Cause of scale-loss bug found!
 

Edit: It turns out my conclusion in this post is wrong. See my third post further down this thread for details.

Hi all.

Please forgive me if this has already been discovered, and please note it may not be the *only* cause of the bug, but...

Certain players, myself included, have occasionally been the victims of a bug where positive Order and Growth scales would disappear after you end the first turn. Sometimes they would start coming back right away, sometimes they would take years of turns to return. So far as I know, no one ever found the cause.

I was putzing around with a modded map, playing with magic sites, so on and so forth, when I hit the bug after making one single change: I added in the "No IndyPendant" (heh) lines. (#setland 1 #poptype 99 #setland 2 #poptype 99 etc etc)

Hmm!

Some experimentation later, and I discovered that if a player's starting province has a specified #poptype in the .map file, then nearly every time the Order and Growth scales will vanish after the first turn!

This is reproduceable. I've just spent the last hour or so creating dozens of games, all with the starting province containing a #poptype. I've tried different maps, poptype numbers, nations, eras, loaded mods, pretenders, land and water starts, etc etc. Every once in a while the Order and Growth scales would remain, but if I recreated the game with the same settings, they would always vanish again.

I'm not a programmer. I can't find the reason *why* this happens. But this is what causes it to happen--or, at least, it's one of the causes. About the only other variable I can't account for is that I'm playing on an WinXP computer. It's possible that Linux or Vista does not reproduce the bug.

Hope that helps!

--IndyPendant.

Re: Cause of scale-loss bug found!
 

Damn. That's a fairly serious problem with using NI maps then.

Re: Cause of scale-loss bug found!
 

IndyPendant
If what you are saying is true, we are all going to need hail you as a hero. Which poptypes exactly are affected?

Re: Cause of scale-loss bug found!
 

http://koti.welho.com/ehalttun/Image...s/icon_eek.gif http://koti.welho.com/ehalttun/Image...s/icon_eek.gif http://koti.welho.com/ehalttun/Image...s/icon_eek.gif http://koti.welho.com/ehalttun/Image...s/icon_eek.gif http://koti.welho.com/ehalttun/Image...s/icon_eek.gif

Forgive you?

*whacks IndyPendant over the head with a clue-by-four*

Now, kindly tell me what in the name of everything holy and unholy made you think you needed to apologize beforehand when posting THAT? As if we haven't been looking for the cause of that particular bug since Dominions 2 and nobody has ever had ANY idea what caused it and it's nearly given several people here apoplexy. This one is about as close to the Holy Grail of Dominions 3 bugs as is possible and such a detailed description is going to help narrow it down greatly.

*runs off to update the shortlist*

Re: Cause of scale-loss bug found!
 

Heh. Well, I'm sick today, but my mind is clear, so I've had time to play around.

Further clarification: It only affects a player's starting province. So every province except the starting province can have a #poptype, and it shouldn't affect the scales. So far as I can determine, it happens with all #poptype numbers.

NI maps should still work--you just can't have random starts. You'll have to specify starting points, and remove the #poptype command for those provinces. If you playing using random-position NI provinces...yes, you'll most likely encounter the bug.

Re: Cause of scale-loss bug found!
 

I am unfamiliar with mod and map commands. What does #poptype mean exactly? Wouldn't every independant province have some poptype?

Re: Cause of scale-loss bug found!
 

#poptype is a map command used to set certain poptype (like wolf tribe barbarians or knights and longbowmen) to a province. If you don't set a poptype with the command, it will be random.

Indypendant, you are a GOD. Consider yourself having one white candle burning in Kuopio, Finland.

Re: Cause of scale-loss bug found!
 

Normally each province has a randomly determined poptype. Poptype numbers run between 25 and 90, but some cannot appear randomly. #poptype <nbr> in a map file sets the poptype of a province to the specified poptype instead of having it randomly determined. If you set a non-existent poptype (i.e. 90+), no units can be recruited from that province.

Re: Cause of scale-loss bug found!
 

