Welcome to Keen Software House Forums! Log in or Sign up to interact with the KSH community.
  1. Hello Guest!
    Welcome to the Bug Report forum, please make sure you search for your problem before posting here. If you post a duplicate (that you post the same issue while other people have already done that before) you will be given a warning point which can eventually lead into account limitations !

    Here you can find a guide on how to post a good bug report thread.
    Space Engineers version --- Medieval Engineers version
  2. You are currently browsing our forum as a guest. Create your own forum account to access all forum functionality.

[1.171] disconnecting connector makes grid explode

Discussion in 'Bug Reports' started by mos, Jan 21, 2017.

Thread Status:
This last post in this thread was made more than 31 days old.
  1. mos Apprentice Engineer

    Messages:
    340
    For this one I'm clueless as to what's happening. I've searched the better part of today for the cause but can't find the root of it.
    So far I have narrowed it down to a connector getting disconnected. This somehow makes other blocks in the grid explode viciously as can be seen in the video at 3:23. That very connector gets connected and disconnected quite often during runtime. At 0:33 everything is fine, at 3:23 everything™ explodes.

    Usually I try to come up with a simple blueprint for reproducing but for the last 10 hours I have failed here. Sometimes it explodes right off the bat as in this video. Sometimes it runs 10+ cycles without any issues whatsoever. It just doesn't make sense to me.

    Logs: https://bei-gertrud.dyndns.org/s/i3OPUIXRddmpG9B
    Video:


    Initial theories:
    • too much weight getting pumped from one container to the next
    • locked landing gear doing strange things
    Both don't really fit as in the video, the thing is almost empty and I deliberately didn't lock the landing gear to the ground.

    Anyone else having issues where stuff magically explodes when you just plain disconnect a connector?
     
  2. MustangDood Trainee Engineer

    Messages:
    66
    It's real. It's happened to at least 3 players on my DS, and once personally to me. I was very skeptical until I experienced it myself.
     
  3. mos Apprentice Engineer

    Messages:
    340
    The thing is, how can it be reproduced constantly? I'd love to throw the testers some world or blueprint that clearly shows the error every time.
    Maybe the grids need to reach some critical mass in terms of complexity.... ?

    --- Automerge --- ---------------------------------------------------------------------------------------------------------

    Maybe this'll do for showcasing. In the video you see a completely stripped down version of the vehicle with a button panel on top to lock and unlock the connector way down in the drill hole. The connector starts off locked so you only need to unlock it. Explosion is at 0:50.

    When you unlock the connector there's two things that usually happen
    • you get kicked into the air violently
    • random blocks in the grid explode.
    Should it not explode on the first try, lock it again with the other button so you can try again.

    World save: http://steamcommunity.com/sharedfiles/filedetails/?id=847884383
    Logs: https://bei-gertrud.dyndns.org/s/FoGZwnYroeAwd9l
    Video:
     
  4. mos Apprentice Engineer

    Messages:
    340
    yup, still happening in [1.172]
     
  5. NikolasMarch Junior Engineer

    Messages:
    927
    i remember seeing earlier test builds moving something up and down with merge block and connector, now i see what its for, interesting contraption....
    does it also explode when disconnecting the connector before activating the drills?
     
  6. mos Apprentice Engineer

    Messages:
    340
    Come to think of it, I never bothered to try but yes, it indeed does explode without the drills. Entirely cutting off the rotor that holds the drills doesn't help. As soon as you unlock that connector down the hole, it explodes.

    The contraption in itself is already quite old but I try to avoid posting it in full. In theory you can grief servers by drilling endless holes and inflating the world size to crazy levels all at the cost of a lot of conveyor junctions and merge blocks. In the end you even get all those materials back because the thing goes into reverse gear, completely grinding itself down until it's safely back up.

    It has been an excellent source of :clang:™ though :woot:
     
    Last edited: Jan 30, 2017
    • Funny Funny x 1
  7. MadMax Trainee Engineer

    Messages:
    62
    Could you try the exact same setup, but remove the pistons on the landing gear and replace them with static armor blocks? I've observed problems with pistons getting hairy whenever a grid is altered.

    Specifically, disconnecting a connector (or merge block, or rotor head, etc...) tends to cause jostling in the structures that were split. Given that pistons can be very volatile in such scenarios, I can definitely see play in the main structure causing the pistons to rupture if under enough stress.

