Bug Report Archive

[sven]

When issues in the "Report - Current Bugs and Issues" thread are resolved, and fixes posted to the 'stable' distro, I move the associated posts to this thread.

Thus, these posts contain a history of all the various reported bugs that have been marked as fixed.

There's not much benefit to reading through this thread -- but, we do sometimes chat about game design issues when responding to bug reports -- and if you do a search on a particular topic, you'll sometimes find interesting info in this archive.

If any of the bugs described in this thread appear to be less-fixed than I'd hoped, please let me know by posting a comment in the "Report - Current Bugs and Issues" thread.

[sven]

There's a known issue that's proving hard to reproduce. What we know is that during tactical combat, sometimes (perhaps immediately after hitting END TURN) the game will throw this error:

Code: Select all

  Lua state\GUI\~TacticalScreen\ObjectSelection.lua:245: looks like you're trying to push an state that doesn't have a valid reference.


Please keep an eye out for this one if you see it -- and, if possible, try to figure out how to trigger it repeatedly. Any additional information on just what causes it happen could be helpful.

Many thanks.

[mharmless]

I'm not at home so can't double check, but I believe that was the content of the error that triggers consistently on the "Immediate Combat Crash 2" save I uploaded last night.

[sven]

Enemy race attacking my system. two pickets and an outpost stationed over the local gas giant, system Pollux. They win the fight, then immediately get a crash about pop_growth.lua types missing type Gas Giant. Load autosave, go into fight again, start clicking around. Looks like selecting the gas giant and then attempting to get a detailed report on it causes the crash. Load the save, start the fight, self destruct all my units. End of fight crash still. Save attached as "Gas Giant Battle Crash.rar". Load the fight, fight to win. Victory does not cause the crash. Carrying forward.

Got this one replicating consistently. Thanks for the save file. A patch is posted on dev, but, unfortunately, in course of fixing it, I've uncovered another related bug. This one is, hopefully, less likely to occur during gameplay. But, if you see the following, it's a known issue. I can replicate it, and I'm working on it:

Code: Select all

  error : failed to resolve the attack at Pollux:
 Lua state\Orders\pending_attacks.lua:858:
  Lua state\Orders\end_year.lua:158:
   Lua state\Orders\end_year.lua:183:end_year:


[sven]

Defending against an attack, immediate crash in tactical combat again. I've not been able to make this reproduce consistently previously. Sent data. Restart game, load autosave, Attacked at Seralta, defend, enemy advances, slight pause, crash. Uploaded again, save stashed. As before, advancing turn and then choosing to go on the offense does not crash. This one seems to crash consistently at least. Uploaded as "Immediate Combat Crash 2.rar". I know, so creative.

I've been trying and failing to replicate this one. Just to be certain I'm doing the replication pattern right, what crashes for you is:

1) Load the game "Immediate Combat Crash 2"
2) Hit "Next Turn"
3) Hit "Defend" (at Seralta).

After the Phidi scouts move forward, the following error should trigger:

Code: Select all

Lua state\GUI\~TacticalScreen\ObjectSelection.lua:245: looks like you're trying to push an state that doesn't have a valid reference.


Of course, if this bug is a symptom of what I'm afraid it is -- specifically, heap corruption somewhere in the C++ side, there's no reason that I should expect to be able to replicate it consistently, even when you apparently can. Still, it would be lovely if I could get this happening consistently inside a debug environment.

[mharmless]

...
1) Load the game "Immediate Combat Crash 2"
2) Hit "Next Turn"
3) Hit "Defend" (at Seralta).

After the Phidi scouts move forward, the following error should trigger:

Code: Select all

Lua state\GUI\~TacticalScreen\ObjectSelection.lua:245: looks like you're trying to push an state that doesn't have a valid reference.


Of course, if this bug is a symptom of what I'm afraid it is -- specifically, heap corruption somewhere in the C++ side, there's no reason that I should expect to be able to replicate it consistently, even when you apparently can. Still, it would be lovely if I could get this happening consistently inside a debug environment.

Load game list says I was using version r11560 when I made that save. I had updated to the Dev version r11564 for tonight, before reading this. Reverted to Stable, which turns out to be version r11562. The issue is not reproducing for me on r11562.

When reproducing the error, I was _NOT_ exiting sis.exe entirely. I was using the restart game button the crash dialog, and loading up the save I was working with from within the game. If you can send me r11560, I can attempt to crash it on my machine again. If it crashes, great you fixed it inadvertently already. If it _doesn't_, then yeah that seems like a symptom of a more serious problem. That particular save point was crashing for me 100% of the time after hitting defend.

edited by sven: i've tried this binary / save combo, and it won't crash for me. the trouble is, even if the bug does replicate reliably on your hardware, if i can't trigger it under a debugger on one of my own machines, it's difficult to get much info from it. the raw dump you've provided might eventually be useful -- but, i'm still holding out hope that i can figure out what's going on without digging into a raw process log.

Maybe the tool procdump? https://technet.microsoft.com/en-us/sys ... 96900.aspx
I'm attaching a dump I just made of build r11562 sitting at the main menu. I can go ahead and generate one with each crash going forward, to possibly help nail down the corruption issue. I've had to generate these before for devs on other projects, but I personally don't have the skill set to examine or understand the content so I'm not sure if it is useful to you in this context.

