succession.) It can easily be crashed by doing that.

I meant that Albertas (or someone else) was telling truth, that Graphic Paging doesn't cause leak - I didn't knew, that this memory leak was EXE fault.

 

[Thunderbrd]

I meant that Albertas (or someone else) was telling truth, that Graphic Paging doesn't cause leak - I didn't knew, that this memory leak was EXE fault.

That's my understanding and recollection. I thought I'd found a leak in paging myself recently, but my assertion proved false. I can't see anything else in that section of code that would cause it, though it is certainly possible that you can press the memory recovery of it too quickly and bring about an overload of memory usage before it can completely clear its previous usage.

 

[Thunderbrd]

CTD at EoT and MinDump

Not only was this legit, this was concerning that we'd not seen it before. It was crashing when it went to check the specialbuilding type of the default building for a buildingclass during a check to see if that buildingclass was somehow limited. Somehow it was checking the BUILDINGCLASS_NATIVE, which is a rare and special case buildingclass that doesn't use a default building type. Thus it hit a NO_BUILDING definition for the default building and used it to call for the SpecialBuildingType for NO_BUILDING and that's where it comes up with a crash.

How this hasn't been a prolific problem already I don't know. I made it check first to ensure that there WAS a default building for the buildingclass before it went to find the building's SpecialBuildingType but something seems fishier about this than simply protecting the code against that illegal call.

 

[TaylorItaly]

So , you think you can fix this?

 

[raxo2222]

So , you think you can fix this?

It was fixed now - update SVN..

10577

• Fixes a reported crash - what is odd is that it appears to be something that would've been a problem before this point. I was able to easily keep it from happening, but the code found a spot where it didn't like the BUILDINGCLASS_NATIVE not having a default building and I'm baffled as to how this wasn't a problem long before now. It was checking if there was a limit on that buildingclass, surely something it's done many times before. Has anything changed in regards to that BUILDINGCLASS setup recently? I couldn't find any sign of recent changes.

Weird bug.
Now I wonder if it caused some wonders to not place free buildings where applicable in all cities - someone reported that water department doesn't place water pipes in some cities, that were eligible for this building.

 

[Thunderbrd]

Weird bug.
Now I wonder if it caused some wonders to not place free buildings where applicable in all cities - someone reported that water department doesn't place water pipes in some cities, that were eligible for this building.

I don't THINK there's a relationship. I'm not sure what that one is about but I bet a recalc would fix it.

 

[TaylorItaly]

Thank you so much!!!!
I was really sad , as i started a new smaller map without some stuff ,like Barb World ....

 

[JosEPh_II]

Thank you so much!!!!
I was really sad , as i started a new smaller map without some stuff ,like Barb World ....

Now you have 2 games to compare against each other as you play them.