Question for Hurley

Other than complaint about bug still existing on various bug threads, the other information was in private communications with Cari. My last offer re debugging was sent to Cari in November, and was never responded to.

I have now reduced my actual time playing in total and in individual sessions and changed galaxy size/style to not encounter the bug. (Note difference in size & submission rate since early December.)

Stardock's response (to date) has been totally consistent.

As is probably too clear I am not impressed. BTW, the reason for editing was to make less like flame. If it still offends I apologise, that isn't the intent. If you want to discuss off-line [email protected]
[Message Edited]
[Message Edited]
[Message Edited]

Thank you

I actually have an update on the status of the invisible ship bug.

We have a special debug mode that lets us run games of Gal Civ where it will just ignore the human player (essentially it's like if you just kept hitting the turn button without doing anything, but the game doesn't end until an AI player wins). 

So I setup a gigantic galaxy with incredible aliens, and let it go, using my laptop to work on other things. About an hour later, the Altarians had won (apparently by culture) and so without restarting Gal Civ, I loaded their EndSaveGame and started looking for missing graphics. 

The graphics that were missing weren't even listed in memory, and just trying to load them may be junkifying memory (I'm still in the process of trying to reverse engineer exactly what happens to cause the bug based on the current results in memory).  Working backwards is, of course, much harder than being able to catch it just before it happens, and watching what happens.  With any luck, I'll be able to find a way to catch it just before it happens with what I find out by working backwards.

So I don't know for certain if the bug is caused by sheer quantity of ships or the quantity of different kinds of ships.  Or both.  If I can narrow down exactly what area of code is causing the problem, I will be able to make a new estimate of how long it would take to find and test a solution to this problem. If the new estimate is more reasonable than 2 weeks, I may be able to get the go ahead to try fixing the bug.  However, the problem is still obviously deep in the guts of Pear, so it is not going to be a quick fix. 

G'day Cari,
some observations that might assist.

It occurs, for me, whenever a new type of ship was built/encountered rather than lots of ships of a few types. So I had it happen after a new type of ship was discovered in an anomoly that I hadn't buily before. (It may occur for many ships in quantity but I don't play that way.)

For others, my consistent way to reproduce was to have other memory hungry applications running (for example outlook, IE with various plugins & java used etc), start galciv (gigantic/crippling) in windowed mode, and switch in/out of galciv every few turns, use windows media player outside galciv maybe other apps. 99% of the time within 30 minutes first missing graphic would occur followed by other dissapearance, interestingly usually starbases! (But that is probably due to me style of play with lots of starbase construction in some games.)

Environment, was as I remember, also I have recently changed for completely separate reasons:
- windows 98 se
- 384 mb memory (windows managed swap)
- 2-3 gb free disk (less than 1% fragmentation - I do that every week at least and always if there is a game problem)
- windows media player
- 1.3 ghz intel
- can't remember graphics & audio but current for about 1999 with latest drivers from turtle beach etc
- fully dx compliant according to dxdiag

Good luck.

Thanks Cari, much appreciated!!

No worries, BTW, ssen sirian's attack on ray in fotr thread?

Hurley, Cari asked me to do some checking on stuff for ya....what was the OS the puter had when you got it new? cause downgrading to a os thats older then the 1 that came with it isn't gonna have the drivers the hardware needs to run right at all

if the puter came with lets say XP home downgrading to 98se isn't a good idea

~SDC~

No worries David and it isn't what I did.
Machine was a Dell dimension T600, with a subsequent powerleap cpu upgrade to 1.4 ghz. Initially installed with 98 and upgraded to 98 SE and always up to date with security etc patches. BIOS couldn't take upgrade to XP so was never done.
I was also very, very careful with drivers as I used the machine for voice recognition software due to a problem with my wrists.
If you want to take off-line [email protected].

G'day HtL, I have posted more details in the thread titled" Question for Hurley". To respond to your questions though. Memory chip fine, was checked with diagnostics, behaviour not happenning with other 'hungry' games/appls such as morrowind/dragon naturally speaking. No memory manager running. Nothing unusual, as far I can tell, also regularly use spyware removers etc to ensure clean as possible. (I will cross post this response to other thread.)
[Message Edited]

Seems correct to me.

