v 26, and above ALL Bugs/Crashes reporting thread

I was getting these due to an error in my recent changes, but though I'd fixed it in my own testing before I pushed it to SVN. Probably you are seeing some variant of the error I fixed hat I didn't quite sort out. Check the BBAI log for the hung turn and see how it ends. Is it basically one unit continually responding to the same contract and finding it's already there?

Whatever it is, I'll try to find time to look into it later today.

yep, spot on.

http://www.filedropper.com/bbai_2
 
The initial error about the MLF is thrown after the CvXMLLoadUtility::logXmlCheckDoubleTypes is called, according to the debugger. When I broke after all of the other errors it was in the middle of kernel code, so I couldn't get any useful info from that.

Yes, it crashed for me in the latest SVN during map generation. I looked at the dump and it's an unhandled access violation in the function CvPlot::isVisible when returning the getVisibilityCount or the getStolenVisibilityCount. For some reason it seems to have a debug variable set to true, even though it's a FINAL_RELEASE DLL. I can't make heads or tails of it, but I'm 99% sure that it's Viewport related. I've attached the new dump (from rev 3848).

These two bugs are still extant, and I wanted to make sure they weren't buried. The first one is preventing me from getting any C2C development at all done, so it's rather major for me.
 
@Koshling & Vokarya:
I've determined what's happening but there's a few things I'm a little confused by. First of all, I have no idea how you could've gotten a unit into this position in the first place except perhaps by having waited for a while until recently to update the SVN. The current wouldn't have allowed it to get all these promos in the first place.

What's taking place is your Honour and Martial Arts II promotions are somehow bypassing every effort to have declared them a 'free' promotion, which must be a result of a failure of the filter set up to designate all promotions on a unit as having been free if the version they were assigned in was previous to a more recent change. This part is Koshling's to work because I'm not sure what changes took place during the rev indicated in that filter or how Vokarya's save somehow got those promotions to bypass this at all.

But the second situation making all of this happen is exactly what Vokarya stated earlier, that Religious prereq'd promos are being removed if you've changed religions. Should happen as soon as any upgrade or promotion takes place after such a change in religion so it suggests there should've been a check on all units to canKeep on all their promos after any change in religion declaration. Since I'm not wanting to include such a heavy processing run after all religion changes, I'm simply disabling the religious obsoletion (at least for now - I have some ideas to flesh that out a bit more elegantly in the future.)

It's also possible that somehow these freely assigned promos are being given to the unit in such a manner as it bypasses the declaration in the code that tags them as having been given free. I'm making an adjustment to make sure this can no longer happen in any case OTHER than having assigned them prior to the SVN changes. Those cases were supposed to be caught by Koshling's new filter so I'll let him adjust that to suit.

Funny that Martial Arts I is a religious based promo... makes some sense I guess but I think there might now be better ways to address that prereq. You were losing MAII and III as a result of losing MAI (took away the promo prereq and thus collapsed the whole line.) You were retraining MAIII properly, but MAII must have somehow avoided being assigned and declared as free so was also indicating a retrain.

Honour as well somehow skipped being designated as free. How you WEREN'T able to retrain those in your first example as a result is a bit odder actually and I haven't seen how that could've taken place. There's a firm bracketing that binds the message to the same condition as adds a retrain to the unit. Let's just put it this way, this structure is not an easy one to make compatible with games in progress but should work just fine once the game works through the quirks that an update might create or through the course of a new game.

I'll hopefully push these 'fixes' to the SVN later today once I clean up another project I was in the middle of here.


Correct: Its just like before, all the archer(s) do there is "bombardment"(bm) but the others wont attack in the stack they just sit there:crazyeye: its BM, BM, BM, BM etc. no attacking.
That was how I remember it working originally, as a bombard mission just like how siege weapons operate. However, I think what you're trying to say here is that there should be a fix made to the DCM stack attack option that made it so that some Archer Bombard takes place against the defender when archers are part of the stack that attacks. That part of the code appears gutted from its original state so its going to be a matter of trying to rebuild the interior of those gutted sections. Stack Attack is potentially buggy from a number of perspectives too so its a bit of an all around effort to make sure the whole structure there is working as intended (and its quite large.) So I'll put that task on my priority list but its not on the immediate front burner at the moment.
 
not interested in changing it, it's simply one of the buildings i'll never build. like a vaccine lab... +20 disease to your city, yea right.

