V24 Bugs/crashes reporting thread

[strategyonly]

CTD again. Minidump along.

Not using latest SVN, rev 3003 I believe.
No Viewport, Dyn Units though.

Cheers

Np offense, but why are you playing on an un-updated SVN version that far back, the problem might have been solved already in the newer updates (SVN 3123)?? Thats 120 updates ago?

 

[BlueGenie]

Because I didn't want Kangaroos to stop spawning as some time back DH placed a Kangaroo resource. Checking the SVN log now for more precise version.

Which means I'm on 3103 as I have journalism but not the Kangaroo updates. Said I didn't know which one it was I was on, only believed. *grin*

Cheers

 

[Dancing Hoskuld]

Because I didn't want Kangaroos to stop spawning as some time back DH placed a Kangaroo resource. Checking the SVN log now for more precise version.

Which means I'm on 3103 as I have journalism but not the Kangaroo updates. Said I didn't know which one it was I was on, only believed. *grin*

Cheers

But, I forgot to make some of them spawn on the resource until yesterday. Even so they don't just spawn on the resource. They are much like the various deer in that respect

 

[BlueGenie]

Oh, ok, I can upgrade then. Sweet, and thanks for the heads up DH.

Cheers

 

[Dancing Hoskuld]

I have found a bug that apparently dates back to RoM: AND. I'm not sure why it hasn't always had RoM: AND and C2C throwing Python errors every time you bring up the tech tree.

The issue is in CvTechChooser.py. There is an array know as "self.aiCurrentState" which is initialized to be an empty list. It is used to track which techs have changed states so as to only update those.

My notes:
Found a bug in CvTechChooser.py: techs that can't be researched by player 0 are not appended to the list of techs to keep track of. But the list is just a list (not a dictionary) and the access later is by using the tech number as an index. Unless all "special" inaccessible techs are at the end of the list, it will try to access elements off the end of the list.
Example: if you had just 3 techs and the 2nd tech was not researchable (tech ID 1, since they start at 0) you'd have an array with 2 elements at indexes 0 and 1 but when accessing it later it would try to use array elements 0 and 2 to look up the two tech that are researchable, going off the end of the list.
Note 1: It only checks player 0 to see if they are not researchable. Not the actual player number for the player looking at the tech tree. This could also be considered a bug.
Note 2: if you ever want to have the unresearchable techs show up, you have to remove the two "if" statements that do the "continue" if the tech is not researchable. As it is, they are never actually added to the screen. (The first of the two is labeled like so: "#Afforess Skip Mountaineering w/o Mountains Mod". The second location is not labeled, it is in updateTechRecords.) These are the culprits that cause the issue in the first place, since with out them the unresearchable techs would not be skipped - but they would show up in the tech tree, which is apparently what was trying to be avoided.

My suggested bugfix:
Instead of appending entries to the self.aiCurrentState array, initialize the array to have the right number of elements first, like so:

self.aiCurrentState = [CIV_TECH_AVAILABLE] * gc.getNumTechInfos()

and then just change the value of the item at the tech's index instead of appending to the list each time.
The initialization would be done near the start of placeTechs, and each "self.aiCurrentState.append(FOO)" to the list in that function is changed to a "self.aiCurrentState = FOO".

I have tried this fix, and it stops the Python errors and in general appears to do the job. It probably corrects the some mistake in updating the tech tree, too, but I haven't actually checked (although I do seem to recall some weirdness in that area).

I have made a task for this.

 

[Hydromancerx]

So I tried random civs (10 civs) with the GEM with the Viewports on (default size). It loaded and played fine for many turns. So I thought maybe I should try the 50 civ one to see if it would work. It loaded like before but after pressing enter it crashed again. Here is the mini dump and game file.

Not sure if its a MAF, the GEM map or viewport related.

 

[Koshling]

So I tried random civs (10 civs) with the GEM with the Viewports on (default size). It loaded and played fine for many turns. So I thought maybe I should try the 50 civ one to see if it would work. It loaded like before but after pressing enter it crashed again. Here is the mini dump and game file.

Not sure if its a MAF, the GEM map or viewport related.

@AIAndy - this appears to be crashing in the property solver

 

[ls612]

@Koshling:

Another Viewports bug here. The first pic shows a viewport at 48x32 tiles, and as you can see the minimap has issues with the size of the viewport. Then, when I shift via the Military Advisor, the viewport is all of the sudden bigger?! (60x40 I think) This is a bug, but if I'm right we may be able to switch viewports sizes somehow. Save attached.

 

[Koshling]

@Koshling:

Another Viewports bug here. The first pic shows a viewport at 48x32 tiles, and as you can see the minimap has issues with the size of the viewport. Then, when I shift via the Military Advisor, the viewport is all of the sudden bigger?! (60x40 I think) This is a bug, but if I'm right we may be able to switch viewports sizes somehow. Save attached.