    I think it may be something worth trying as a sort of root-cause analysis. If that does stop the carnage, then try placing a rotor at the top of the piston rather than affixing the piston directly to the main structure. Rotors can act as a sort of shock absorber in these situations.
     
  8. mos Apprentice Engineer

    Messages:
    340
    As far as I remember it exploded with just one landing gear as well but I can try that tomorrow.
    There was a reason the thing was suspended on pistons but I somehow forgot what it was.
     
  9. MadMax Trainee Engineer

    Messages:
    62
    I don't think the number of landing gears is all that important. My bet is that something is triggering the piston to rupture, especially considering that they are fully extended whenever the disconnect occurs.

    I've had a piston blow up on me when the only thing I was doing was adding on to a ship that was attached to the end of it. If I added or welded a block, then it would cause the entire structure to jump around. Eventually, the piston ruptured and everything went poof!

    If I could figure out how to make those pistons rupture on demand, then I could easily make the mother of all nukes. Nothing seems to compare to the destructive potential of those things for some reason...
     
  10. mos Apprentice Engineer

    Messages:
    340
    I've been tinkering with the world save for a while just now and results are mixed but there is still some blowing up going on. Something is odd with the landing gear but it's getting late and I need more time to tinker with that. Initial observations are:
    • add one landing gear on the frame
    • lock it to the voxel
    • unlock the landing gears on the pistons
    • all landing gears are now unlocked

    • add one landing gear on the frame
    • lock it to the voxel
    • cut away the pistons, don't bother unlocking the gears
    • newly added landing gear somehow isn't locked right and whole grid moves around
    • (optional steps)
    • unlock remaining landing gear and quickly lock it again
    • upper rotor down in the drill hole now shakes violently and explodes when connector unlocks
    Right now I'm preparing a fresh world with a landing gear directly on the frame but drilling down without block damage to let the grid grow some critical mass takes time. Most likely I'll have something posted tomorrow.

    The reason why the pistons are there is [this bug] by the way. Using landing gears on pistons works around that issue.

    All in all, it shouldn't explode while suspended on pistons but simplifying the grid even more to identify the root cause sounds like the most sensible approach.
     
  11. MadMax Trainee Engineer

    Messages:
    62
    I absolutely agree that the piston should not be exploding. However; identifying instances where we can make it blow up when we do not expect it to are gold because they give us an insight into where to look for the root cause.

    I had a really weird problem last night where I built a hinge with two rotors that spun a large section of a roof from a building I am constructing. Then, two pistons would lift everything up 5m. The roof is lifted perfectly, but something crazy happens on the way down.

    If the roof is rotated back into position after the pistons were retracted, then the roof section would merge back with the building flawlessly. If; however, the rotors go back into the neutral position before the pistons were retracted, then ENTIRE structure would violently explode whenever the rotors reached their lower limits. It was like the brakes on the rotors caused the building to erupt if there was nothing for the roof section to rest on whenever the neutral position was reached. The explosion was consistent. I HAD to retract the lower pistons prior to rotating the sections back into place.

    It is things like that, as well as the issue you are having, that makes me believe that rotors and pistons both suffer from a mathematical error in the determining the forces that are applied to them under certain circumstances. Isolating the conditions that make them go boom is more beneficial than you may realize...
     
  12. mos Apprentice Engineer

    Messages:
    340
    hmm.. results with directly attached landing gear are still inconsistent at best.... this will need more experimenting and i'm sort of short on time during the week.
    --- Automerge ---
    Well, this is something.. it exploded nice and loud just now so there is some sort of consistency at least :p
    Now for narrowing things down...
    --edit--
    Something that I've noticed so far is, the landing gear seems to be in some sort of half-locked state when loading the world. It shows solid green but the grid in itself is still moving against the voxel and either the piston or the rotor doesn't like that state at all.
    Right now, bets are on some occult things going on inside the landing gear, affecting the other blocks down the line
     
    Last edited: Feb 1, 2017
  13. MadMax Trainee Engineer

    Messages:
    62
    I feel your pain, my friend. My aforementioned experiment in creating a pressurized, planet-bound hangar with a retractable roof failed miserably last night. I was using landing gears to attach to arms that extended from pistons, that where themselves attached to large a hinge I'd build between two rotors, in order to stabilize the roof as the rotors moved it from one position to the next.

