Dwarf Fortress Bug Tracker - Dwarf Fortress
View Issue Details
0002230Dwarf FortressWorld Generation -- Generalpublic2010-06-09 14:302010-06-12 17:08
ChickenLips 
Footkerchief 
normalmajorsometimes
resolvedno change required 
AMDWindowsXP
0.31.06 
 
0002230: Persistent crash during WorldGen with certain seed
I'm getting a repeatable crash during WorldGen when using a certain random number seed along with my settings, which I'll include in comment below. It occurs toward the latter end of history generation, although I didn't catch the exact year. I've run WorldGen on these settings with other seeds, without a hitch. It's just this one, and it has crashed each of the two times I've tried it (seemingly at the same point, but I'm not sure).

AppName: dwarf fortress.exe AppVer: 0.0.0.0 ModName: msvcr90.dll
ModVer: 9.0.30729.4148 Offset: 0006ccb5
1) Use the WorldGen settings in my comment below with seed 28734.
2) Create and let it run (seems to take about 5 or 10 mins to get to the crashing point on my box).
No tags attached.
Issue History
2010-06-09 14:30ChickenLipsNew Issue
2010-06-09 14:31ChickenLipsNote Added: 0008008
2010-06-09 14:32ChickenLipsNote Edited: 0008008bug_revision_view_page.php?bugnote_id=0008008#r3011
2010-06-09 14:33ChickenLipsNote Edited: 0008008bug_revision_view_page.php?bugnote_id=0008008#r3012
2010-06-09 14:34ChickenLipsNote Edited: 0008008bug_revision_view_page.php?bugnote_id=0008008#r3013
2010-06-09 15:50FootkerchiefNote Added: 0008013
2010-06-09 16:23ChickenLipsNote Added: 0008015
2010-06-09 21:08ChickenLipsNote Added: 0008024
2010-06-09 21:09ChickenLipsNote Edited: 0008008bug_revision_view_page.php?bugnote_id=0008008#r3016
2010-06-10 10:32Knight OtuNote Added: 0008059
2010-06-10 11:00ChickenLipsNote Added: 0008062
2010-06-10 11:02ChickenLipsNote Edited: 0008024bug_revision_view_page.php?bugnote_id=0008024#r3028
2010-06-10 20:33ChickenLipsNote Added: 0008101
2010-06-10 20:49FootkerchiefReproducibilityalways => sometimes
2010-06-10 20:55ChickenLipsNote Edited: 0008101bug_revision_view_page.php?bugnote_id=0008101#r3045
2010-06-10 21:16ChickenLipsNote Edited: 0008101bug_revision_view_page.php?bugnote_id=0008101#r3048
2010-06-10 21:37ChickenLipsNote Edited: 0008101bug_revision_view_page.php?bugnote_id=0008101#r3049
2010-06-10 22:12ChickenLipsNote Edited: 0008101bug_revision_view_page.php?bugnote_id=0008101#r3050
2010-06-11 10:34ChickenLipsNote Added: 0008149
2010-06-11 11:05ChickenLipsNote Edited: 0008149bug_revision_view_page.php?bugnote_id=0008149#r3071
2010-06-12 15:41ChickenLipsNote Added: 0008220
2010-06-12 15:49FootkerchiefNote Added: 0008221
2010-06-12 15:50FootkerchiefNote Edited: 0008221bug_revision_view_page.php?bugnote_id=0008221#r3109
2010-06-12 15:51FootkerchiefIssue Monitored: Footkerchief
2010-06-12 15:51FootkerchiefNote Edited: 0008221bug_revision_view_page.php?bugnote_id=0008221#r3110
2010-06-12 15:52FootkerchiefNote Edited: 0008008bug_revision_view_page.php?bugnote_id=0008008#r3111
2010-06-12 15:59ChickenLipsNote Added: 0008223
2010-06-12 17:08FootkerchiefNote Added: 0008231
2010-06-12 17:08FootkerchiefStatusnew => resolved
2010-06-12 17:08FootkerchiefResolutionopen => no change required
2010-06-12 17:08FootkerchiefAssigned To => Footkerchief

Notes
(0008008)
ChickenLips   
2010-06-09 14:31   
(edited on: 2010-06-12 15:52)
My WorldGen settings:

(The main things I've changed relate to civs and caves.)

