Catching the city overflow bug

[Leoreth]

I don't know but you seem to be disproportionally affected by it (or you just play DoC a lot).

 

[migck]

Given how much I've played this mod in ctrl-Z mode, I really should have investigated this... because hell if I haven't seen all kinds of weird overflows, almost in every game.
Sometimes it's tech: Moors will start getting a tech every turn, usually in the 600AD scenario, until I notice something weird's happening with a world congress being called in 1300AD. Other times it's culture, I've seen Seljuk Isfahan only being able to work 1 tile, and upon checking I read it "has" -1 billion culture. Sometimes it's gold, with France suddenly getting billions of gold and making spy looting impossible due to the sheer cost of the percentages. Funniest must be when it's GP points, along the usual percentages you'll see one with no sign and a 100000%+ chance for a main battle tank, or a lion, which indeed ends up spawning.
So many wasted opportunities to report savegames. I'll try to be more vigilant from now on.

 

[citis]

I don't know but you seem to be disproportionally affected by it (or you just play DoC a lot).

I'm affected in any game after 600AD. Could you please add some button or at least a python code, so I can hanndle this?
If you can add a function to python that resets the values of a city with given the coordinations, and clears the buffered values (for example oveflow from previous research, culture, espionage and wealth) is enough to me.

Else, you could add checkpoints at any function and log it in the file saved with the game's data. So any time the bug happens you can have the full code flow. Else you can just log the random values of a gameline, so you can recreate a game with them.

 

[Leoreth]

Yes I'm pretty sure I need a serious logging offensive here.

 

[morff]

Does this belongs here:
it happened to me more than few times, but when i try to load some older save, and repeat the issue, it doesn't happen again (always)...
3000 BC scenario, i'm playing let's say babylon or rome, and in indian city Delhi (with the old name)... for some reason they just have all buildings they shouldn't have (because they don't have the technology yet, like airport); and almost all wonders. When i tried to edit the city... it just changed back next turn (then i tried to delete it, but it just got worse and on next turn million exception windows displayed about counting score but that's not the important part).
Also the wonders are on the city / wonders screen (f8) but the year of completion is like 20-something thousands or so...
Looks like there is over- under-flowing something somewhere...

 

[migck]

I have one! It happens consistently in Indian Varanasi at 1904AD, when they finish the hydro plant there, the power bonus causes out of whack production and commerce. I'm attaching the save the turn before.
This is from the 3000BC scenario, started as Maya, switched to Aztec and then to Colombia. And yeah, totally non legit use of WB as usual with me.
(BTW, Colombia failed their third UHV goal as soon as I switched to them, not sure what's going on there)
Great, the file is 2,01MB big and the limit is 2... 7zip, do your thing!

 

[Leoreth]

Unexpected, but awesome.

 

[merijn_v1]

Maybe completely unrelated, but I say it anyway. I'm playing a game of C2C right now, and also here the city overflow bug seem to exist. (extreme amount of science, food and production)

One thing that did catch my eye was that the target city had a +900% (ish) science boost from buildings, whereas most cities only had 100% max. Maybe the DoC COB (city overflow bug) has something similar, but hidden. (buildings that give a bonus out of nowhere)

 

[Leoreth]

Okay, new stuff:

Due to Tigranes' latest save, I could track the issue back to base yield values. Base yield can be changed by a number of events, so I need more logging to figure out what exactly causes the problem. So now, every time the base yield changes, it is checked if it is less than 0 or more than 1000 afterwards (reasonable values can be expected to be within these bounds).

If that is the case, the following will happen:
- a message "Overflow in city [city name] for method [method name]" will appear in red in the interface. This is your cue to report it here
- the same message will also be written into the sdkDebug log in case the game crashes
- an autosave will be created regardless of autosave settings (always during autoplay)
- autoplay will immediately end

At least, that is supposed to happen, not sure if breaking the autoplay actually works.

In any case, if that happens to you, please send me a save of this situation (or use the autosave) and tell me the affected city and the mentioned method name.

 

[Tigranes]

Nice. Very nice. We are on something here.

 

[migck]

Okay, new stuff:

Well in case this is still being investigated, I have triggered such an event. Attached screenshot shows the city at which it happened, and the four R overflow events in the log. It seems the place for city name and event method are swapped.

This was playing in 3000BC with Ethiopia, the overflow happened when the Arabs flipped Greek Buto to Kem Kasir. The flip confirmation window asked me for San'a and Tyrus, though as Ethiopia I only controlled the former, the later I hadn't ever set foot into and was controlled by Rome.

EDIT: and of course, first attempt to replicate it from 620AD doesn't show it... also, I don't think stability is working at all in this game. Forgot to mention I've used 1.13 SVN 946 throughout it.

 

[Leoreth]

Thanks for the report, didn't expect to hear more from this. Too bad I apparently messed up formatting the message. As expected, it's related to flips. I'll look into it some more when I have the time.

 

[Royal Tenenbaum]

It's never happened to me. I must add I only play 600 or 1700, never 3000.

I also never play Normal, I play Epic or Marathon.

 

[LacsiraxAriscal]

I only play 3000, all speeds, and I've never seen it either.

 

[Theodorick]

I've played a good 500 or so games from this mod, using all scenarios, and have never seen it (normally) either.

The only time I have seen it was when I downloaded a save where it could be reproduced. Could that mean anything? Is there something unique about the people who see it every other game? Is there something they might be doing that I am not doing, that can start it off?

For reference, I play marathon ONLY. I don't remember the last time I've played on any other speed. I also never switch civs. I have had bad luck with that mechanic. I swear I get every bug under the sun when I do use it, like cities not existing, or the names vanishing. As a result I just avoid it altogether.

 

[merijn_v1]

Well, it was quite rare. And now Leoreth has included some fixes a while ago, it's even rarer. So it's not that strange you didn't experience it.

 

[Leoreth]

I actually had my hopes up that it's gone for good, but finding out more about it is useful in any case.

 

[Theodorick]

Well, it was quite rare. And now Leoreth has included some fixes a while ago, it's even rarer. So it's not that strange you didn't experience it.

I've been playing this mod since nearly the beginning, so I wasn't referring to recent games.

It's never happened to me even though I've gone on 12 hour a day binges for weeks on end playing only this mod. (Thanks Leoreth. This mod is more addictive than crack sometimes. )

 

[Leoreth]

Excellent.

 

[calad]

Never seen that bug either. Sounds really bizarre.