I played for hours no invisible ships but some times ships disappear 30 minutes into a section. The common factor the graphics seem to disappear when I get a ship for the first time.

I kind of got used to the bug.

Ships go away to me means call it a night

and in the past it was understable stated the resources needed to fix this bug could be used in other places.

Good to hear that you are still tracking it down.

I agree with what technician says.

The time I had it with Starbases was when I had built say 4 or 5 mods to it.

It's difficult to say for me how long it is before it happens but normally within an hour or so.

Other times I have had this problem is when I've invaded a civs planet and that planet is building say an overlord and I dont have that technology, the graphic will not appear in the build screen.

the specs you have typed out Hurley it should run Xp with no problems the Bios should have nothing to do with whether XP runs or not the key factors are p 233 or higher and 96 meg of ram (yes they say it'll run in 64 but its slower then heck) your machine is more then capable....

~SDC~

Thanks David but moved on from that Machine. Also the specific bios that dell use seems to sometimes be problematic with upgrades. I appreciate that is a marketing decisions but I didn't need xp anyway.

(This is being typed on xp pro laptop machine with 512 mb, 1.3 ghz)

Hurley -- thanks for the clarification
Carielf -- thanks for your efforts

I want to point out that hurley's techniques do not consistently reproduce the problem. I grant that by performing certain random actions on his system using 3rd party software for a random period of time the probability of the bug exhibiting within a set time is extremely high.

On the other hand, taking any directx game in windowed mode while running office, IE, and other programs, running resources down in win98, and rapidly switching between them will eventually crash any system. Not too long ago most directx games could not be reliably minimized or toggled (alt-tab) so these keys were often disabled.

What we are looking for is a set of step by step instructions that will reproduce the bug exactly and predictably (like 1 time in 10 perhaps). So that the game engine can be examined before the bug happens and after.

I also concur that the bug most often occurs when a new ship is built or discovered. Using my Lightning Alpha Strike system I have seen it many times when creating death stars. I have also seen CTD's and corrupt save games on the last few turns. It seems to really strain the system to create many death stars or destroy many planets at once. Unfortunately I cannot be assured it will happen at any given time.

I am glad the openness of the discussion has allowed people to discuss the issue a little less, um, emphatically!

I am wondering if a test system can be set up to run a very minimalist version of the game -- mod range to zero and disable research. Move the ships randomly overnight, then explore an anomoly that will give a new ship. Will this produce the bug a significant oercentage of the time?

I have seen the problem occur during my own games, and I do not play on Maso difficulty or Gigantic size maps (Usually no larger than a Medium) but still encounter the problem after a period of time. Ships vanish, graphics for planets also vanish

P4 1700
384 Meg RAM
128 Meg GeForce4 Graphics Card
Win XP Pro
No other apps running

Quiting GalCiv then reloading the save game does seem to resolve, but is a pain to have to keep doing that

Hope this helps

I can reproduce the bug now by running the game in the AI-only mode, so reproducing it is no longer a problem.  I can work on my laptop while trying to reproduce it on my desktop.  It does still take a long time, but I don't have to babysit it.

After taking a look at memory after reproducing the bug yesterday, I think that it has more to do with the number of different graphics in the game as opposed to the number of instances of same graphics.  I'm running additional tests, but the last two have locked up the game.  It seems likely that my tests are the cause, so I'll have to tweak them a bit today.

Thanks Cari.



Agreed, I can't predict exactly how many turns it will take and I don't have my actions scripted. The behaviour of the other races in the game, planet distribution, anomolies etc also affect this, and I can't control that.






Maybe, however the other games don't exhibit behaviour like this bug in the period of time before a crash and the system does not crash even after playing other than galciv in this manner for 10-12 hours.

very cool -- that's what we needed!

Has anyone noticed that if you reload a game over and over the time it takes to reload gets longer and longer until it freezes, then a restart of the game will fix things completely? Seems strange to me, something is going on there that's significant (cache build up?) . . . not sure if that information is useful to you . . .

I have noticed that if I reload a game several times doing little or nothing in between I get the wierd bug there nothing will move and the "turn" stays an hourglass. Memory leak somewhere