[EDITED: Changed oops in seed below.]


http://pastebin.com/QSzuK9qD [^]

In case it matters, I also have these in the d_init:

[SHOW_EMBARK_TUNNEL:ALWAYS]
[SHOW_EMBARK_RIVER:ALWAYS]
[SHOW_EMBARK_POOL:ALWAYS]
[SHOW_EMBARK_M_PIPE:ALWAYS]
[SHOW_EMBARK_M_POOL:ALWAYS]
[SHOW_EMBARK_CHASM:ALWAYS]
[SHOW_EMBARK_PIT:ALWAYS]
[SHOW_EMBARK_OTHER:ALWAYS]

(0008013)
Footkerchief   
2010-06-09 15:50   
Are you using any mods? If so, you'll need to upload a zip of those as well.
(0008015)
ChickenLips   
2010-06-09 16:23   
Nope, no mods.
(0008024)
ChickenLips   
2010-06-09 21:08   
(edited on: 2010-06-10 11:02)
Oh, damn, I forgot to change the first seed back to the one I used in the file above. Should be XXXXX, not XXXXX; editing the original to reflect that.

EDIT: ARGH! 28734, not 28735.

(0008059)
Knight Otu   
2010-06-10 10:32   
I haven't been able to reproduce this crash so far in a clean 0.31.06 install. For comparison, the parameters above generate Nir Lathon, The Land of Myth. The last site is zooms to is the cave Storlutamas Um Ûbom, in the Esteemed Forests.

I do notice that you say that you have a lot of stuff in your d_init.txt that shouldn't be there (everything but tunnels), but I don't see why that would cause a crash on a limited number of seeds. I'll retry with those init options.
(0008062)
ChickenLips   
2010-06-10 11:00   
Yeah, I was trying those out -- I imagine they're no longer used. I'm running it again to see if I get it again, and I'll try other nearby seeds in case I goofed and it's actually 28733 or 28735 or something. I wasn't being as rigorous about tracking it when I ran this one, so it's rather possible I goofed it. I'll pipe up when I've confirmed it.
(0008101)
ChickenLips   
2010-06-10 20:33   
(edited on: 2010-06-10 22:12)
[EDITED MORE: Tested quite a bit more and have changed report below to reflect it.]

Okay, testing it further, I find that I lied twice. However, it seems the neglected issues are fairly immaterial (see below).

First, that reproducibility field should be changed to a very strong "sometimes," as it's not completely consistent, but successful completions do seem to remain the exception. Out of 13 test runs, it crashed 9 times -- an approximate 66% failure rate.

Log of test runs made *after* initial report:

Attempt 1: Crashed while placing civilization 35.
Attempt 2: Crashed during history generation at year 363.
Attempt 3: -- Completed successfully.
Attempt 4: -- Completed successfully.
Attempt 5: Crashed during history generation at year 100-something (couldn't catch exactly; was obscured by another window at time of crash).
Attempt 6: -- Completed successfully.
Attempt 7: Crashed during history generation at a year I couldn't catch at all.
Attempt 8: Crashed at placing civilization 24.
Attempt 9: Crashed during history generation at year 121.
Attempt 10: Crashed at placing civilization (number was obscured by another window).
Attempt 11: -- Completed successfully. (Again, was not running a second instance of DF while I ran this WorldGen, so that may be influencing it.)
Attempt 12: Crashed during history generation at year 108. (Before I ran this one I removed the extraneous d_init finder entries to isolate them from the equation, and the crash still occurred.)
Attempt 13: Crashed at placing civilization 9. (Mineral raw files were restored to unchanged originals before running this attempt.)

