raptop changed the topic of #principia to: READ THE FAQ: http://goo.gl/gMZF9H; The current version is Galileo. We currently target 1.5.1, 1.6.1, 1.7.x, 1.8.1, and 1.9.1. <scott_manley> anyone that doubts the wisdom of retrograde bop needs to get the hell out | https://xkcd.com/323/ | <egg> calculating the influence of lamont on Pluto is a bit silly… | <egg> also 4e16 m * 2^-52 is uncomfortably large
egg|anbo|egg has quit [Remote host closed the connection]
egg|cell|egg has joined #principia
egg|anbo|egg has joined #principia
egg|cell|egg has quit [Ping timeout: 189 seconds]
egg|cell|egg has joined #principia
egg|cell|egg has quit [Ping timeout: 202 seconds]
egg|cell|egg has joined #principia
egg|cell|egg has quit [Ping timeout: 378 seconds]
Wetmelon has joined #principia
egg|cell|egg has joined #principia
egg|cell|egg has quit [Ping timeout: 378 seconds]
egg|cell|egg has joined #principia
egg|anbo|egg has quit [Ping timeout: 202 seconds]
<discord->
itsRyan. — I'm not sure if this has been reported already, but I was seeing multiple reports in rp-1 channels of oddness around spin stabilized upper stages, especially with regard to early orbital craft, so I reproduced the issue here. This is with Galileo, but it behaves identically in Fuches. This video first shows with Principia enabled, then a reload of the same situation with hack gravity where you
<discord->
itsRyan. — I'm not sure if this has been reported already, but I was seeing multiple reports in rp-1 channels of oddness around spin stabilized upper stages, especially with regard to early orbital craft, so I reproduced the issue here. This is with Galileo, but it behaves identically in Fuches. This video first shows with Principia enabled, then a reload of the same situation with hack gravity where you
<discord->
itsRyan. — I'm not sure if this has been reported already, but I was seeing multiple reports in rp-1 channels of oddness around spin stabilized upper stages, especially with regard to early orbital craft, so I reproduced the issue here. This is with Galileo, but it behaves identically in Fuches. This video first shows with Principia enabled, then a reload of the same situation with hack gravity where you
<discord->
itsRyan. — OK, I'll see if I can track down a) where the initial off-axis momentum delta is coming from as it is present in both cases. b) If there's a workaround on the part side.
<discord->
egg. — the off-axis momentum is very visible before decoupling
egg|anbo|egg has quit [Read error: Connection reset by peer]
<discord->
itsRyan. — Yeah, this may just be the case that we were trained in the pre-Frobenius world to be too sloppy with initial conditions.
<discord->
egg. — I think most of the mass is in the big cone behind you
<discord->
egg. — which means the decoupler is quite far from the centre of mass, and the mild off-axis movement is not so mild on the jettisoned needle
<discord->
itsRyan. — That makes sense.
<discord->
egg. — Note that with gravity hacked, you definitely have something fishy going on, because there is a marked angle between the cone and the needle
<discord->
itsRyan. — OK, thanks, so it's an issue, but not with Principia. So folks building Juno 1 style orbiters are just going to have to get a lot more precise with their spin axis to make it work. (or spin a lot faster).
<discord->
egg. — or putting more mass into the orbiter
<discord->
egg. — or put more mass into the orbiter (edited)
<discord->
itsRyan. — Tsiolkovsky would take issue with that solution 😀
<discord->
egg. — yeah, it is not the most practical one :-p
<discord->
DanieI. — how does principia manage to handle so many nbody's without intense lag
<discord->
DanieI. — anything I make cant handle more than 20
<discord->
egg. — I am going to need some context to make sense of that question
<discord->
egg. — also I am going to need some sleep, since it is 6:30
egg|anbo|egg has joined #principia
<discord->
DanieI. — so ive tried making nbody simulations
<discord->
DanieI. — really simple ones with point masses and 2d
<discord->
DanieI. — they all run at like 3 fps when more than 20 objects
<discord->
DanieI. — they all run at like 3 fps when more than 20 objects are being simulated (edited)
<discord->
egg. — what are frames in this context
<discord->
DanieI. — frames per second, with a time step of about 1/60th of a second
<discord->
egg. — so three time steps per second?