    I was able to connect the arms to the connectors on the roof, lock the landing gears, unmerge the roof sections from the main structure, and rotate/lift the sections off of the hangar. Perfect!

    I was even to lower the sections back into position and completely merge them back with the hangar. The clang; however, struck whenever I tried to unlock the landing gears once the roof sections were back into position.

    The result was a spectacular explosion that completely eviscerated my hangar along two parallel lines that aligned with where the arms had been positioned. From examination of the debris, it looks like one of the landing gears blew up and caused a chain reaction to the rest of them. The damage is much more significant where one of the landing gears was, with the severity decreasing near the other gears as you move away from the apparent culprit.

    This is further evidenced in that the tower that was closest to the first landing gear was completely obliterated, while the top of the tower that was immediately across from that gear was sheared fro the base and tossed 750m away. This tower section also created a 5-10m deep impact crater near the landing site, so the forces involved had to have been pretty extreme considering it was made from light armor.

    I think what made matters worse; honestly, is that the arms I mentioned were constructed from pistons with a connector on top of it and 4 blast door blocks on top of the connector. When everything blew up, I am pretty sure that those blast doors turned into missiles that took no mercy on my creation. =(

    I'll see about getting some pictures together tonight for posterity's sake if nothing else. The pattern to the destruction is actually pretty cool once you get past the fact that it exists within a building that I'd poured nearly 50 hours in to when you look at design and build time as a single unit...
     
  14. mos Apprentice Engineer

    Messages:
    340
    That sounds pretty extreme and I feel a little bad that it made me laugh halfway during the pretty graphic description. ;)

    Maybe the landing gear is innocent in this here case. Right now the thing resides on 3 blast doors as feet without any landing gear whatsoever and it seems to constantly explode. Lets see where this trail leads
     
  15. Forcedminer Senior Engineer

    Messages:
    2,227
    tfw space engineer problems.

    I remember disconnecting landing gears and connectors and the piston drills on the front of my ship would explode.
     
  16. MadMax Trainee Engineer

    Messages:
    62
    It's OK. I had to laugh just to keep my inner Hulk at bay. My keyboard almost suffered a horrible end...

    It is interesting that things are still blowing up without a piston or a landing gear involved in your case. Is it still blowing up at the base like your prior video?

    It is starting to sound like a big bug in the collision detection code that is executed after a disconnect of some kind occurs. Given that the connector is deep underground whenever the detachment occurs, is it possible that the game sometimes thinks that part of the base should also be underground?

    If you were interested, then a test could be to try your contraption in space to see if it still blows up. It may not be able to actually drill anything, but testing in a vacuum could help to identify problems with collision detection. If it still blows up, then we know that we're getting somewhere - even if it is not in the direction of a viable workaround.
    --- Automerge ---
    Actually...

    What is directly behind that large container? I've never seen what is there, but it appears that the explosion is not random at all and is centered around that location. Your first video saw the container destroyed completely, whereas the second video shows the damage in the exact same are with the only difference being that the container survived rather than being destroyed outright.
     
  17. mos Apprentice Engineer

    Messages:
    340
    Well.. i've created a meat grinder apparently :woot:

    http://steamcommunity.com/sharedfiles/filedetails/?id=854919693

    There's a button panel on the side. Push first button to lock connector, push second button to get killed by... the force?
    No explosion though.

    Testing in Space sounds like a good idea, no voxels getting in the way so nothing to collide with. It'll start rotating due to some force inevitably but the thing in itself was already working perfectly fine in zero-G when I lifted my old mobile base into space with that thing attached :)

    ---
    There's not much around the container. one conveyor line going to the front for docking purposes and 6 of them going to the sides for stone eject. Below was one other rotor with a grinder to remove the landing gear but I think i removed that one in the world save.
     
  18. MadMax Trainee Engineer

    Messages:
    62
    Random, unexplained death? Sure sounds like a collision detection bug to me, but I can't think of what would make this happen only when using connectors...
    --- Automerge ---
    Here are some shots of things going terribly wrong...


    --- Automerge ---
    Update!

    I was able to verify that my problem was absolutely due to some bug in the collision detection code. By rearranging the sequencing in which I completed each step, I was able to isolate the problem as being one where a prong on one of the arms was making contact with either the landing gear or the station itself after the landing gear was released.

