0009479: Growth of floating tree causes random collapses on clear cut area.

Notes
(0034457) Detros (manager) 2016-01-18 13:26	Trees seem to like to respawn in the same area you cut them out, not caring about what is or is not there (0008752, "Trees spawn on up stairs"). Can that "can't be designated for cutting" be caused by 0008700, "Trees can't be designated for cutting when other part of given tree is already designated for that" so you can't designate their accessible part for cutting because you already have some other, unaccessible, part of them designated?

(0034468) Superrob (reporter) 2016-01-19 02:52	I checked, and there was no other part of the tree designated. I even lost a miner to a tree that I couldn't designate. I even mined underneath it, and mined out the roots.

(0034476) Detros (manager) 2016-01-19 11:04	Shoot that tree with ballista then. Or bring magma. In one my fort I build a roof so that trees spawned onto/into it rather than in midair where immediately collapsed. Trees are weird.

(0034528) Superrob (reporter) 2016-01-23 11:53	SO, is that what is happening, then? They're spawning, then collapsing?

(0034535) Detros (manager) 2016-01-23 16:43	Yes, that's my guess.

(0036564) PatrikLundell (reporter) 2017-05-20 01:20 edited on: 2017-06-01 02:26	This report is the closest one I could find for the issue of trees spawning and falling, creating cave-ins, so I add my findings to this report. I dug away the level below the surface under a large area (about 40*40) with numerous trees, built a support, channeled away the sides of the surface area, cut down the trees outside the area whose branches tangled with those of trees on the inside, hooked a lever to the support, and pulled the lever. The area caved in, as intended. It was also noted that roots of trees that had been plunged to destruction still remained in the soil outside of the area, indicating the fallen trees weren't removed properly (when cutting a tree the roots magically revert to soil). Shortly after that the fortress was interrupted by reports of cave-ins in the area, as trees came crashing down. Using DFHack to investigate the issue, it was found that one difference between a standing ordinary tree and the same tree after it has been cut down is that df.global.world.plants.all[<index>].tree_info (in DFHack terms) points to a struct describing the tree before it's cut, but the pointer is nil after cutting. Quite a few trees were found within the cave-in area that did have pointers to tree descriptions. In further testing the trees in half the cave-in areas had their tree_info pointers set to nil, while half were kept. This caused cave-ins to happen only in the area where the pointers were kept. Reverting to a save before pointer manipulation and removing the pointers on the other half instead cause cave-ins to switch side. A further indication of something being amiss is that in the half of the area where the pointers were retained, the ground (one level down) showed patches of leaves, flowers, and fruit in the same way as those "contaminants" appear under the trunk of a tree if you dig out the tile beneath the trunk. Conclusion: DF does not check cave-ins for trees to be removed as a result of falling, so those trees remain and have effects (leaves, roots,...). I haven't confirmed this, but my guess is that collapses happen when trees grow. Edit: Further testing has confirmed that the tree most likely to have causing a cave-in has morphed from after the previous cave-in to the pause caused by the latest one. Twigs and a trunk appeared, and a number of tiles were marked as blocked. Edit 2: Building a floor over parts of the area where tree cave-ins were induced caused trees to gradually appear on top of the floor, including 4 trunk tile (i.e. mature) ginko trees. In addition to that, one tree appeared suspended next to the floored over area, but stopped from falling by having the branches tangled with those of a tree standing on the floor.

(0041363) MrDigger (reporter) 2023-01-01 17:32	I encountered this today while undertaking a mega project to bring sunshine down to the magma sea. I dropped a massive block (31x37x52) of intact tiles into the magma sea creating very large empty space. That large block included layers of the lowest cavern. There were plenty of uncut trees on those cavern layers that were instantly vaporized by the magma flow and every 20-30 minutes I get random collapse announcements on exactly the z-levels that the trees once stood, they don't occur anywhere else in the 52 z-level deep empty space that was created. It's fairly annoying so I'm going to attempt to drop a single 31x37 z-level of natural stone onto a support wall, muddy it, see if the trees regrow, then cut them. Long story short, do not collapse tiles containing trees because they will grow back in midair causing random cave-ins.

Issue History
Date Modified	Username	Field	Change
2016-01-16 03:24	Superrob	New Issue
2016-01-18 13:26	Detros	Note Added: 0034457
2016-01-19 02:52	Superrob	Note Added: 0034468
2016-01-19 11:04	Detros	Note Added: 0034476
2016-01-23 11:53	Superrob	Note Added: 0034528
2016-01-23 16:43	Detros	Note Added: 0034535
2017-05-20 01:20	PatrikLundell	Note Added: 0036564
2017-05-20 07:12	PatrikLundell	Note Edited: 0036564	View Revisions
2017-05-23 10:37	lethosor	Assigned To	=> lethosor
2017-05-23 10:37	lethosor	Status	new => acknowledged
2017-05-23 10:37	lethosor	Assigned To	lethosor =>
2017-05-23 10:37	lethosor	Status	acknowledged => new
2017-05-23 10:38	lethosor	Assigned To	=> lethosor
2017-05-23 10:38	lethosor	Status	new => acknowledged
2017-05-23 10:38	lethosor	Assigned To	lethosor =>
2017-05-23 10:38	lethosor	Status	acknowledged => new
2017-05-31 21:16	Loci	Assigned To	=> Loci
2017-05-31 21:16	Loci	Status	new => acknowledged
2017-05-31 21:19	Loci	Summary	Random collapses on clear cut area. => Growth of floating tree causes random collapses on clear cut area.
2017-06-01 02:26	PatrikLundell	Note Edited: 0036564	View Revisions
2017-06-08 18:40	Huntthetroll	Issue Monitored: Huntthetroll
2023-01-01 17:32	MrDigger	Note Added: 0041363

Relationships