I hope this gets fixed in the last patch. I got hit hard with this in a game where I had order 3 prod 3 taken away early in the game. T'was a hard kick in the nads.

Jazzepi

Re: Cause of scale-loss bug found!
 

I was burned badly in Epic Heroes II by this infamous bug! A conspiracy I say. A conspiracy. http://forum.shrapnelgames.com/image...es/biggrin.gif

Re: Cause of scale-loss bug found!
 

IndyPendant, this is, as far as I am concerned, the most important post in this
forum in the last year. If I have to point out to the most annoying bug, this
is it. And now there is a chance it gets fixed.

I suggest the next patch contains a mercenary sage named IndyPendant, with at
least 20 research, and bad event cancellation. :-)

Re: Cause of scale-loss bug found!
 

While IndyPendant's observation is very keen and may prove useful, I'm quite sure I've gotten the bug in places no #poptype was specified, at least for water nations.

Re: Cause of scale-loss bug found!
 

Still, it shows the connection between pop types and this bug. Very very important find. I was hunting the cause of this bug since Dom2 beta. Never really managed to reproduce it.

Damn good job IndyPendant.

Re: Cause of scale-loss bug found!
 

If it's linked to poptypes, it could be linked to certain specific poptypes more than others and if the underlying poptype is a problematic one, it could cause the problem. So, as Daynarr said, this is a very important thing. It's the first concrete clue since Strages Sanctus managed to reproduce the bug on a specific map in a specific province and a damn sight more informative and general than that one too.

Re: Cause of scale-loss bug found!
 

I am wandering around this forum since at least one year and heard full of angry comments against this bug.
Personnaly, I never experiment it (or more probably, I never paid attention to it because I'm playing for fun and never look at the scales except at some very specific moments, like cold temperature when attacking C'Tis ... http://forum.shrapnelgames.com/images/smilies/wink.gif).

But nontheless, I have to applause IndyPendant for his finding.

I fully agree with Tuidjy : the next patch should include an hommage to this Great Man (not even a God !!! Damn' !).

It has to be said.

Now, back to work http://forum.shrapnelgames.com/images/smilies/frown.gif

Re: Cause of scale-loss bug found!
 

I have a single data point. A couple of weeks ago I tried a map edit with #poptype 30 in my starting province and did not have the problem. I also placed a #knownfeature in the starting province and a buffed pretender. Nation - MA Tien Chin.

This was on a approx. 210 province map generated using random map maker. Approximate settings were height 3000, width 2000, 210 provinces, water 12%, mountain 50%, rugged 50%, river 700. (Values are pretty close, all my random maps are made with approx. same settings.)

May be data, may be noise. Good luck.

Re: Cause of scale-loss bug found!
 

The point of #poptype is to search wich poptypes generate the problem. it´s not the command itself guilty for the bug but certain poptypes (domsuckers) according to what IndyPendant said.

Re: Cause of scale-loss bug found!
 

Most interesting. I'll take a look at it and see if it works the same for me. Thanks!

Re: Cause of scale-loss bug found!
 

Heh. Thanks guys. I may have to bow out though, redfaced.

My conclusion appears wrong after all. : (

While I was experimenting last night, one of the things I did was remove the #poptype command from the starting province, and I kept the positive scales on the second turn. I didn't think to recreate *those* game settings though, just in case that was a fluke. I went to bed, ill and tired, but confident I had it right.

It seems I don't.

This morning, I went ahead and created a game to play, with the poptypes edited out--and the scales vanished.

Oh, crap.

Some further troubleshooting later, and I can still reproduce the bug with like 98% reliability or so--but I've discovered it's not the #poptype command after all. By making this post last night, I now feel like I have some 'ownership' in finding the cause. So I've spent some more time experimenting with this, to see if I can track down the real cause after all.

So far, I've only been able to find one reliable pattern: when playing MA R'lyeh on certain maps, I can create the scale-loss bug nearly every time.