    Apparently, the landing gear was holding the hinge's rotors in place such that they were 1 degree away from the limit I had set for each. By flipping the rotors into reverse prior to unlocking the landing gear, I was able to get the arms to raise without blowing up.

    Once they were clear, I threw the hinge back into reverse so that I could reconnect everything and move on. As soon as the rotors got back to the end of the limit on the trip back down, they blew up - again....

    By setting the limits such that they match the values they were when they were being held in place by the landing gear prior to the initial release, I was able to safely drop the arms back into place. The process can now be repeated multiple times without incident. I was even able to lift and replace the roof with the rotors turning a full 1 mps. It was pretty cool to watch it moving so fast without even a slight wobble!

    The big question now is - why on EARTH would a 0.1 velocity on the rotation cause that kind of destruction?!? I wonder if there is any way that I can harness this in the form of a bomb. That thing would be incredibly light, agile, and immensely destructive. Then; maybe, devs would have to take it seriously...

    Unfortunately, the process requires me to manually convert the roof sections into ships after they un-merge from the station before the arms can move them. Otherwise, bad things happen...

    The consequence of that is I cannot automate the opening of the roof. Additionally, the thing has more leaks than I presently know what to do with given the current mechanics available to me. Oh well, that will just have to wait for a future update before I can implement, I guess...
     
  19. mos Apprentice Engineer

    Messages:
    340
    It was some pretty looking base turned trainwreck. That is some kind of nasty trench there and the outer construction completely flew away even. Actually if you can build something small to harness that destruction in the same way, maybe the devs can get rid of the clang.

    As for digging into space... I think we hit something.
    [​IMG]

    Woke up and it had dug almost a 5km hole and is still going. Good thing it has a beacon because it drifts a kilometer or two even with a set of small ion thrusters to keep it stationary.
     
  20. MadMax Trainee Engineer

    Messages:
    62
    Interesting. The collision engine is having a hard time inside of gravity wells it would seem. That sounds like a logical problem to have. My understanding is that gravity in general did not come in to play until well after the initial game engine was established.

    Maybe someone made a simple mistake in an equation somewhere? It does seem like collisions on a planet cause more damage faster than bumping around in space does, but that could just be the 12:45 AM talking at this point.

    I have not done anything with the gravity generator, but I wonder what it would take to create an artificial gravity well strong enough to get the clang to rear its head on a large asteroid? If I could figure out how to get the solution to build in VS and load a saved game, then I'd love to be able to debug your machine as it goes the way of the Challenger shuttle myself...
     
  21. mos Apprentice Engineer

    Messages:
    340
    So, from what I can see it is still behaving the same way after the patch just now.
    A also friend suggested to drill through an asteroid to rule out gravity wells. I'll slap some thrusters on the thing now

    I tried to put the car upside down this morning to have it drill into the sky but found it blew up after work. Can't say for sure what happened because I didn't see it though. Debris suggested the usual clang but I need to try again.
    --- Automerge ---
    Alrighty... drill's running on an asteroid. I'd say it explodes the same way
     
  22. MadMax Trainee Engineer

    Messages:
    62
    I'd think that the collision issue would present itself irregardless of whatever direction the drilling apparatus was moving in so long as you're in a gravity well and connected to the ground. It seems like slight contact is calculating to a much greater force than it should in these environments. I've yet to see an example of explosions like this without gravity - be it on an asteroid or otherwise.

    I'm willing to bet that there is some sort of overlap between the hit box for your station and the voxels that make up the ground. If that is the case, then the code that splits grids from one another could be detecting that but miscalculating the force generated by the collision due to some bug related to the existence of gravity in the environment.

    Thinking about it, I wonder what would happen if the drilling machine were to simply suspend above the ground while it drilled rather than do so while it is in contact with the planet. Slapping some thrusters on that bad boy and letting it hover while it drills should tell us one way or the other. If that works AND you are able to drill on an asteroid while making contact with it (as it works now,) then we will have proven our hypothesis that something is wrong with collision detection within a gravity well.

    I know that they work a lot with floating point decimals in this game. I wonder what kind of precision they are trying to maintain in those calculations. It is possible that throwing gravity into the computation while trying to maintain too tight of a precision spec could be resulting in some sort of arithmetic error.
     
  23. mos Apprentice Engineer

    Messages:
    340
    So far no explosions, this is astonishing. It's about 140m down in the asteroid but still intact. Drilling is slow because it has to wait for all the stone to eject each turn.