Batting rate so far: Out of 13 attempts, 9 crashed, 4 completed without a hitch. The only real difference in situations I can think of is that, most of the time, I was doing something else (usually playing around with another session of DF); during the times it didn't crash, I wasn't doing much on the computer aside from reading, and hence only one processor was being significantly utilized.

Second, I just remembered that, while I don't have any mods proper, I did change the mineral and gem frequency from 100 to 20 in the raws (inorganic_stone_gem and inorganic_stone_mineral) -- basically find/replaced VEIN:100 -> VEIN:20 and CLUSTER_SMALL:100 -> CLUSTER_SMALL:20. However, after Attempt 12 I changed them back to normal, and the crash still recurred, so that seems to eliminate them as a likely cause.

(0008149)
ChickenLips   
2010-06-11 10:34   
(edited on: 2010-06-11 11:05)
Alright, as can be seen by my repeatedly-edited comment above, I've run some more serious testing on it, and come up with results that seem to indicate a definite problem, and one that seems to be independent of the d_init entries and changed mineral frequency raws that I had during the earlier tests.

*However*, I've also set up a bone-stock installation of the game (with minor changes to the init -- no intro, no sound) and tried those WorldGen settings, and it does *not* fail. No CTDs in 12 consecutive run-throughs.

Meanwhile, I'm now noticing intermittent CTDs with *another* seed in WorldGens run from my play installation (the WorldGen settings for which I've only changed to set the caves to 50 and 75 instead of 150 and 225), which is strange, because I haven't changed the settings since I installed it, and I wasn't running into CTDs for a lot of other WorldGens.

So, it would seem there's something in my play installation causing it, and that whatever it is doesn't occur in all WorldGens, but some.

Given that I've removed the variations in mineral frequency and the erroneous SHOW_ entries in my d_init, and that the only other difference, aside from installation location, is the inits (again, I play unmodded), it seems something in them is causing/triggering the CTDs.

So, for now I'm adding each change in the inits one at a time, then running five WorldGens with the same WorldGen settings and seed discussed here, and I'll pipe up if I run across a setting that triggers it.


EDIT: Here are the changed settings in my play installation's inits. Any advice on which are more likely than others to be problematic would be helpful. Aside from that, I'll be slowly going through them.

init

[FONT:curses_800x600.png]
[BLACK_SPACE:NO]

[COMPRESSED_SAVES:NO]


d_init

[AUTOSAVE:SEASONAL]
[AUTOBACKUP:YES]
[AUTOSAVE_PAUSE:YES]


[PAUSE_ON_LOAD:YES]
[EMBARK_WARNING_ALWAYS:YES]
[SHOW_EMBARK_TUNNEL:ALWAYS]

[ECONOMY:NO]
[ZERO_RENT:YES]


[LOG_MAP_REJECTS:YES]
[EMBARK_RECTANGLE:5:5]

[BABY_CHILD_CAP:12:8]

(0008220)
ChickenLips   
2010-06-12 15:41   
Alright. I'm not seeing the crashes replicate in my test installation, with all those settings active. Could it be caused by running two instances of the game from the same executable? (One playing, one running WorldGen.)

Still looking for other differences.
(0008221)
Footkerchief   
2010-06-12 15:49   
(edited on: 2010-06-12 15:51)
Alright. I'm not seeing the crashes replicate in my test installation, with all those settings active. Could it be caused by running two instances of the game from the same executable? (One playing, one running WorldGen.)

Um, yeah, that could probably do it! Particularly if the "playing" save is saving/loading during the world generation process.

The fact that DF allows multiple instances was actually a bug under the old system:

# 000020 □ [technical] (http://www.bay12forums.com/smf/index.php?topic=21812.0 [^]) stop multiple copies of the program from being open

(0008223)
ChickenLips   
2010-06-12 15:59   
Aaaggh! Alright. Go ahead and close this one. :facepalm:
(0008231)
Footkerchief   
2010-06-12 17:08   
No problem, glad it's figured out. If it shows up again without multiple instances, feel free to reopen this.