<discord->
egg. — how did you write the simulation
<discord->
DanieI. — I used a runge kutta integrator to calculate forces
<discord->
DanieI. — in c#, in Unity
<discord->
DanieI. — I disabled pretty much everything except rendering and my code
<discord->
DanieI. — i even tried implementing an octree to partition the space but that just made it more laggy
<discord->
egg. — The RK integrator does not calculate forces, it integrates them; it does its thing once per time step and that is not a lot of work
<discord->
egg. — the force computation itself is what should be the bulk of the computational cost, and if that takes a third of a second for 190 interactions something is extremely wrong
<discord->
DanieI. — oh wait its made in python and pygame
<discord->
DanieI. — i have 2
<discord->
DanieI. — i think I actually came to you with a question similar to this several years ago in irc lol
<discord->
egg. — Python or C# is not really the question, be it written in COBOL or in lisp this is orders of magnitude too slow to be sensible
<discord->
egg. — the question is what you do
<discord->
egg. — "used a runge kutta integrator to calculate forces" is not a useful description when the question at hand is whether you wrote an integrator and a force computation sanely
<discord->
egg. — Python or C# is not really the question, be it written in COBOL or in Lisp this is orders of magnitude too slow to be sensible (edited)
<discord->
DanieI. — how do you deal with super high levels of timewarp?
<discord->
Falcon. — Hey Egg, thanks for that patch, everything is functioning perfectly atm
<discord->
egg. — @DanieI there is nothing magical to do, doing it in a straightforward way is just very fast. With the numbers you give, the problem you have is not that you are missing some clever trick, it is that you are doing something so extraordinarily wrong that it runs one thousand times too slowly
<discord->
egg. — if you managed to do that independently twice, the issue is likely that you fundamentally misunderstand what you are doing
<discord->
DanieI. — the unity one runs a lot better 😅
<discord->
egg. — which is why I was trying to ask you to explain what you were doing with that integrator
<discord->
egg. — but anyway, it is now seven in the morning, I should be sleeping
<discord->
DanieI. — there is a lot of glaring issues in other parts of the code
<discord->
DanieI. — i try to render a line with over 6000 segments every frame for every object
<discord->
DanieI. — as the history
<discord->
lpg. — with build 2626, got a crash twice in a row while working on a flight plan:
<discord->
lpg. — `F0712 00:59:31.438285 14304 discrete_trajectory_body.hpp:644] Check failed: t_min() <= time (-1.28353034162678909e+09 s vs. -nan(ind) s)`
<discord->
lpg. — will provide full report tomorrow
<discord->
egg. — @lpg if you can reproduce it, make a journal, that speeds debugging quite a lot
<discord->
egg. — 💤
egg|anbo|egg has quit [Remote host closed the connection]
Blu3wolf has joined #principia
Iskierka has quit [Ping timeout: 204 seconds]
Wetmelon has quit [Ping timeout: 378 seconds]
Iskierka has joined #principia
<discord->
Butcher. — > i try to render a line with over 6000 segments every frame for every object
<discord->
Butcher. — @DanieI a 6000 vertex line should not be a problem on modern hardware.