I hadn't noticed this until now. I think I'm going to change that to a local instability penalty. I think when I designed it I didn't quite understand the difference between a local instability penalty and a national instability penalty - the text of local instability says "in EACH city", when it should probably say "in THIS city". The purpose of the instability is to show the effects of disruptive artistic movements. In terms of raw game-effects, it allows me to put a lower hammer cost on the Wonder than you would expect for an Industrial-Era Wonder.
 
Hey, in my current game I got some yet unidentified bug that causes the game to simply and immediately shut down when trying to move on to the next turn. It is a quit huge game and it took some time, but I managed to isolate the issue to a stack of units. When I deleted those and only then, the crash does not occur.
However, I am still playing v 25 (I believe), so maybe this got already fixed. In any case, it took enough hassle for me to find the cause that I decided to upload the game save. Maybe it is of some use.

The stack I am referring to is the SoD of the Cherokee (the violate ones) which moves within the boarders of the Assyrians (the pink ones).

Save can be found here http://www.mediafire.com/?6gvs7wifs1bn58r

I can't seem to find the SoD you mention (can't see any Cherokee troops in Assyrian territory), but the game plays our the end turn ok for me (latest SVN), which may indicate that whatever was happening is fixed. Can you clarify where the offending stack is (either by reference to a city and/or with a screenshot showing me where exactly to look?
 
@Vokarya & Koshling:
Noticed another issue with religious promos given by buildings not being tagged as free. So that had a lot to do with how a situation like the one given would've presented an issue. Now, all promos not assigned via leveling selection will be considered 'free' and never subject to potential retraining.

Update to SVN pending soon.
 
I have found 3 issues which exist in v26+patch, one of them potentially crash inducing (unless C2C changed the way some relevant stuff works in the DLL):

1) Serious (probably) issue: there is a bad unit class - the default unit it points to does not exist. Unless C2C changed how part of the code works, this can lead to the DLL accessing the data for an array element of -1, which does not really exist...

In the main CIV4UnitClassInfos.xml file UNITCLASS_ALARIC is defined with default unit type UNIT_ALARIC, which does not exist. You can delete UNITCLASS_ALARIC since UNITCLASS_GOTH_ALARIC is also in there and it already points to the valid UNIT_GOTH_ALARIC.

Less important related issue: UNIT_GOTH_ALARIC uses TXT_KEY_UNIT_GOTH_ALARIC for its description but UNITCLASS_GOTH_ALARIC uses TXT_KEY_UNIT_ALARIC. Those two text keys have the same text. I would suggest pointing the description for the unit class at TXT_KEY_UNIT_GOTH_ALARIC and eliminating the redundant TXT_KEY_UNIT_ALARIC from CIV4GameText_HEROES.xml. Note that UNIT_GOTH_ALARIC uses TXT_KEY_UNIT_ALARIC_PEDIA and there is no TXT_KEY_UNIT_GOTH_ALARIC_PEDIA (so don't delete TXT_KEY_UNIT_ALARIC_PEDIA).

2nd issue: BUILDING_SLAVE_MARKET has an invalid trait for its ProductionTraits. It has TRAIT_PROGRESSIT which should either be changed to the valid TRAIT_PROGRESSIST or removed.

3rd issue: TECH_FORENSICS has an invalid quote: TECH_FORESICS_QUOTE instead of TXT_KEY_TECH_FORENSICS_QUOTE (which does exist already).

It's possible that some (or all) of this has already been fixed in the SVN, a lot happens in there, but I don't think it has been.
 
@Vokarya & Koshling:
Noticed another issue with religious promos given by buildings not being tagged as free. So that had a lot to do with how a situation like the one given would've presented an issue. Now, all promos not assigned via leveling selection will be considered 'free' and never subject to potential retraining.

Update to SVN pending soon.

I think something went very wrong in what you just did. I just updated to SVN 3888 and the modifier recalculation stripped EVERY free promotion off my units. Even things like the Amphibious trait from Isolationist were lost.
 
