Single Player bugs and crashes v36 plus (SVN) - After the 24th of October 2015

Status
Not open for further replies.
Dancing Hoskuld,

Many thanks , I think your instructions have cured my problem - now back to 3 of the dll's ( instead of 5 ) and it has updated successfully.

DavidR
 
Thanks a lot! So we have 2 different stack attack buttons: One in game options, one in BUG options (RevDCM).

That´s what I see (bad: unit is not active after attack; good: unit is active after attack):

Both buttons turned off: bad
Both turned on: good
Only BUG turned on: good
Only Options turned on: bad

So only the BUG button seems to have an effect.

Well this is interesting. My test results showed:

Both buttons turned off: good - and I stress tested this with 6 different situations. Singular selections attacking cities, multiple selections attacking cities, groups attacking singular defenders in the wild, groups attacking multiple defenders in the wild, singular selections attacking single defenders in the wild, singular selections attacking groups in the wild. In all cases it behaved as it should've.

Both turned on: not tested
Only BUG turned on: not tested

These weren't tested because the BUG option is unsupported and will have numerous bugs in it due to there being a lot of work done that has never been updated to this mechanism since it was broken. What you're likely experiencing with it being 'good' is that whatever bug has been revealed was due to other bugs being fixed in the normal combat resolution coding.

Unfortunately I have yet to experience the problem so as to solve it.

Only Options turned on: bad (maybe - in one certain situation)
When I attacked with stack attack on against the city I found one situation where if the lead attacking unit attacked twice and the city was captured the whole stack was put to sleep afterwards - that seemed to work like you're explaining. Selecting and dividing the stack allowed me to then move the individual units that still had movement. I think I can probably fix that bug by adding a
checkRemoveSelectionAfterAttack(); line after the advance command.


Otherwise, something unique must be happening here... are units withdrawing, being pursued, or what? Are there other units there to defend? (Not that this went without testing.) I also varied tests between units with blitz and units without blitz. Still not finding the problem manifesting here in almost all cases.
 
I have produced a savegame where it is easy for me to replicate. Only 2 groups of Corsairs have movement points left. I can attack with them nearby ships as a stack or as a single unit. When I have both buttons turned off I get the bad behaviour in any case.

Maybe we should compare the settings of our games. Where do I find them for the current game?

The other thing I have reported before occurs about once or twice every turn. After a successfull attack the unit can´t move anymore. I also can´t click on any other unit or city. Every option for the current stack is available but nothing else. Only reloading the savegame helps. Unfortunately not reproducible. The SVN from last week doesn´t show this behaviour.

PS: I can also confirm the behaviour reported by Toffer where the unit doesn´t follow when the defender withdraws.
 

Attachments

"You should use the BBAI log too see what the AI does everything else is just quessing."

Is enouth on this problem because you can see one city using meager and the other dont. Logs for coders, WB tricks for real gamers. ;) Both has its advantages... how long to fix that?
 
"You should use the BBAI log too see what the AI does everything else is just quessing."

Is enouth on this problem because you can see one city using meager and the other dont. Logs for coders, WB tricks for real gamers. ;) Both has its advantages... how long to fix that?

What if that city just finished it's last production?
 
Otherwise, something unique must be happening here... are units withdrawing, being pursued, or what? Are there other units there to defend? (Not that this went without testing.) I also varied tests between units with blitz and units without blitz. Still not finding the problem manifesting here in almost all cases.
Nah, With the current SVN I NEVER experience that units retain selection after attack, (with both stack attack options OFF) no matter the situation.

checkRemoveSelectionAfterAttack() has at least been removing selection on units after capturing 0:strength: units since pre v30
Perhaps the new problem is related to that old one as it behaves exactly the same, only now it kicks in after proper combat too.
 
What if that city just finished it's last production?

Its on 10 AI citys and also no production when you move one more turn forward, i already have tested this, i had that in mind also, but only if i give the AI oral tradition then they start building next turn knowledge inhertiance.... also you see they build nothing when you look into the AI city my savegame.
 