egg|cell|egg has quit [Ping timeout: 198 seconds]
<_whitenotifier-d13c>
[Principia] LUH5147 commented on issue #2634: Near FTL capability!? - https://git.io/JJOfe
<_whitenotifier-d13c>
[Principia] pleroy commented on issue #2634: Near FTL capability!? - https://git.io/JJOfz
<_whitenotifier-d13c>
[Principia] pleroy opened pull request #2638: Mention proper library versions in the loader error message - https://git.io/JJOfH
<discord->
Stonesmile. — but with the build that was recommended to test the issue is appearing on every other time warp, and correcting itself on the next
<_whitenotifier-d13c>
[Principia] clc333 edited a comment on issue #2643: Game Crashes When Launching New Game (RO-RP-0) - https://git.io/JJOt2
<discord->
Stonesmile. — but with the build that was recommended to test, the issue is appearing on every other time warp, and correcting itself on the next (edited)
<discord->
egg. — 1. have you checked that it occurs only with Principia;
<discord->
egg. — 2. if so, which version introduces the issue
<discord->
Stonesmile. — I'm still not sure this is Principias fault, as the issue only appears when a ROCapsule is root part
<discord->
Stonesmile. — The issue only appears using principia, but that could be due to something with the part
<discord->
egg. — well that is an interesting data point ; the part might have some strange property
<discord->
egg. — so now, 2.
<discord->
egg. — when does the bug appear
<discord->
Stonesmile. — My best idea is something with CoM offset, as the capsules might do something with that for stability
<discord->
DRVeyl. — All ROCapsules are not alike. They're not a system like ROTanks, they're copies of parts from other mods. So probably need to be specific about which one.
<discord->
DRVeyl. — All ROCapsules are not alike. They're not a system like ROTanks, they're copies of parts from other mods. So need to be specific about which one[s]. (edited)
<discord->
Stonesmile. — This happens with both Mercury and Mk 2 @DRVeyl
<discord->
Stonesmile. — I can test the other capsules as well
<discord->
DRVeyl. — Follow egg's RFIs. He knows what to look for. I'm just clarifying. 🙂
<discord->
Stonesmile. — The issue is re-introduced in 2602
<discord->
egg. — OK so it is not the CoMoffsets
<discord->
egg. — @Stonesmile you are using a new ship each time?
<discord->
Stonesmile. — Yes, it's a new vessel
<discord->
egg. — if you load a Galileo ship with a pre-galileo version and vice-versa you could conceivably get some CoMoffset-related nonsense
<discord->
Stonesmile. — I can try to assemple a new vessel file in the VAB if that is what you mean
<discord->
Stonesmile. — I can try to assemble a new vessel file in the VAB if that is what you mean (edited)
<discord->
egg. — as long as you launch it anew that should be fine
<discord->
egg. — just don’t reuse the same already-flying vessel
<discord->
Stonesmile. — The vessels are simulated in KRASH, but the issue does appear on "real" vessels, would you like me to send a real onw up with 2602?
<discord->
Stonesmile. — The vessels are simulated in KRASH, but the issue does appear on "real" vessels, would you like me to send a real one up with 2602? (edited)
<discord->
egg. — yeah please don’t use KRASH, it is known to interact weirdly
<discord->
egg. — > the EVA parachute directly sets the angular velocity of the Kerbal
<discord->
egg. — Back to the parachute mines
<discord->
egg. — @ferram4 you were right in recommending that I consider it all cursed
egg|cell|egg has quit [Ping timeout: 378 seconds]
<discord->
ferram4. — Everything is broken \o/
<discord->
egg. — OK so maybe I could look at the angular velocity before and after KerbalEVA does its thing to infer the equivalent torque, and apply that
<discord->
Stonesmile. — Sorry for delay, can confirm that the issue remains on a non-KRASH vessel with 2602
goblin has quit [*.net *.split]
icefire has quit [*.net *.split]
goblin has joined #principia
icefire has joined #principia
_whitelogger has joined #principia
<discord->
egg. — Ok, if it remains with test2626 I have no idea then
<discord->
Stonesmile. — The issue is avoidable in my usecase, since i can make another part the root part, it might just be a couple parts that don't play nicely with principia
armed_troop has quit [*.net *.split]
armed_troop has joined #principia
egg|anbo|egg has quit [Remote host closed the connection]