3256 crash
Moderator: Oberlus
3256 crash
after loading save - just press next turn
- Attachments
-
- save.zip
- (152.37 KiB) Downloaded 43 times
- Geoff the Medio
- Programming, Design, Admin
- Posts: 13603
- Joined: Wed Oct 08, 2003 1:33 am
- Location: Munich
Re: 3256 crash
This looks to be a similar problem to several others that have popped up recently... Players are sent universes that contain all the objects they've ever known about, and the AI doesn't know the difference between stuff that used to exist but doesn't now, and stuff that exists on the current turn. As such, it thinks it has ships and fleets available that it doesn't, and tries to do stuff with them, which leads to crashes.
You can see a similar effect in 3256 by dragging your battle fleet onto your home fleet, then ending your turn. Order the home fleet to move, and there's still a fleet icon for the battle fleet, even though it doesn't show up in the fleets window.
This needs to be fixed on two levels - so that orders stuff that doesn't exist doesn't crash things, and so that the AI and UI knows what really exists.
This may make 3256 rather frustrating to test, unfortunately...
You can see a similar effect in 3256 by dragging your battle fleet onto your home fleet, then ending your turn. Order the home fleet to move, and there's still a fleet icon for the battle fleet, even though it doesn't show up in the fleets window.
This needs to be fixed on two levels - so that orders stuff that doesn't exist doesn't crash things, and so that the AI and UI knows what really exists.
This may make 3256 rather frustrating to test, unfortunately...
Re: 3256 crash
here's my 10th turn:
there are strange ship-sign at Spica.
and some starlanes remains visible (marked ~)
while some other are not (marked - - -)
save attached
there are strange ship-sign at Spica.
and some starlanes remains visible (marked ~)
while some other are not (marked - - -)
save attached
- Attachments
-
- AS_Human_0010.zip
- (128.81 KiB) Downloaded 40 times
- Geoff the Medio
- Programming, Design, Admin
- Posts: 13603
- Joined: Wed Oct 08, 2003 1:33 am
- Location: Munich
Re: 3256 crash
There's probably not much point in testing visibility related issues in 3256 as it's in the midst of an incomplete transition, and the UI and AIs partly can't and don't know how to deal with previously-but-now-not visible stuff properly...
Re: 3256 crash
so, than tell me what to focus on? maybe there is some sort of "changes list"? because game itself almost unplayable for now
Re: 3256 crash
http://cia.vc/stats/project/FreeOrion/mZhura wrote: maybe there is some sort of "changes list"?
Geoff's point was not to provide a playable game, but to keep us updated with all the recent changes.
- Geoff the Medio
- Programming, Design, Admin
- Posts: 13603
- Joined: Wed Oct 08, 2003 1:33 am
- Location: Munich
Re: 3256 crash
I don't know if there's much you can test in 3256 that involves actually trying to play the game. One benefit is that 3256 should be able to read Bigjoe5's latest set of content files that use the new CreateShip effect, so you can look at the stuff he's added or modified.mZhura wrote:so, than tell me what to focus on? [...] because game itself almost unplayable for now
You can also try playing without any AIs, which should eliminate the crashes they cause, and just work around the phantom fleet icons appearing around the map from your own previously-existing fleets or ships.
Even if it's not much use to further test with, sending this version has already been a useful, in that it revealed the scope of the current problems were a bit larger than I'd noticed and already attempted to work around.
I should probably make a temporary version that uses the old no-memory visibility system, though, which is a relatively simple code change (undoing revision 3247 should work, I think) although is a bit awkward for me to do because I'm in the midst of a bunch of other rewrites that will support the empire memory stuff better. I'll just have to check out a separate copy and build that with the necessary changes.
Re: 3256 crash
understoodGeoff the Medio wrote:I don't know if there's much you can test in 3256 that involves actually trying to play the game. One benefit is that 3256 should be able to read Bigjoe5's latest set of content files that use the new CreateShip effect, so you can look at the stuff he's added or modified.mZhura wrote:so, than tell me what to focus on? [...] because game itself almost unplayable for now
"my 10th turn" in 3rd post here already have no AI'sGeoff the Medio wrote:You can also try playing without any AIs, which should eliminate the crashes they cause, and just work around the phantom fleet icons appearing around the map from your own previously-existing fleets or ships.
you're welcomeGeoff the Medio wrote:Even if it's not much use to further test with, sending this version has already been a useful, in that it revealed the scope of the current problems were a bit larger than I'd noticed and already attempted to work around.
but that's a regress in development? you don't actually have to worry about my complain of unplayability. AI's for now didn't add much fun so playing without is a good solution. if you want i'll continue with 3256.Geoff the Medio wrote:I should probably make a temporary version that uses the old no-memory visibility system, though, which is a relatively simple code change (undoing revision 3247 should work, I think) although is a bit awkward for me to do because I'm in the midst of a bunch of other rewrites that will support the empire memory stuff better. I'll just have to check out a separate copy and build that with the necessary changes.
- Geoff the Medio
- Programming, Design, Admin
- Posts: 13603
- Joined: Wed Oct 08, 2003 1:33 am
- Location: Munich
Re: 3256 crash
Not really regressing... It's easy enough to switch between the two, and most of the meat of the changes are still there, but just aren't used noticably in 3258, which has the modification I mentioned. The empire-memory-enabled version will be what I use to make further modifications... (and actually, I'll add a compile-time switch to make it simpler to alternate between the two in future, like the already-present switches to make techs or ships cheap and fast to research or build.)mZhura wrote:but that's a regress in development? you don't actually have to worry about my complain of unplayability. AI's for now didn't add much fun so playing without is a good solution. if you want i'll continue with 3256.
Use the 3258 I sent a few minutes ago.
Re: 3256 crash
maybe it's too late but there's crash of 3257:
freeorionca.log not found ai_3.log attached
and by the way: without AI's works well. but there's one thing marked (~) - you see it's my 2nd turn so i cannot know anything about those starlanes - i never visited nearby systems.
freeorionca.log not found ai_3.log attached
and by the way: without AI's works well. but there's one thing marked (~) - you see it's my 2nd turn so i cannot know anything about those starlanes - i never visited nearby systems.
- Attachments
-
- AI_3.log
- (9.05 KiB) Downloaded 42 times
Re: 3256 crash
3258 looks exactly the same only now with AI_2
- Attachments
-
- AI_2.log
- (9.05 KiB) Downloaded 41 times
- Geoff the Medio
- Programming, Design, Admin
- Posts: 13603
- Joined: Wed Oct 08, 2003 1:33 am
- Location: Munich
Re: 3256 crash
Please include saves and all the log files...
Edit: don't bother, I can reproduce it myself. The starlanes you can see are there because I forgot that the visibility changes I reverted were a bit more complicated than I thought...
Edit: don't bother, I can reproduce it myself. The starlanes you can see are there because I forgot that the visibility changes I reverted were a bit more complicated than I thought...
Re: 3256 crash
please make some on-off switch somewhere in options or in newgame settings for this damned "unknown" thing i've already had a headache when counting 10-planet-production/minerals/food-balance. i really have no need for any additional difficulties
Re: 3256 crash
i have a lot of doubt about shown path is fastest
Re: 3256 crash
i guess i found an answer. starlane marked (~) while showing by the UI is still invisible to the AI. somehow