I have produced a savegame where it is easy for me to replicate. Only 2 groups of Corsairs have movement points left. I can attack with them nearby ships as a stack or as a single unit. When I have both buttons turned off I get the bad behaviour in any case.

Maybe we should compare the settings of our games. Where do I find them for the current game?

The other thing I have reported before occurs about once or twice every turn. After a successfull attack the unit can´t move anymore. I also can´t click on any other unit or city. Every option for the current stack is available but nothing else. Only reloading the savegame helps. Unfortunately not reproducible. The SVN from last week doesn´t show this behaviour.

PS: I can also confirm the behaviour reported by Toffer where the unit doesn´t follow when the defender withdraws.
Thanks for the save. I'll take a look tonight.
 
The other thing I have reported before occurs about once or twice every turn. After a successfull attack the unit can´t move anymore. I also can´t click on any other unit or city. Every option for the current stack is available but nothing else. Only reloading the savegame helps. Unfortunately not reproducible. The SVN from last week doesn´t show this behaviour.

I assume you mean once or twice per game not per turn as that is what I experience.;)

This is a known issue. Clicking anywhere in the minimap allows the game to continue. The problem appears to be that a unit is not finishing its move but since it is not repeatable it is very difficult to find.
 
I assume you mean once or twice per game not per turn as that is what I experience.;)
In fact I do mean once or twice per turn. And I never observed this before (it started 2 or 3 days ago and when I returned to an older SVN version because of the other issue I didn´t see it, but I have to admit I only played for a short time). But thanks for the advice with the Minimap, I´ll try this.

Added later: I had it happen 3 times last turn. And yes, you can continue if you click the minimap, thanks again.
 
In this game, as soon as I use the Ornithopter to investigate undiscovered territory, the game crashes.

Also, the bug with units that have valid attack moves left (e.g. hero units) is still there - after the first attack, the unit goes into turn skip mode and has to be activated manually.
 

Attachments

What if that city just finished it's last production?
That's a good point. Could be a non-issue entirely. Not to say its not worth investigating further though. This could explain the civs that seem to get completely frozen (if STRIKE doesn't explain all of those entirely.)

I have produced a savegame where it is easy for me to replicate. Only 2 groups of Corsairs have movement points left. I can attack with them nearby ships as a stack or as a single unit. When I have both buttons turned off I get the bad behaviour in any case.

Maybe we should compare the settings of our games. Where do I find them for the current game?

The other thing I have reported before occurs about once or twice every turn. After a successfull attack the unit can´t move anymore. I also can´t click on any other unit or city. Every option for the current stack is available but nothing else. Only reloading the savegame helps. Unfortunately not reproducible. The SVN from last week doesn´t show this behaviour.

PS: I can also confirm the behaviour reported by Toffer where the unit doesn´t follow when the defender withdraws.
Well... this is frustrating even further still. I carefully reviewed my options and both stack attack options are off. I load your game and ... both grouped and ungrouped corsairs retain selection and readiness to act after attacking.

i further reviewed various options and one I wonder about is quick moves and quick combat (offense) and quick combat (defense). I'll further test on those but I'm not seeing any others that could indicate an issue. I have them off at the moment.

Furthermore, reading the code and following much of its processing isn't illuminating any potential problems either. I'm NOT denying there IS a problem but I'm starting to feel like the mechanic that's testing a car for the issue the customer is complaining about and it's always working perfectly when its in the shop.

Again... stay close by because I'm testing further.

Nah, With the current SVN I NEVER experience that units retain selection after attack, (with both stack attack options OFF) no matter the situation.

checkRemoveSelectionAfterAttack() has at least been removing selection on units after capturing 0:strength: units since pre v30
Perhaps the new problem is related to that old one as it behaves exactly the same, only now it kicks in after proper combat too.
See... that suggests that it must be a setting difference somewhere. Always is a fairly sweeping condition and so far my experience is 'never' so something isn't the same here.