[mharmless]

Figured I'd start posting things here as I encounter them

During ship design dragged components have a drop hotzone which is not congruent with the graphical depiction of that location.
Build r11576 on 04 Media PC (AMD A10-6800K, 8gb of RAM, AMD A10-6800K APU graphics, 120gb SSD)

Issue appears to happen with all components, on all designs, by all approaches. I mentioned this in the earlier reports, however I haven't seen anybody else mention this, so concerned it may be another AMD specific problem.

Component Drag and Drop Issues.png (425.62 KiB) Viewed 68218 times

[mharmless]

Enemy scout cruiser design does not use all available component slots. Appears to be missing an item in the System slot.
Build r11576 on 04 Media PC (AMD A10-6800K, 8gb of RAM, AMD A10-6800K APU graphics, 120gb SSD)

I believe this is related to my earlier report about a default design having two warp lane amplifiers. The enemy scout design pictured below does not have all six items it should have.

Enemy incomplete ship design.png (524.8 KiB) Viewed 68217 times

[mharmless]

Enemy ship design uses components to add to munitions, but has no weapons which consume them.
Build r11576 on 04 Media PC (AMD A10-6800K, 8gb of RAM, AMD A10-6800K APU graphics, 120gb SSD)

Got into a fight with it just to make sure. Yup, can't launch anything.

Destroyer with extra missiles, no launch tubes.PNG (636.22 KiB) Viewed 68217 times

[mharmless]

Significant death rate due to overpopulation can cause a crash when examining racial details on the planet screen.
Build r11576 on 04 Media PC (AMD A10-6800K, 8gb of RAM, AMD A10-6800K APU graphics, 120gb SSD)

Brought 1 million Gaiads to a world with 1.4 million of my own people, at a 2 million max. Dropped em down, then clicked on em for info. Crash. Bug report sent in. My own race was not causing a crash when clicked on. Picked up my race, world was now 1m Gaiads and nobody else. World size dropped to 1, no growth. Clicking on them was fine. Dropped another Gaiad on the world, 2/1, clicking crashes. Dropped my race down again, now 3/2. All races are listed with negative growth equal to current population. Clicking my race now also crashes. Sent that in too. Appears to crash if you can force the death rate high enough, but I'm not sure what the threshold is.

As an aside, fractional millions of pop die on their own. Possibly a design decision, just noting that here in case it isn't intentional. 1.4 to start, drop gaiad, pick up own race, planet limit is 1/1 and the 400k of my own race all die off. Drop a whole unit of own race down, pop is now 2/2 and both are stable.

[sven]

Significant death rate due to overpopulation can cause a crash when examining racial details on the planet screen.
Build r11576 on 04 Media PC (AMD A10-6800K, 8gb of RAM, AMD A10-6800K APU graphics, 120gb SSD).

Cool. Got it replicating. Thanks for that.

As an aside, fractional millions of pop die on their own. Possibly a design decision, just noting that here in case it isn't intentional. 1.4 to start, drop gaiad, pick up own race, planet limit is 1/1 and the 400k of my own race all die off. Drop a whole unit of own race down, pop is now 2/2 and both are stable.

Working as intended -- though it's an awkward comprise we might be able to improve on at a later date. Basically, in many parts of the game we're summarizing planetary populations in terms of "units", which are required to be small integer values. But, most of the game logic actually works in terms of real numbers, which can have fractional values. Most of the time, that doesn't cause much trouble. But, having <1 population unit worth of several different living together on the same world feels like it could lead to a lot of potentially strange edge cases. So, the current rules are such that it's just not allowed.

[sven]

Enemy ship design uses components to add to munitions, but has no weapons which consume them.

This, and the double warp lane amplifier issue, are both known issues. They fall under the category of "the ai is currently really bad". They're not that hard to fix, but, with other critical stuff involving tactical systems still badly broken i.e. -- no working anti-missiles or torpedoes, these bugs are not actually at the top of that particular priority queue

[mharmless]

Gaiad colonists do not display a racial graphic above their description.

So far every other race has a high transparency image in that empty spot.
EDIT: Spoke too soon. Just captured some Threshers for the first time. They do it too. All minors?

Gaiad colonists do not display racial graphic.PNG (508.89 KiB) Viewed 68333 times

[sven]

Enemy scout cruiser design does not use all available component slots. Appears to be missing an item in the System slot.

This is actually working as intended, though, I'd appreciate your feedback on whether or not it's actually a good idea.

Basically -- my intention for the tactical UI is that if you have a weapon with 5x attacks, you can actually direct those 5 different attacks at 5 different targets, if you so choose. The UI hooks for this are currently in the game (right and left clicking on the weapon group in question should change how many are queued up for firing).

One implication of this is that, when you have multiple weapons in separate hardpoint mounts that are otherwise functionally identical, it seemed to me reasonable that they should be merged into a single effective weapon group when the ship transitioned from a design to a tactical / strategic object.

I think it is confusing. It was intended to simplify things. And it's a choice I now find myself questioning. Forcing all attacks from weapons that share the same hardpoint to fire together makes some sense. And I suspect that giving players precise control over which attacks they want to direct at which enemies is only really likely to make the tactical game more fiddly than it actually needs to be.

Um -- so file this one under "questionable design choices I'm planning to revisit."

[sven]

Gaiad colonists do not display a racial graphic above their description.

Arguably working as intended. The high transparency overlays are an example of the sort of context sensitive images that I'd like to work into the interactive text displays. Arioch will be redesigning these screens eventually anyhow -- and when he does, the nature and location of those images is certainly going to change. Personally, I think it could be a great excuse to show people more of the cool concept art content that otherwise wouldn't make it into the game.

Given that they're really just temporary placeholders/mockups it's probably not worth setting them for all the races atm.