I can't really see what's wrong with the first image - can you be more specific? Also I'm in the middle of some AI work right now (feeling rather burnt out on the multi map and viewport stuff to be honest), so it's likely to be a few days before I get round to looking at this.

 

[ls612]

I can't really see what's wrong with the first image - can you be more specific? Also I'm in the middle of some AI work right now (feeling rather burnt out on the multi map and viewport stuff to be honest), so it's likely to be a few days before I get round to looking at this.

Look at the minimap. I'm pretty sure it is smaller in the second pic than in the first. Also, in the second pic the map ends at smaller X dimensions than in the first pic I'm pretty sure.

Also, did you figure out anything to deal with the revealed-tiles bug? If you don't, please don't feel pressured, I can deal with it manually by closing my eyes for the time being.

 

[AIAndy]

So I tried random civs (10 civs) with the GEM with the Viewports on (default size). It loaded and played fine for many turns. So I thought maybe I should try the 50 civ one to see if it would work. It loaded like before but after pressing enter it crashed again. Here is the mini dump and game file.

Not sure if its a MAF, the GEM map or viewport related.

Just to make sure: This is not one of the GEM versions with a typo in one of the civic definitions, right?

 

[Cyrusfan]

It'll be fixable. Worst case (which would suck admittedly) is that I'd have to have an intermediate transition to totally blank (which would double switching time)

If you have to do that, can you get it to do a quick check as to whether any unrevealed tiles are in the immediate viewing area? The vast majority of viewport shifts once the game gets going wouldn't need the blank transition, and having all those times doubled would suck a lot.

Also, my viewport seems to want to carry my great wall around with it. The wall seems to be in the right spot on reloading, but when the viewport switches, you can sometimes see part of the wall in territory nowhere near me (that should happen during this turn) and when it switches back to my general area, if it's not to the very same unit that went first on the load, the wall will be misaligned from my border.

View attachment 326448

SVN 3103 or 23? (the volcano fix-this version has been so stable I'm hesitant to update) 48x30,2 VP. As I said you probably won't see anything on loading, I just thought it would be helpful to have a save with the wall in existence and a VP switch coming up.

Edit: Just one correction, I went back after leaving this comment and found the wall out of place on loading the save. So I must have started on just the right unit earlier for it to be in place.

 

[ls612]

If you have to do that, can you get it to do a quick check as to whether any unrevealed tiles are in the immediate viewing area? The vast majority of viewport shifts once the game gets going wouldn't need the blank transition, and having all those times doubled would suck a lot.

Also, my viewport seems to want to carry my great wall around with it. The wall seems to be in the right spot on reloading, but when the viewport switches, you can sometimes see part of the wall in territory nowhere near me (that should happen during this turn) and when it switches back to my general area, if it's not to the very same unit that went first on the load, the wall will be misaligned from my border.

View attachment 326448

SVN 3103 or 23? (the volcano fix-this version has been so stable I'm hesitant to update) 48x30,2 VP. As I said you probably won't see anything on loading, I just thought it would be helpful to have a save with the wall in existence and a VP switch coming up.

That sounds like the same issue I reported. It is becoming rather annoying.

 

[draz]

Updated to latest SVN, and almost all units can perform inquisitions now.

Edit: Disabled DO_CACHING, no more inquisitions.


Archery units can get Flaming Arrows II with just Combat III - I guess the pre-requisites are meant to be both Combat III and Flaming Arrows I?

Also, there's 3 units with TXT_KEY_UNIT names instead of their proper ones (listed after T95 in Units list of Civilopedia).

 

[Hydromancerx]

Just to make sure: This is not one of the GEM versions with a typo in one of the civic definitions, right?

Whatever is on the SVN is what I am using.

 

[AIAndy]

Whatever is on the SVN is what I am using.

Which seems to be the bad one still.

 

[AIAndy]

Updated to latest SVN, and almost all units can perform inquisitions now.

Edit: Disabled DO_CACHING, no more inquisitions.

Bug found and fixed.

 

[Alornal]

Waiting for civ error. 15% into an eternity game. It seems completely inconsistent with when it appears in any of my games, but it appears each time. I don't think its specs at all, because of the varied maps it shows up on. No minidump, because the game won't save one with the game bug.

 

[Koshling]

Waiting for civ error. 15% into an eternity game. It seems completely inconsistent with when it appears in any of my games, but it appears each time. I don't think its specs at all, because of the varied maps it shows up on. No minidump, because the game won't save one with the game bug.

Just to clarify - you mean it hangs forever when you press end turn?

 

[Alornal]

Just to clarify - you mean it hangs forever when you press end turn?

Yes. In other saves, I've left it for hours without result. This is the third - I was content to leave it for several minutes (other turns took under 30 seconds).