I assume you mean once or twice per game not per turn as that is what I experience.;)

This is a known issue. Clicking anywhere in the minimap allows the game to continue. The problem appears to be that a unit is not finishing its move but since it is not repeatable it is very difficult to find.
I thought that was only a bug on the Rom stack attack... hmm... hopefully I can find and catch this issue.

In fact I do mean once or twice per turn. And I never observed this before (it started 2 or 3 days ago and when I returned to an older SVN version because of the other issue I didn´t see it, but I have to admit I only played for a short time). But thanks for the advice with the Minimap, I´ll try this.

Added later: I had it happen 3 times last turn. And yes, you can continue if you click the minimap, thanks again.
Perhaps I can replicate by playing out a turn or two. But it's not promising so far.
 
In this game, as soon as I use the Ornithopter to investigate undiscovered territory, the game crashes.

Also, the bug with units that have valid attack moves left (e.g. hero units) is still there - after the first attack, the unit goes into turn skip mode and has to be activated manually.

I'll look into it when I can.
 
I said:
Well... this is frustrating even further still. I carefully reviewed my options and both stack attack options are off. I load your game and ... both grouped and ungrouped corsairs retain selection and readiness to act after attacking.

i further reviewed various options and one I wonder about is quick moves and quick combat (offense) and quick combat (defense). I'll further test on those but I'm not seeing any others that could indicate an issue. I have them off at the moment.

Furthermore, reading the code and following much of its processing isn't illuminating any potential problems either. I'm NOT denying there IS a problem but I'm starting to feel like the mechanic that's testing a car for the issue the customer is complaining about and it's always working perfectly when its in the shop.

Again... stay close by because I'm testing further.
VOILLA! Quick Attack (Offense) is our offender!

Now... HOW it's the offender is still a mystery but at least I'm replicating the process. It MAY explain the units not being able to move at all at times but I haven't seen it cause that yet.

Anyhow, I've caught the scent of the rascally rabbit and am on the trail.
 
Is there a reason why 2 folders in the "my mods (unloaded) area are exactly the same??

EDIT: @ DH current SVN commit i get this error message (pic 2) then it CTD's

They aren't the same, your dyslexia is showing;)

Blast! forgot to sync the art definition file:blush: I copied the old over my newer one :blush:
 
I believe you need stack attack (BUG) "ON" for this one to occur.
Probably true given that I have long heard reports of this symptom on that before I started modding.

I'd LIKE to for now just remove the option entirely since it cannot possibly work properly at the moment and won't unless I can get an example of what it was like before I murfed it up when I started work on the combat mod so long ago.

@DH: is it possible to simply disable this option and hide it so it's not obvious the coding is still in place for it to a player who doesn't realize it would be hopelessly bugged? As a 'modderoption' I'm not sure how to go about doing that.

However,

I've FIXED the reported problem! It was actually very deep and involved an old hack to fix a problem with quick attack (offense) that inadvertently caused a problem with stack attack. I don't think the programmer who made this hack realzied he was breaking stack attack.

When I fixed stack attack I inadvertently re-broke quick attack (offense) in the process but I did not understand that at the time. The root problem is a complexity in the graphics engine apparently. As the complexity of the problem became more clear, for a moment it was looking like it would be impossible to solve for both options.

However, while it took some SERIOUS thinking to figure out how to fix both AND keep stack attack from displaying the de-selection of the stack problem as well, at this point, stack attack alone, quick attack (offense) alone, both together and neither all work properly according to all tests. (The time to applause would be now. ;) )

But of course, stress test and if you find some funny bizness afoot, let me know once I've got this fix on the SVN. If we keep having trouble with str 0 units, which I can see have some relationship to this situation and I haven't tested, let me know... I MAY yet be able to figure that out as well.
 
@DH: is it possible to simply disable this option and hide it so it's not obvious the coding is still in place for it to a player who doesn't realize it would be hopelessly bugged? As a 'modderoption' I'm not sure how to go about doing that.

It is quite simple, I think, it is in the XML somewhere. If I get a chance I'll look at it before I head off but that is not looking likely at the moment.
 
Status
Not open for further replies.
Back
Top Bottom