Getting a little bored with simply tweaking the hippies by offering them a non-descript "featherwood training sword", I decided to embark on a project to develop a slightly more significant tweak button. I'm imagining something like this:
This is a masterwork obsidian mug. It is infused with the essence of mongoose demon (beware it's insta-death bloody goo). It is encircled with bands of nether-cap (to keep your pansy strawberry juice cool). It menaces with slivers spikes of glumprong. On the object is an image of an elf in goblin-cap. The elf is screaming. On the object is an image of elves in blood thorn. The elves are dying.
I've seen the dwarven checkerboard idea and several others, but considered it too "safe" - where's the risk of one particularly speedy flier getting through the setup before Urist McLeverPuller finds his way to his station, or the risk of Urist McBoozer running off to find a drink instead of attending to the lever? So, I started setting up a semi-elaborate clown infusion obsidian farm. The setup is basically this:
- Level Z - a 20x20 maze of constructed fortifications that the clowns have to enter on one corner and exit on the opposite corner; will be filled with kittens or some such to keep them occupied for a bit to help with timing
- Level Z+1 - empty space
- Level Z+2 - an 18x18 floor suspended by a support as the first line of offense - allow the maze to fill and drop the floor first to handle the first wave
- Level Z+3 - 4 10x10 retracting bridges covered to 7/7 with magma
- Level Z+4 - empty space - the wiki and other sources seem to indicate you want to dump water from at least 2 Z-levels above the magma, otherwise you just get steam
- Level Z+5 - 4 10x10 retracting bridges covered to 7/7 with water
The bridges on Z+3 and Z+5 are all hooked to the same lever; the idea being that both the magma and the water will fall, maintaining some separation, until the magma splashes down in the maze, followed very quickly by the water, to turn the inhabitants of the maze into shiny obsidian blocks. Seemed like a sound design on paper. Water above magma, because the wiki and other sources say that's the better plan.
Anyway, decided to make a copy after I completed construction to do an un-dwarvenly test run. It was a spectacular, steam and magma-mist filled glorious..... failure. The magma and water both fell, as expected, they mixed mid-air, producing lots of steam, but absolutely no obsidian. When the steam cleared, I had a maze full of magma, with no water in sight. Which led me to play around a bit with DFHack liquids in arena mode, where I found some things that, at least to my initial searching, had not been sufficiently investigated before. A new branch of !!science!!, so to speak. So I thought I'd put my results here, in case someone else ever wonders.
In my brief searching, I did also come across a couple other ideas for obsidian drops, that seem to work, but it now seems that, it's probably a case of it working stochastically - i.e. it depends on the exact timing of whether the magma or the water hits the open space first, which will naturally vary, and give some mix of successful firings with mis-firings.
So, in the arena, I tested a) putting the water above the magma vs putting the magma above the water, b) placing the water first vs placing the magma first (assuming this would sort of simulate the timing difference when bridges are used, due to their construction and triggering order, even when attached to the same lever) and c) placing one liquid directly above the other vs leaving a Z-level in between. I'll use the notation MW/M0 to indicate "magma placed first, then water / with magma on the bottom, with zero Z-levels between". The lowest liquid was 2Z above the floor in all cases. Four separate rounds were run. The results I found were:
- For the cases where the magma was on the bottom (*/M*), no obsidian at all was formed. All water was converted to steam during the fall, leaving only magma on the floor.
- For the water on the bottom cases (*/W*), obsidian was reliably created, although not always in the shape of the dropped liquids. 3 of the */W0 cases exhibited some spread, while 5 of the */W1 cases spread out. The spreading was after the liquids hit the floor (or the top of newly formed walls), though, so is not particularly unusual.
- The cases where there was more separation between the liquids (*/W1) also exhibited some cave-ins, as obsidian formed in unstable locations. The */W0 cases did not seem to have any caveins.
- The order of placing the two liquids (MW/* vs WM/*) did not seem to have a significant influence. The additional timing factors of using bridges vs DFHack liquids might make some difference, but more testing is necessary, and takes a bit more time to set up.
So, it would seem to me that, while the general lore's recommendation to drop water onto magma instead of the other way around works just fine for the cases where you are dropping water onto magma that's already stable on a floor, when both liquids are going to be falling, it's definitely better to do things in the other order - drop the magma on top of the water. Not quite sure if these results are due to Dwarven physics, or just order of evaluation of events in the simulator (a side note - it seems, after single-stepping through this simulation, that the order in which tiles are processed for fluid-gravity interactions is randomized somewhat, because I would see some tiles fall before others, and it would be some number of ticks before an entire block of liquid had descended one level; not sure if the order of tile processing is the same on every Z-level, or if it might be different on each level - I suspect it's the same, though).