View Issue Details

IDProjectCategoryView StatusLast Update
0001791Dwarf FortressMap Featurespublic2015-12-04 21:51
ReporterKefka Assigned To 
PrioritynormalSeveritymajorReproducibilityhave not tried
Status newResolutionopen 
PlatformUnknownOSUnknownOS VersionUnknown
Product Version0.31.03 
Summary0001791: Magma sea breaching into Hell
DescriptionSometimes in the divisions between the local map quad's there's such a steep cliff that the hell layer moves so upwards it breaches the magma sea from the side. Obvious effects include lag due to absurd flow simulation, magma pipes drying out and demonic invasion.

Here are two threads on the forums relating the same bug:

(Hence why I put the "unknown" tag on the platform/OS. I don't know what those people used)

Given how obscene cliff sites are popular as fort locations, this bug jeopardizes many fortress construction efforts.
Steps To ReproduceEmbark on a site with very very very extreme cliffs.
Dig down to hell or reveal the map.
Additional InformationMy own theory about this :P

This happened because DF generetes the terrain on each one of its 48x48 quads more or less independently, and each of the 48x48 quads seems to have some "height" measure of its own.
Whenever you are in a map with global worldgen terrain "steps" like this one, the quad's internal height values don't match properly, and you have those maps where there are one or two Z-levels of discrepancy between the magma sea on the "high" and "low" tiles (specially common on mountainsides). However since the magma sea is thick, and so is the SMR layer below it, we don't have many issues (other than the odd dwarf digging into the magma sea from the SIDE, yes it can happen).
In this case because the fort sits on an obscene cliff, the "fault" between the quads is so great that the magma sea's side breached into hell, and voila, you have this.. There's a perfect match between where the giant cliff starts and where the magma sea breaches into hell.

Some screenshots in stonesense (save is also available somewhere in that thread)
TagsNo tags attached.


related to 0000051 resolveduser6 Cavern collapse at embark 
related to 0003310 resolveduser6 areas of semimolten rock and magma sea intersect with higher geological features in a very rectilinear manner. 
related to 0000127 new Muddy open space in magma pipe 
parent of 0002411 resolveduser6 Volcano that pierces water cavern layer will drain into magma sea plus cave-ins. 
related to 0008190 new Embark Location has 2000+ Z Levels. Crashes on initial unpause 


Urist Imiknorris

2010-05-13 05:50

reporter   ~0006636

It's official. This is the best bug this game has ever had.


2010-10-31 18:37

reporter   ~0013573

I believe I have another instance of this bug. Tapped into the magma sea at a weird place and unleashed a hoard of demons. However, they don't appear to be able to path through the magma to reach my fortress.


2010-12-12 00:14

reporter   ~0014549

Sometimes a Deep pit or a cave will do this too. Had this problem on my current civs won't pickup non military equipment bug fort. Used DFhack dfliquids to seal the leaks and boosted my fps a little, although not as much as I'd hoped.


2011-03-27 09:08

reporter   ~0016674

Last edited: 2011-03-27 09:18

Actually, i managed to gen a map, where hell not only reaches the magma sea, but goes straight through it on two grid cells, almost reaching the cavern layer below.

Normal hell starts at Z-68, except for two cells, that start at Z-45. Suffice to say, i was able to dig straight down into it without going past any obstacles.

The two adjacent cells in question are situated under a 'peak' of sorts, so the relative difference in height between hell and the surface is about the same as for the rest of the map, except that in this case, that relative difference places things a bit higher than usual.

Also besides the two cells mentioned above, there are 5 others that behave the opposite way. In them, the rock layers go deeper than they should, skipping both the magma sea and the slade layers and ending straight in hell.

image of Z-67. Z-68 is universally hell in all grid cells. the two hell cells on the right stretch up to Z-45.

The embark area has some 10+ 20+ cliffs, thanks to a volcano.



2012-03-06 08:58


The OP's original save did not appear to exhibit save corruption.


2012-03-06 10:35

reporter   ~0021178

This should be easier to replicate at will be lowering the number of layers. In theory, it should be possible to have hell breach the surface by removing all cavern layers and then embarking on an extreme cliff.


2015-12-02 01:23

reporter   ~0033248

I believe I also have this bug. My entire map dropped two z levels and crashed onto whatever was below.

I located a backup and replayed, but the bug did not duplicate, so I am not including a save file on DFFD.


2015-12-02 11:04

reporter   ~0033282

That may not be the same bug. The "entire map" should never crash, as it is supported from the side. Therefore, no amount of cliffs should cause a cavein.

Add Note


Issue History

Date Modified Username Field Change
2010-05-06 18:39 Kefka New Issue
2010-05-06 18:41 user6 Relationship added related to 0000098
2010-05-06 18:41 user6 Category Geology => Map Features
2010-05-13 05:50 Urist Imiknorris Note Added: 0006636
2010-07-14 11:35 user6 Relationship added related to 0000051
2010-07-14 11:35 user6 Relationship deleted related to 0000098
2010-07-14 11:39 user6 Relationship added parent of 0002411
2010-10-31 18:37 Karlito Note Added: 0013573
2010-12-05 15:41 user6 Relationship added related to 0002713
2010-12-12 00:14 Videospirit Note Added: 0014549
2011-03-27 09:08 csebal Note Added: 0016674
2011-03-27 09:09 csebal Note Edited: 0016674
2011-03-27 09:10 csebal Note Edited: 0016674
2011-03-27 09:18 csebal Note Edited: 0016674
2012-03-06 08:58 user6 Note Added: 0021172
2012-03-06 08:58 user6 Relationship added related to 0003310
2012-03-06 08:59 user6 Relationship deleted related to 0002713
2012-03-06 08:59 user6 Relationship added related to 0000127
2012-03-06 10:35 Granite26 Note Added: 0021178
2012-03-07 20:27 user11 Sticky Issue No => Yes
2014-08-31 09:53 lethosor Relationship added related to 0008190
2015-12-02 01:23 romeofalling Note Added: 0033248
2015-12-02 11:04 jwoodward48df Note Added: 0033282