Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[RelaxAyax]

I not sure exactly what happened, but my magma piston just collapsed (the game said it was a cave in) and got completely destroyed at the last possible time. When it happened one of my miners was channeling around the piston in the level above, where the magma furnaces were supposed to go, and also there was a huge fire going on the 3rd cavern area where the piston went thru.
I dont know if this happened because of the channeling, the fire or because i had a couple of constructed walls to cover small holes in the piston, but now the only thing i have left is a huge hole connecting my fort to the caverns and a bunch of stone (i presume from the piston itself) liying around in a magma pool. Anybody has an idea of why this happened?

Thanks a lot.

[Starver]

First thoughts (not knowing much more about your chosen piston design) is that the chanelling was entirely at fault.

A fire shouldn't deconstruct anything that supports, etc, so removal of the very last true supportive-connection by either digging or directed deconstruction is the probably cause (glitches aside).

If you had something like a levered-support prepping for you ultimate piston release, holding it until after you had severed the mass from the surrounds, could that lever have already been pulled by accident/mistake/other? (Gremlins have traditionally messed this sort of thing up, in-game, and maybe you hadn't seen this when 'safely' designating the other bits of severing.)

You probably need to be sure about what the chanelling was doing (and was intended to do) with the delinking of tiles from tiles, which normally needs a good 3D model in your head. Easy to not get quite right, even if you already know things like extended bridges don't serve as a valid supporting structure when all else is removed.

[RelaxAyax]

First thoughts (not knowing much more about your chosen piston design) is that the chanelling was entirely at fault.

A fire shouldn't deconstruct anything that supports, etc, so removal of the very last true supportive-connection by either digging or directed deconstruction is the probably cause (glitches aside).

If you had something like a levered-support prepping for you ultimate piston release, holding it until after you had severed the mass from the surrounds, could that lever have already been pulled by accident/mistake/other? (Gremlins have traditionally messed this sort of thing up, in-game, and maybe you hadn't seen this when 'safely' designating the other bits of severing.)

You probably need to be sure about what the chanelling was doing (and was intended to do) with the delinking of tiles from tiles, which normally needs a good 3D model in your head. Easy to not get quite right, even if you already know things like extended bridges don't serve as a valid supporting structure when all else is removed.

Yeah i thought so myself, i think the support retractable bridge that holds the floors that connected to the piston was activated for some reason, and to add to that i did had a gremlin pop out in the caverns some time before all this happened, but it just wandered around and leave the map. What im most curious about is why did the piston just disappeared instead of falling to the ground beneath. And also the magma did come up as i planned, but because the piston was not there anymore it just falled again to the magma pool.
I'm gonna try this again, maybe channeling the stuff above first and having a different support design will help.

Nonetheless you were of great help, thank you!

[Quietust]

i think the support retractable bridge that holds the floors that connected to the piston was activated for some reason

But retractable bridges don't provide support - if that was the only thing holding your piston up, then it's no surprise that it collapsed early. As Starver stated above, you need a Support linked to a lever.

What im most curious about is why did the piston just disappeared instead of falling to the ground beneath. And also the magma did come up as i planned, but because the piston was not there anymore it just falled again to the magma pool.

Your magma piston wasn't falling down onto semi-molten rock, was it? In order for a magma piston to function properly, it needs to be dropped onto solid stone. since SMR will "swallow" any terrain that falls into it.

[RelaxAyax]

As Starver stated above, you need a Support linked to a lever.

Im so stupid, when the wiki said to add a retractable support it didn't even crossed my mind that it referred to a building literally called Support.

Your magma piston wasn't falling down onto semi-molten rock, was it? In order for a magma piston to function properly, it needs to be dropped onto solid stone. since SMR will "swallow" any terrain that falls into it.

It wasn't, i pumped the magma 1 z-level above the magma sea, but now that you mentioned it, the game tells me that there are open spaces in the floor of the magma reservoir just below where the piston was, weird because i didn't put those in, and i checked before filling it up that it was empty and in 1 piece. Maybe the piston collapsing broke the floor and it went right thru the molten rock, idk.

[Starver]

Falling landscape punches holes in both natural and constructed floors, if they have no support below, and possibly even destroy built walls (and any floortop above) but I'm afraid I'm not enough of being in the habit to cave-in to know the latter.

