OOS troubleshooting

[Kael]

Somewhere around patch "i" we started getting OOS errors again. I cant find any issues in code review, but I have a number of suspects. I will need help from all of you to figure out which suspect is the one causing the OOS errors.

I will be attaching various CvGameCoreDLL.dll files to this post. Please test them in multiplayer games and see if your OOS errors go away. There are a few disclaimers:

1. All of these CvGameCoreDLL.dll require patch "j".
2. All of the test files will need to be renqamed to CvGameCoreDLL.dll.
3. All of the test CvGameCoreDLL.dll files will need to be copied into the \Mods\Fall from Heaven 2\Assets\ directory.
4. Everyone in your multiplayer game will need to be using exactly the same version of the test CvGameCoreDLL.dll file.
5. These test files are not for normal play. They all have different features disabled. They are only for OOS testing.

Test Versions:

Version 0: Unmodified patch "j" dll
Version 1: Python callouts blocked

Angrenost- Validated OOS with this version​

Version 2: Block specialized choose production and emphasize

Angrenost- Validated OOS with this version​

Version 3: Disable Groupflags
Version 4: Disable Regenerate Map Fixes

Sowy- Reported that OOS issues didn't occur in one rest with this version
Need someone else to validate OOS issues with earlier test versions and no OOS issues with this version​

If you could grab these versions with your friends and test out a short multiplayer game to verify if the OOS issues still occur and report back in this thread it would be greatly appreciated.

If you would like to validate full patch versions, here are the full patches to test with:

Patch "g"
Patch "h"
Patch "i"
Patch "j"​

ps. remember that all the above CvGamceCoreDLL.dll's all require patch "j". So when you test the oos dll's you will have to have patch "j" installed.

 

[Angrenost]

I tried them with another player and about 8 AI's on pangaea maps with no special game settings and we started each round with OOS with each DLL. The "python callouts blocked" dll made me crash on map start but when loading a save OOS happened.

We don't have OOS problems with vanilla BtS.

 

[Sowy]

Just wondering, do we at least know if the OOS are due to AI change issues?

 

[Emptiness]

Any change could potentially have introduced the problem(s). Since the OOS problems seem to have spiked after Patch I, rather than after Patch H (when the majority of the AI changes were introduced), it seems likely that it is located elsewhere. Either way, the process of tracking down the source is probably the same.

 

[Kael]

Thanks to Angrenost for the testing.

I added a new dll version to test as well as 4 full patch versions to test with. Thanks to the help from everyone isolating this. Once we get it pinned down to the specific change we should be able to get it fixed.

 

[Sowy]

I tested out Dll4 and finished the game on round 277 with 0 OOS errors. However, the AI players merely stayed in its town and built the upwards of 30 warriors, which I thought was weird for one that disabled regenerate map fixes. Well hope this helps. =) Thanks again for all your hard work.

(Edit)

It was a Tectonic -> Pangaea (Large/Temperate/LowSea/Ancient/Quick/RandomShoreline) map with 5 AI (Kandros/Flauros/Decius/Faeryl/Jonas) and 2 Player (Keelyn/Auric). The map settings we had selected were: No Technology Brokering, Permanent Alliances, New Random Seed on Reload, Wildlands, Blessing of Amathon, Living World, All Unique Features No Acheron

 

[Kael]

All the DLL's are cumunitive (they disable everything in the versions about them). I do that because its entirely likely there is more than one thing thats causing OOS's and if each DLL only disables one thing in that scenario we will never isolate an issue.

But thats good to hear. I need to know if patch 3 and earlier caused issues and if anyone else can validate that test 4 fixes oss issues.

 

[Angrenost]

We tried versions 3 and 4, albeit with different computers than last time. With every version our problem is that we start the game in OOS, but with 3 and 4 we could load a save and continue without OOS (we played two games until 100 and ~80 turns with standard settings).

I can start with 2 civs (two pop-ups at start, then starting as either one), and when reloading dungeons and lairs and special land marks switch places, as well as spawns, at least from my point of view. The host naturally doesn't seem to have such problems.

There should be more testers =/

 

[Emptiness]

Considering how many people have posted OOS reports the (apparent) support for this effort to correct them has been surprisingly lackluster.