I'd like to ask someone else's help now. Here's what I'd like you to try: start the following games on the following maps, and see if you repeat my findings:

Change *only* the following settings when creating the game: Middle Age, Human player from Random to R'lyeh, Pretender to Master Lich (though the bug appears for me with most Pretenders), Scales to Order-3, Production-3, Cold-3, and Growth-3. Don't change anything else, then when the game starts end the first turn. If you don't get the bug, recreate that game with the same settings just in case. See if you get these findings:

--Silent Seas: bugged.
--Silent Seas Wraparound: bugged.
--Lonely Mermaid 1 & 2: not bugged.
--Glory to the Gods: bugged.
--Cradle of Dominions: not bugged.
--Aran: bugged.
--World of Geometry: not bugged.

I...am at a loss to determine a pattern from those maps. Some have been edited by me, some have not. Some are wraparound, some not. No maps seem to have any particularly unique commands or characters that would cause it. At first I thought it might be all the maps installed with Dom3--but Cradle isn't bugged. Other nations such as Formoria, EA Atlantis, MA Ermor, and MA Abysia aren't bugged, on the same maps that R'lyeh is. The #poptype command works safely on the Lonely Mermaid map.

Still, reliable repetition is the single most important step in tracking down any bug--and I can at least achieve that, even if I can't figure out the damned pattern. ; ) (Was Cradle not available in Dom2, but Silent Seas, Aran, and Glory were for example? If so, then all the non-bugged maps weren't around for Dom2...)

Sorry to get your hopes up, guys. The good news is that you can keep using the #poptype command; it doesn't look to be the cause. I'll edit my original post accordingly.

...bah.

Re: Cause of scale-loss bug found!
 

We are not able to reproduce the bug. Neither on win nor Mac.

We tried Silent Seas with your scale settings without success.

How far into the game does this happen? First turn or after a while?

Re: Cause of scale-loss bug found!
 

IIRC this happens on the second turn. First turn scales are normal, next turn all positive scales are gone.

Re: Cause of scale-loss bug found!
 

What about copying the game folder and testing for the bug with that spesific file? IndyPendant can reliably cause it in his own computer. With even one game that causes the bug reliably, on his computer, we can test if it happens on the devs' computers reliably. It could be it only happens on certain OSs, or with certain hardware, or something similarly obscure.

Re: Cause of scale-loss bug found!
 

How many other nations did you use?

Did you do anything except press end turn?

Re: Cause of scale-loss bug found!
 

Heh. This bug is starting to seriously annoy me. --In a good way, though; I love a true troubleshooting challenge! : )

I can still reproduce the error, but now with only about a 50% success rate. Still, I've got a game I've created, copied to another folder, then ended the first turn. Got the bug, copied that same game--the exact same files--back into the savedgames folder, ended the first turn again, no bug. Then no bug, bug, bug, no bug, no bug, bug, bug, bug.

I've attached the files for anyone to try that would like to. Game folder is 'Bug1', just as it is in the archive. I'm using the default Silent Seas wraparound map that came with the game. I'll be interested to see if anyone can reproduce this bug. If the bug doesn't occur, please try ending the first turn at least five times total to see if you can reproduce it. (Note: DO NOT end the same game five times, to reach Turn Six! I mean delete, recopy, and end the first turn up to five times to see if you can reproduce the bug.)

K.O., it didn't matter if I gave orders or not. If the bug kicked in, it *always* kicked in after ending the very first turn. When I first encountered the bug, I was giving my first-turn orders then ending the turn. But when I began my troubleshooting, I would just create the game and immediately End Turn, and I still got the bug. The bug takes three forms, so far as I can observe: 1) loss of all positive scales, which will come back slowly similar to converting from enemy dominion; 2) loss of all positive scales, that only come back very slowly, so that even after two years and more it still hasn't returned to normal; and 3) slight loss of positive scales, which spend the next year or two 'making up their minds' whether to go up or down, but never quite vanish or return to normal.

I have checked a few other nations, but could only reliably reproduce it with MA R'lyeh.