    This is a noteworthy change though. This round, cargo capacity is 1x instead of the default 10x. It's getting late over here so I'll set it to 3km depth and see if it survives tomorrow. Then maybe another round at 10x with a fresh start.

    Suspending it above ground might be a challenge because the inter-grid-forces that make it drift away in zero-g would also make it drift away while hovering. Drifting doesn't matter much but hitting the ground in the process kind of would :p
    It could however... would it be possible for it to navigate to stay at a waypoint in mid air via remote control autopilot constantly?
     
    Last edited: Feb 2, 2017
  24. MadMax Trainee Engineer

    Messages:
    62
    Yup, there should be at least. Just use four remote controls. Move the rig into the position that you want and assign the current GPS location of each remote control as each one's way point. Then, just turn on autopilot.

    I am a little surprised that this is a problem for you though. Throwing thrusters on there and setting their thrust override to the same value across the board does not hold it in place?
     
  25. mos Apprentice Engineer

    Messages:
    340
    Not sure about the thrusters, actually I never tried hovering a drill mid air:stare: Normally the thing should just wobble a bit because of the drills. Floating in space it rotates and moves around though, even with dampeners attached. I'm guessing it'll crash land while in the atmosphere but it might be worth a try just hovering with dampeners.
    I blame the circle formed by mergeblock - piston - rotor - connector that somehow confuses the physics engine so it applies some force to the grid where it really shouldn't.
     
  26. MadMax Trainee Engineer

    Messages:
    62
    Gotcha. Well, then the remote control approach should do just fine. If they won't consistently travel to that location, then try setting up a timer block that is constantly telling the remotes to travel to their set way point. It should keep the rig planet in place for the most part.
    --- Automerge ---
    Yah know, I was thinking about this last night and I wonder if all of the "inter-grid forces" that you mentioned are actually the problem here.

    It sounds crazy, but can you reinforce that section of the rig around the large cargo container? It looks like the only connection holding the thing together is actually the connection to the conveyor ports on the large container and (seemingly) nothing else. Running some beams from the back and towards the front to solidify the structure may just solve the problem.

    If that is the case, then this may be a "by-design" thing. If it is, then they definitely should have some sort of mechanic to indicate as much. Maybe just a different destruction animation would do the trick?
     
  27. mos Apprentice Engineer

    Messages:
    340
    I'm still running that test on the asteroid because after a while either the piston head or the attached rotor vanishes. Just now the thing kicked me away upon disconnecting the connector as happening in that world save ->
    --- Automerge ---
    16:12 - hmm... to sum up what I've seen so far

    no gravity - empty space - works fine
    no gravity - drilling asteroid - explodes
    1G earth - drilling voxel - explodes
    --- Automerge ---
    16.50 - blew a nice deep shaft with some warheads to avoid collision while drilling in a gravity well. (and partly because I was too lazy to fiddle with thrusters ;))
    Let's see if this explodes or not.

    [​IMG]

    [​IMG]
    --- Automerge ---
    17:54 - well, it reached 300m in one piece. Things by and large point to collision and not really the gravity well
    --- Automerge ---
    18:39 - it almost brings a tear to the eye... it went down to 300m and just now fully hauled itself back in without any explosions
    --- Automerge ---
    18:55 - lets see if it can collide with itself. emulating a mining shaft with blocks now.

    [​IMG]
    --- Automerge ---
    20:40 - thing went down to 300m and safely back up without explosions. If collision is the issue, it doesn't collide with itself. It did however kill the astronaut upon connector disconnect, when getting too close.
     
    Last edited: Feb 4, 2017
  28. MadMax Trainee Engineer

    Messages:
    62
    Gotcha. It would seem that collision detection between a celestial body and a ship grid is the culprit. I bet you could just throw blocks down and park your rig on them. After setting the brake, just start drilling. I'd be surprised if it blows up again.
     
  29. mos Apprentice Engineer

    Messages:
    340
    drilled a 300m hole without block damage and re-drilled into that same hole with block damage worked fine for 300m. While going back up, the piston head apparently ripped off. I'll try parking the thing onto a station block like you suggested. lets see where that blows up.
     
  30. MadMax Trainee Engineer

    Messages:
    62
    I would not assume that the piston head getting ripped off is a related phenomenon. They are notoriously finicky in their own right.
     
Thread Status:
This last post in this thread was made more than 31 days old.