 

[CurseUppl]

Indeed. If it weren't for me waiting for my RAM to arrive I'd be testing

 

[idiot2]

Me and a friend would like to test these dlls. We are wondering about the influence of Kuriotates and City of a Thousand Slums however, since supposedly these can cause OOS too.

Should we exclude them from our games?

EDIT: When playing normally (patch j, 2 players, via direct ip), we usually get persistent OOS around turn 300-350. Is this late OOS even the kind you are looking for ?

 

[readercolin]

hmm... I just noticed this page now. I'll see if I can test these next week. A few things to note from my own playing however:

Sometimes the game crashes on turn 1, but can be loaded from a save. Some games when doing this are completely playable to a finish after that point - other games are a lost cause. This doesn't seem to have an effect on OOS issues.

If you get OOS's in the frist 100 turns, you game WILL be toast. Note, if you get just 1, it might be salvagable, but if you get more than 1, your game will be a lost cause.

Sometimes you get random OOS's between turn 200 and 300. About half the time the game is salvagable from these OOS's.

Anyways, I'll try to test these over the next week. If I can't though, I can guarantee testing time the week after that.

-Colin

 

[dieffenbachj]

Considering how many people have posted OOS reports the (apparent) support for this effort to correct them has been surprisingly lackluster.

Eh, this thread is located midway down on the forums. I've been using patch g since patch j came out since patch j is unusable to me, have frequently complained--yet it's only because I was bored and felt like reading through the forums did I notice this thread =P

I'll set up a game tonight or tomorrow and test a dll.

 

[idiot2]

We tried version 2 (pangaea, small, 7 civs, 2 players, living world, blessing, no tech brokering, perm.alliances) and didnt have any OOS until the game ended in turn 360 (We didnt build City of a thousand slums and there were no Kuriotates in the game). We will try to cross-check the other versions in the next week or so.

 

[Bill Bisco]

OOS caused after Clan Worldspell was cast. Another OOS later in the game sub turn 70. You really need to play your own game with a score of online matches before releasing a public patch.

The game is unplayable for multiplayer now.

 

[mimic]

The game we played last night, I used the clan worldspell and there was no oos in fact I haven't had any oos errors during mp games for ages. Also Kael's asking players to test it isn't he? just fixing these errors is probably a crapload of work on its own (nevermind how long it would take to test a "score" of online games), no offense but its better to help than post negative criticism's. (or at least think about what your posting)

 

[Dionesos]

OOS caused after Clan Worldspell was cast. Another OOS later in the game sub turn 70. You really need to play your own game with a score of online matches before releasing a public patch.

The game is unplayable for multiplayer now.

1. This is no game, but a FREE mod
2. The support for this free mod is much better then for most of the full-price-games

greetings,
Dio

 

[Emptiness]

OOS [...]. Another OOS [...].

[...] there was no oos in fact I haven't had any oos errors during mp games for ages.

Communicating information that Kael can use is important here. Add to those posts what Mod versions you're using, and your game settings. Then, if you are experiencing OOS errors, test with the four test versions (using the same game settings) and post the effect on the nature, character or frequency of OOS errors.

 

[mimic]

Vanilla FFH patch J (as it is most stable for MP) game settings vary depending on who is hosting...but usually along these lines:
Living World
All unique features
require complete kills
No technology Trading/brokering
Raging barbs (sometimes)
Aggressive AI (we rarely use AI though due to the nature of our games, they take away from the role playing aspect)
AI no building requirements
AI no level requirements
No Orthus (if we use a classical start)
No Acheron (if map is small)

All seem to work with no problems.


Amount of players can be around 3-5 +, OOS errors are usually caused because someone has a modmod installed or is on a different patch from everyone else. I believe there was an OOS problem with the Infernals where it was if AI or human took a city it would OOS? Cant remember if that was fixed or not and I never encountered that particular problem.

If I do get any OOS errors that aren't caused by the usual stuff then I'll be sure to test it with the test files Kael posted though as of yet everything seems to be working fine.

 

[Emptiness]

I believe there was an OOS problem with the Infernals where it was if AI or human took a city it would OOS? Cant remember if that was fixed or not and I never encountered that particular problem.

That was fixed in patch H. I'm not sure whether the fix was a reversion to the previous code (which would have restored the bug where the Infernals always reduce the AC when destroying a city, even if it did not originally have AV present) or if it was new code that corrected both the OOS and the invalid AC decrease.