I'm sick again today, so I'm going to be around for at least another four hours, and I'll probably be around tomorrow as well with the way I'm feeling. So I've got time to work back and forth on this if you'd like. Gives me something to do. ; )

Edit: Took me a short bit to figure out how to upload it, since it wouldn't take the .trn file. Zipped it is!

Re: Cause of scale-loss bug found!
 

Well, this is a bit annoying.

I've just tried a few dozen more starts, and I can no longer reproduce the bug on the Aran and Silent Seas (nonwrap) maps. Just in case the previous file doesn't work though, I've attached a zip of a Glory game I've created where I can also reproduce the bug. My Glory .map file is heavily edited, if it matters. ; ) --And after doing all that, I went back and copied Bug1 in again and tested it: still getting the bug semi-reliably.

Oh, another detail, in case it wasn't clear: if the bug doesn't show up after ending the first turn, then it never shows up.

Re: Cause of scale-loss bug found!
 

...and now Cradle is bugged for me, when it never showed after dozens of tests earlier today. *sigh*.

Here's yet another file for you to try. Cradle is also heavily edited by me btw. When I tested this file, I got bug-no bug-bug-bug.

I'll stop for now, and see what happens when you try these files, before I continue troubleshooting on my end.

Re: Cause of scale-loss bug found!
 

Okay, I lied. I also remembered reading somewhere about the diagnostic command. Here's a log file of Bug1. To generate this log file, I created a Dom3.bat file with the command "Dom3 -dddd". Then I ran the batch file, selected "Bug1", ended the turn without entering any commands whatsoever, selected "Save and Quit", and then quit Dom3 entirely. I did nothing else.

Turn 1 I had Order-3, Prod-3, Cold-3, and Growth-3. At the start of Turn 2, all I had displayed was Cold-3.

Hope all this helps you track it down! --Now I'm really done, for now at least. ; )

Re: Cause of scale-loss bug found!
 

You wouldn't have zip (instead of rar) perhaps http://forum.shrapnelgames.com/image...es/redface.gif

Re: Cause of scale-loss bug found!
 

Hmm...I don't. WinZip costs money, and I only pay for things I'm going to use reasonably regularly. I almost never bother compressing things any more. ; )

WinRAR is available for download here. The version I have is quite old, but also completely safe; I'm a bit of a paranoid when it comes to spyware and viruses. The new version should be safe too.

Alternatively, someone who has both WinRAR and WinZip might be willing to download all three of my rars, and zip them up together into one file to upload...?

In the meantime, here's the Bug1 .trn file. I just renamed it .txt; the data should remain the same. Hopefully you can change the extension back to .trn and take a look at it. If this works, and you need me to, I could upload the other two files this way as well...

Re: Cause of scale-loss bug found!
 

WTF? You can create zip archives just fine with WinRar, I do that all the time.

Re: Cause of scale-loss bug found!
 

I played the game on my pc, and the scalbug happend 4 out of 5 times, so its quite reliable, i attached the output from:

./dom3_x86 -dddd > scalebug.txt

i hope you can use the file

Re: Cause of scale-loss bug found!
 

And there is an "unrar" available for linux if that's the issue.

Re: Cause of scale-loss bug found!
 

...And I just finished figuring out how to do it, too. Heh. Never done it before.

Hopefully my last upload. I've used WinRAR to convert the files to .zip format; please let me know if they don't work for you, K.O.

Re: Cause of scale-loss bug found!
 

in Winrar you can just select to save as zip (well in my version atleast, maybe it even came with vista?) though zip produces a bit larger files..

It is strange though that the inferior program is used here while winrar is just as easily available..
strange pplz here

Re: Cause of scale-loss bug found!
 

WinRAR may be better than WinZip, but I have been using PKZIP.EXE since 80s... and picked WinZip from first batch of graphical user interface heirs. I'm sure that many are like me :-)

And I prefer WinZip's interface to WinRar's interface.

But no, we are not strange. We are old :-) (I'm over 60, and too damn old to learn good English, or change my programs with no good reason)

Re: Cause of scale-loss bug found!
 

R'yleh with cold executes a non-standard Income temperature scale function.