(I've occasionally 'built a wall off the edge of a bridge', which you can try yourself (above a drop) to observe how the game lets you build them, but the instant they are complete they will fall (assuming you have cave-ins enabled, as per default), but I think a construction falling just doesn't have the coherence of a hollowed-around landscape. There are obvious experimentations you can make to try various natural cave-ins in a prepared drop, though. And it's also possible to line a bridge[1] with walls by constructing them from the edge(s) first[2] creating a fully bridging construction itself when the final wall goes in...)

If you'd previously dug directly below your pumped-magmacistern, or it was raised up in constructed form, I could see those holes being an additional design-inevitability. But I know some people do this much more deliberately, so would be able to confirm this from experience and how they know their cave-in was arranged, and not just post-mortem after a rare unexpected error).

This is, of course, one of the !!fun!! parts of DF. Consider it a learning process, and if you can work it out yourself from the aftermath of this little fuss then you're immediately more knowledgable than myself in this aspect of the game.


[1] In mid-air, at least 1Z gap vertically between it and any lower/upper supportive structure/cavern.

[2] Remembering that construction jobs not yet assigned get taken up on First In Last Out order, if you're mass-/serially-designating. And even then, completion-time might go out of order if the job-takers have different long distances to travel to collect material and arrive to start, never mind the issue of mid-job stoppages. There is an art to job-designation of co-dependent supporting (and/or path-blocking) constructions, at the best of times, and when I 'line a bridge' (for an exotic aerial walkway, usually) I don't take chances and generally wait until the necessary side-support is at least being directly worked on before designating the next construction along.... But I digress.

[PatrikLundell]

As mentioned, falling "terrain" punch through any floors not supported by a solid construction directly underneath the falling "terrain", so your piston probably ended up swallowed by the SMR after having punched through the floor. It's a mistake that's not uncommon to fail to (remember to) check that the floor the cave-in is designed to fall to is supported beneath.

Natural stone (including cast obsidian) falling in a cave-in fuses with the rock beneath when it stops, but I don't think that happens with constructions: I think constructed walls just turn into stone/blocks/logs/..., but I won't swear on that. Being uncertain, I make sure to cast obsidian if I want to ensure it fuses.

@Starver, point [2]: You're quite a bit out of date. The construction order went out the window in 0.40.X (19?) when construction became a common labor task rather than a mason/carpenter/... task. Now dorfs take up the "closest" job.

[Starver]

@Starver, point [2]: You're quite a bit out of date. The construction order went out the window in 0.40.X (19?) when construction became a common labor task rather than a mason/carpenter/... task. Now dorfs take up the "closest" job.

You (re?)learn something every day. I've definitely been using this assumption most of the time since 40d, etc. But not as the only safeguard, so maybe it's been covered by other microtweaks and safeguards I employ. (I haven't done one of my old tower designs for a while, but I suppose my tendency to pipeline my exterior corners first, while I'm completing the floors between the walltops emerging from below, just makes the assumed parallelism work anyway, etc.) I must see if I can 'break' it in my next fort, to rid myself of that habit of working.

Cheers for the heads-up!

[Quietust]

Natural stone (including cast obsidian) falling in a cave-in fuses with the rock beneath when it stops, but I don't think that happens with constructions: I think constructed walls just turn into stone/blocks/logs/..., but I won't swear on that. Being uncertain, I make sure to cast obsidian if I want to ensure it fuses.

I'm not 100% sure how it works in the current version, but I know for a fact that constructed walls were sufficient to halt the collapse of a cave-in back in 40d - the main thing that made them different from natural stone walls is that if they fell as the result of a cave-in they would revert to item form, but as long as they stayed in the same place they were just as strong as natural stone.

[PatrikLundell]

Thanks Quietust, this matches my recollection, and it would make patching holes in a magma piston messy, as you'd have to repair the piston after each step.

[Salmeuk]

As mentioned, falling "terrain" punch through any floors not supported by a solid construction directly underneath the falling "terrain", so your piston probably ended up swallowed by the SMR after having punched through the floor. It's a mistake that's not uncommon to fail to (remember to) check that the floor the cave-in is designed to fall to is supported beneath.

