Autosave Bug

Has anyone done any research or come to any conclusions as to why having autosave active causes the odd CTD (at the time of saving from what I've seen)? Are any mods able to run with no CTDs while having this active?

Anyone can confirm whether or not 'Quick save' is susceptible to the same bug (it seems to be)?

Thx.

Only thread I ever saw that mentioned autosave related CTD was
https://forums.totalwar.org/vb/showt...autosave+names
with problem with character names - that was more than occasional though!

I never turned it off on End of Days and never saw problem, for vII its still on and I always do manual save each turn as well with no problem.


How occasional is it? I'm just wondering if that could be behind my about one in 400 turn problem I'm blaming on script (yes I'm getting desperate)

I'm pretty confident fATW is coded well

Names have been checked a lot.

Normally a game, or an ai game, runs through to campaign end without any CTDs. But with autosave on there is a CTD every so often (last time I checked). Probably about 20 - 30 mins into an -ai fast run so quite a few turns...

The problem with autosave bug game saves is they never load (crash at end of loading bar)so it's hard to see what's wrong. the crahs always occur at the timestamp of the save.

Could it be that enum files are required for autosaves? Though we have them done for names I believe, we do not for buildings.

Maybe being a BI provinical modfodlered mod causes a problem?

Of course, it's easy enough to leave the autosave off but if Quick Save causes the problem too then that might be more difficult to avoid; and why on earth does a normal savegame cause no problems? How does a save game and a auto (and quick?) save game differ?

Whatever it is - something is diffferent between fatw and eod and it's not just the vampires...

I just took autosaves off EoDII and it did still crash on scripted version - its not 100% conclusive that non-scripted version won't go with autosave on but I've not seen it do it yet and have been doing a lot of year jump / -ai testing to try and narrow down source of problem.

Maybe being a BI provinical modfodlered mod causes a problem?

wouldn't explain any difference between FATW and EoDII. I did recently find apparent problem with a newly named faction in M2TW mod-folder not wanting to look up main folder textures properly (and erratically) - if you have new culture in that would be only equivalent I could think of that might cause problems, as I think we're both on vanilla faction names.

Could even be machine related, I could try running FATW -ai with autosave enabled and see if it behaves on mine, saves do get bigger as game advances, so there could be something about a progressively increasing load, have you tried watching CPU/memory usage while its running? I was also wondering if complexity of script might be causing me problem rather than specific bug - as having half script contents in EoDII seems fine regardless of which half it is... (symptom to my crash with script is pretty similar to the "30 mins in, occasionally" on -ai )

If it is the amount of stuff getting checked that's problem and you're not scripted you could try running without trait and anc triggers, you're probably got more on that than I have.

I haven't done changes for enums or lookups.

no idea why saves would be different between normal/auto, its easier to speculate on why it might go wrong on -ai etc as I suppose it could be trying to do the operations too fast.

Well, just ran fatw until I got bored (200 turns) on -ai with autosave set to true and no CTDs at all...so no problem there.

I've played with some quick save files sent to me and, as far as I can tell, I seem to get a ctd on the ai turn when I have fog of war on (I never have it on so I would never see any fow-related problem)....so...what would cause it to ctd with fow on, but not with it off?

I find it hard to believe it really exists, where your tests supplement my belief DG ...

I never run a game without autosave and always just use quicksave whenever I make a manual save, tested my game by running lots of -ai campaigns, only thing I havent done is actually play my game for a 100+ turns campaign ... yet I hardly ever encounted a CTD, seems that or v1.5 or my modding has improved the stability of my game (lets say its 1.5 ... :D)

Wouldnt be able to say where the problem may be, but I'm wondering if the problem still shows when you play vanilla (I cant confirm it (as in; could claim the opposite), but last time I've played vanilla iirc was 1.3)

Only CTD's I had where mostly before 1.5 and where randomly within the players turn iirc (course except for the very few random ones you could expect in about any game/program)


Edit;
Do note; I've made sure each any every file in my game is updated acording to changes, meaning even "enum" files have been modded ..
Also note that I used to play and mod Rome v1.5 and now play and mod mostly Rome under BI v1.6, where both seem the same in this ..

Edit2;
Just recalled something, something you kind of helped me solve Mak. ..
When I first made a map I could not get around a CTD when I scrolled to rebel-controlled terf. or whenever an army of brigands actually came in my view .. the same happened after [random] number of turns, if I prevented the other by staying zoomed in on my own capital, this CTD always happedned roughly around the autosave/ai-turn time, wich had something to do iirc with rebels and who owns what wich is set up in descr_regions in combination with who and what was present in my game.
Probably not the same but as I seem to recall it as it happening at the end of the players turn ... who knows .. ;)



G

Maybe it does not exist - I have forgotten the frequency of RTW vanilla bugs since I think I last played it seriously only in its first version :)

Too much time coding and none playing!

Anyway, I have most certainly had CTDs occur on autoresolve and then not load again...but this has not occurred for some time.

Much of it may well be vanilla.

And - as far as we can tell - this case turns out to be a vanilla problem which we should post up soon in the appropriate place.

Red Spot, you've come along way since your noob days ;) You're the 1 in 50 that actually become real modders. Mak's another (she used to just make the tea)

Afaik there is no problem, at least I remember from the last campaign I really played that I only ever encountered a single CTD.

Memory about it is a bit vague, but iirc it was v1.3 modded to include 3 additional Roman faction (eg; Numidia was a 2nd scipii, etc) and add these to the superfaction Senate creature (so I actually made it less stable!!) quit when I was somewhere slaughtering eastern factions, so I as Scipio had taken pretty much everything south of Italy ...

Thats ofcourse if everything is suplied in such way the game likes it ... not saying there may be an error, just something the game doesnt like eventhough its implemented correctly. Like you cant make a faction horde if its part of superfaction creature thingy, even if you have every bit of individual code correct.


ps;Thanks for the compliment, though this is what started me on serious modding (C++, 3D-Max ....), do have to note I had like a 5 year break in between .....
http://www.planetbattlezone.com


G