Patala, Lanka, Abysia?, and some others will also call a "temperature adjustment" function that is non-standard.

Has anyone experienced the scale bug on a race with normal temperature scales?

Re: Cause of scale-loss bug found!
 

I have ran into the bug with machaka.


More severly In a game I am in now, I ran into the bug as eary age r'lyey, and the scales have not come back to what they should be, and it is past turn 60. I was supposed to have order 3, but had to play the entire game with order 0, misfortune 2. (And I did have cold 3 for the record.)

Re: Cause of scale-loss bug found!
 

Machaka prefers Heat Scale +2

Re: Cause of scale-loss bug found!
 

Since it must occur on the first turn, and the first turn is early spring, I suspect that Cold-3 would more likely flip the switch. But Heat-3 might do it as well.

It may be an overrun to Cold-4/Heat-4 that does it, maybe for races with a Non-Standard Income for temperatures. This triggers a not so heavily tested function.

Re: Cause of scale-loss bug found!
 

I use 7-zip, or TugZip.

Free is good!

I've never had a scale bug with Agartha or late-era Atlantis.
The first doesn't have a special interaction with heat, the latter does...

I wonder if I can produce the scales bug on my machine. I've never had it, but my sp 'try-out-a-new-nation' games often involve scales of zero all around and a random pretender.

I'll have to try a race like R'lyeh.

Re: Cause of scale-loss bug found!
 

I use alzip, also free. User friendly and can open and zip in dozens of formats.

Re: Cause of scale-loss bug found!
 

Just to confirm, I can reproduce the bug...
This is the first time I've ever seen the bug, but thanks to IP, I know a decent how...

Windows Vista Home Premium Laptop
Acer Aspire 9410-2028.

Re: Cause of scale-loss bug found!
 

Still happens if I do it with the same pretender, dormant.
Same pretender, settings (lich, ma R'lyeh) - reproduc
Same pretender, settings, but dormant - reproduce
Same pretender, settings, dormant, and 8 dominion - happened not once (though I did lose one order scale, one productivity scale in one case) in five or more tries.

As soon as I stayed dormant but dropped my dominion back to three, I got it immediately. Then I started a couple with dominion 8, and avoided the problem.

I swith to a MA Pangaea with the original lich, or a lower dom. Manticore, no problem.

I switch to MA Atlantis, no real problems (just one missing cold scale, twice) in five tries.

Four AM thoughts (brainfarts): Dominion overflow, random numbers, nations out of bounds into random seed, L. Ron Hubbard

My original thoughts lied in the direction of changing the scales, but something led me to focus on dominion instead.

I'll try a few other things sometime, see if I can reproduce with R'lyeh with various chassis- focus on inherent dominion High, low, then tweak.

P.S. Going back to the start, dropping all scales to two, and adding luck or going wasted points does not result in a bug. I have to wonder what happens if I remove particular scales. Before I lose attention, I thought the results of switching the cold for misfortune 3 was interesting, though too small a set of data.

LBob has that handy thread with people mentioning what nations they've managed to get the bug with- some mention of scales as well...

Re: Cause of scale-loss bug found!
 

@Lord Bob
Good thinkings, but i tested it with several nations and normal temperature got the bug in the same matter heat/cold nations get it. So temperature isn't the problem!

I thought that dominion spread my cause the problem so i did some tests with imprissoned/awake pretenders and start dominion 1/10. Well i found out that the bug doesn't correlate with that.

What i found is that the bug only occurs in some provinces. And when it occurs than in a 60-75% case.

Well i can't find any special in this provinces. I'd the feeling thats provinces with a lot of neighbours but there are also provinces with only 4 neighbours.

However from my tests with silent sea (wrapable):

The critical provinces were:
1, 11, 15, 28, 30, 35, 44 and 49

No problems with:
3, 4, 7, 8, 12, 13, 14, 17, 19, 38, 39, 41, 43
(i'd 5 runs without the bug but maybe a few number are bugged anyway)

The rest provinces aren't tested.