I think something went very wrong in what you just did. I just updated to SVN 3888 and the modifier recalculation stripped EVERY free promotion off my units. Even things like the Amphibious trait from Isolationist were lost.

Ok. I have a theory as to what may have done that and it'd be an oops. I'd test against games like yours if they weren't the kind of thing that takes 15 min to load.

EDIT: Ok, should be fixed now. I can't test the recalc on your save cuz it causes me a MAF. But before recalculating, I selected your crusader and gave it Combat I and no problems arose.
 
I have found 3 issues which exist in v26+patch, one of them potentially crash inducing (unless C2C changed the way some relevant stuff works in the DLL):
It's possible that some (or all) of this has already been fixed in the SVN, a lot happens in there, but I don't think it has been.

It is now fixed, and btw, Thanks for pointing this out.;)
 
Got a CTD everytime I try to end this turn after recalculating.

Since the game worked fine yesterday I guess the problem to be in the 3 newest updates, that came when updating this morning.
Currently on SVN 3894
zip contains savegame and minidump.
 

Attachments

Because I set a Great Scientist to wait till a specific Research come up the End turn didn't come up after I set the commands for all my units.
I specifical waited till the Scientist was the last unit then set the Research he should wait for and after I realised it is set to sentry I clicked on a other unit and hit space but no endturn flash up.
 
Got a CTD everytime I try to end this turn after recalculating.

Since the game worked fine yesterday I guess the problem to be in the 3 newest updates, that came when updating this morning.
Currently on SVN 3894
zip contains savegame and minidump.

Fixed in latest SVN
 
Ok. I have a theory as to what may have done that and it'd be an oops. I'd test against games like yours if they weren't the kind of thing that takes 15 min to load.

EDIT: Ok, should be fixed now. I can't test the recalc on your save cuz it causes me a MAF. But before recalculating, I selected your crusader and gave it Combat I and no problems arose.

Do you have viewports on? I had to switch to Viewports because I was getting MAF's due to the size of the map.

I think your latest solution fixed almost everything -- I tried a recalc and the only lost promotion was a Female Shinto Missionary that obsoleted its Morale promotion. Morale was a bonus from the Famen Temple wonder. Did you do anything with Morale? It can be granted for free by both Famen Temple (Missionaries only) and Kong Miao.
 
New bug, not related to the promotions issue. The ability of Great Persons to lightbulb technologies appears to have gotten messed up. My current game is in the early Renaissance. I have several Great Engineers, Merchants, Prophets, and Doctors available. All of them show Axe Making as their current lightbulb technology. If I use a Great Person to discover a tech, no tech is discovered, the Great Person is expended, and no other Great Person can discover a tech - the lightbulb button is gone (I went one turn beyond using the GP to see if it was just a one-turn problem or if it persisted). I hope the attached savegame helps.
 
I can't seem to find the SoD you mention (can't see any Cherokee troops in Assyrian territory), but the game plays our the end turn ok for me (latest SVN), which may indicate that whatever was happening is fixed. Can you clarify where the offending stack is (either by reference to a city and/or with a screenshot showing me where exactly to look?
Picture is attached, the stack with the green mark.
And I see that to play the save with version 26 is no trouble? Perhaps I can just see for myself if this fixes it.
 

Attachments

  • Civ4ScreenShot0018.JPG
    Civ4ScreenShot0018.JPG
    325.3 KB · Views: 51
I have a bug probably related to multi features. (Hotseat save attached. SVN 3895)
The tile directly south of Trireme Alpha (Hean Teav) is a tropical ocean tile, while it should be a tropical coast. IIRC that tile was not always ocean, but I don't know when or what was the cause of the change.

Also there are two tiles near the city of Nesi which is now ocean, but was coast before.
Unfortunately I don't have the before change savegame :sad:.
 

Attachments

  • Civ4ScreenShot0000.jpg
    Civ4ScreenShot0000.jpg
    303.3 KB · Views: 40
  • Civ4ScreenShot0001.jpg
    Civ4ScreenShot0001.jpg
    357.6 KB · Views: 43
  • multifeaturebug.7z
    multifeaturebug.7z
    1.7 MB · Views: 29
Back
Top Bottom