Natural stone (including cast obsidian) falling in a cave-in fuses with the rock beneath when it stops, but I don't think that happens with constructions: I think constructed walls just turn into stone/blocks/logs/..., but I won't swear on that. Being uncertain, I make sure to cast obsidian if I want to ensure it fuses.

@Starver, point [2]: You're quite a bit out of date. The construction order went out the window in 0.40.X (19?) when construction became a common labor task rather than a mason/carpenter/... task. Now dorfs take up the "closest" job.

This is contrary to my direct experience when building things, gonna have to test this later. Was this specifically mentioned by Toady, or something you've noticed yourself? I swear dwarves still operate under FIFO, though your statement has certainly shaken my understanding. As you usually do Patrik lol

RE: OP, try attempting it again and double check your designations. If you were dropping right into a magma pipe that explains why you don't see any leftover plug - it fell through the magma sea and into the depths of hell. I would hate to be that demon it lands on. .*

*not actually a game mechanic

[PatrikLundell]

Regarding construction FIFO:
I don't remember if Toady mentioned the construction order change explicitly, although I would expect that he did. However, if you designate a large number of construction jobs (e.g. a wall) and look at the jobs list shortly afterwards, you'll find that the jobs marked with TSK aren't strictly the top ones, but rather spread out over the whole range of available construction jobs.

[Starver]

I woudn't know about a large number of instantaneous constructions. If I do a block of constructions[1], it's because I know it's not my concern what order things are done in. But in the same pause I will apply subsequent solo/multiple blocks to give me the most convenient order of job take-up when play commences.

And surely it's not First In, First Out, but LIFO? It'll be the weekend before I can test this (don't ask!), but I'm currently convinced that's the principle that has always worked for me.

Spoiler (click to show/hide)

If I designate a swathe of floor over a gap, immediately cancel (still in the construction-pause) the corridor of floors that leads straight through the middle and to the opposite side, then redesignate them for building afresh then I get that corridor prioritised (though restricted by access, so >1 builders would start to do non-'corridor' floors from the first block, which is toletable) and thus get the access to the opposite side for... whatever continuation of work I'm rushing towards (another similar wide and long block of floor, possibly, or the cavern wall point that I'm going to dig onward from).

If I get to the point I then add later-accessible floorways, from the 'advanced guard' end-access, I may need to de-designate not-in-process building jobs and re-apply to shuffle these back to the (possible) top of the take-up queue. Before, of course, re-adding the new-block's newer-corridor which I still intend to be the primary primary build priorty. Easier to imagine than to explain, I think.

...but I don't do quite so much open-air/mid-air mega construction these days[3]. I have, though, siezed upon Dig Priorities to extend this principle to tunnelling[2], which both removes most of the FIFO/LIFO effect and probably isn't under Common Labour Task rules anyway. (And I also guide miners, according to my own ideas of efficiency, by personalised burrows, which the Dig Priority addition makes advantageously more trivial to set up/maintain the micromaagement in Fire And Forget mode.)



I need to check all this myself afresh, though. Just for my own peace of mind. I may need to totally restructure my internally-mapped tactics, lest my current mindset fails me badly the next time it counts.


You don't need to know half of all this, of course. But I get carried away with this internal mulling over, sometimes, as all the old-hands here should be fully aware of by now. 



[1] Walls (obv. not more than two thick on the thinnest axis, assuming build access from both sides, usu. 1x10ish at most) or Floors (self-prioritising to 'internal' elements where floating over a gap).

[2] Priority 1 of narrow corridors to reach out, with lower priority rooms. Or a form of the opposite to enforce a complete room to be dug before the lower-priority 'plug' dig into the next room that reveals the next high-priority room (by this point, all the first room has been/is being dug by pre-commited miners... And unrestricted by the size and logistical restrictions (in vanilla, unhelped by DFHack extensions) that limit the equivalent practice by floors in mid-air.

[3] I used to do skyscrapers quite a bit, with alternating pinwheel floorplans that meant some practised forethought of how to use the wall-tops of the level below in some places and overhanging floors in others. Slightly simplified by the fairly recent change where the deconstruction of temporary floors (adjacent to where I needed to build floating external corners) is no longer dropping their material below but pulling it back onto the build/remove standpoint.