03:03 <discord-> A​cer_Saccharum. — Just say no to patched conics

03:18 <UmbralRaptop> Lord of CVEs out there. Patch your conics

05:50 <discord-> S​ir Mortimer. — Wow

06:05 <discord-> S​ir Mortimer. — Don’t you love it when people write short and concise issues that focus on one specific point instead of writing a rant with dozens of bullet points that they then accidentally post on github instead of tumblr

06:59 <discord-> l​amont. — some of the old mechjeb issues were a spew of like 4 or 5 ideas that all are simple to state but take years to fix and can't be done without graduate level math.

07:26 <discord-> S​ir Mortimer. — re-reading that issue I could say that there is, *maybe* (can't confirm atm) one or two points that are somewhat valid. the rest is him a) not understanding what principia is or does and b) having a very, very, very bad day.

07:29 <discord-> S​ir Mortimer. — curious thing is that he seems to be a kind of software developer himself, he has 16 repositories on github. one would assume that a person like that would know how to write constructive bug reports.

07:46 <discord-> e​gg. — @XKdiver the “advance one orbit” thing is a bit like that

07:47 <discord-> e​gg. — Undergrad maths but grad astrodynamics I'd say

07:48 <discord-> e​gg. — Under∇

08:03 <discord-> e​gg. — @Sir Mortimer there's also "make the trajectories clickable to add/drag nodes that way", which is an obvious idea since stock does that and which I have no idea how to reimplement

08:05 <discord-> e​gg. — @Sir Mortimer don't even know what to do with the stupid issue, if I close it saying that issues should be focused I'm going to get another rant distributed over 30 issues >_>

08:11 <discord-> S​ir Mortimer. — well you can't do something to his satisfaction. kindly reply that some of his ideas have merit while others are just the way they are, and going to stay that way. If he knows how to help with the clickable node, he's more than welcome to provide some sample code that demonstrates how such a thing could be achieved. if he feels that certain points really should be dealt with, please open new t

08:28 <discord-> e​gg. — @Sir Mortimer you may be thinking too highly of developers :-p

08:30 <discord-> e​gg. — In lighter news, there's a fun discussion about procedurally generating geopotentials in stock which led to discussing cube-shaped planets

08:30 <discord-> e​gg. — https://inspirehep.net/record/1118525/files/CubeLake.png

09:02 <discord-> S​ir Mortimer. — wait that's a lake on a cubic planet?

10:27 <discord-> e​gg. — Yup

10:30 <discord-> S​ir Mortimer. — so, procedurally generated geopotentials. mascons for kerbin?

11:18 <bees> but you need to account for lake mass changing geopotentials, changing lake mass and shape again, changing geopotentials again, changing lake mass and shape aga...

11:29 <discord-> S​ir Mortimer. — Pah. Fluids have no mass, everybody knows that. Otherwise they wouldn’t be called fluid, but masdive. Duh

11:29 <discord-> S​ir Mortimer. — Pah. Fluids have no mass, everybody knows that. Otherwise they wouldn’t be called fluid, but massive. Duh (edited)

15:18 <discord-> R​urouniDonut. — Huh

15:19 <discord-> R​urouniDonut. — I tried making a flight path to go to the moon from a polar sat and it seemed pretty simple

15:19 <discord-> R​urouniDonut. — I think I'm gonna try it out

16:42 <discord-> S​tandecco. — question is, why is he checking the remaining burn time in the stock navball instead of the flight plan window?

16:46 <discord-> R​urouniDonut. — Is there anyway to make RT flight computer or mechjeb to execute a principia maneuver? My burn starts when I'm out of contact

17:03 <discord-> K​obymaru. — honestly, just disable the strict RT mod

17:03 <discord-> K​obymaru. — honestly, just disable the strict RT mode (edited)

17:05 <discord-> K​obymaru. — there isn't really a way

17:06 <discord-> K​obymaru. — last time i tried it, I programmed the burn time with the flight computer. But principia and stock don't agree on the time, so your burn will be a few seconds/minutes off (depending on how long before the burn you program it)

17:06 <discord-> K​obymaru. — plus, there is no way to do ullage with RT

17:07 <discord-> K​obymaru. — except for using the H key, which is kind of a cheat already

17:11 <discord-> R​urouniDonut. — What about with Mecjeb?

17:11 <discord-> R​urouniDonut. — *Mechjeb

17:20 <discord-> R​urouniDonut. — Welp with a late burn and rcs I can impact

17:20 <discord-> R​urouniDonut. — https://cdn.discordapp.com/attachments/480397772248580098/610522908632285195/screenshot68.png

17:51 <_whitenotifier-5dfc> [Principia] pleroy pushed 3 commits to master [+0/-0/±3] https://git.io/fj5GP

17:51 <_whitenotifier-5dfc> [Principia] pleroy b0ba718 - Avoid unbounded behavior.

17:51 <_whitenotifier-5dfc> [Principia] pleroy 0dff847 - Typos.

17:51 <_whitenotifier-5dfc> [Principia] pleroy 1bf7998 - Merge pull request #2281 from pleroy/2261a Avoid a unbounded behavior introduced in #2279

17:51 <_whitenotifier-5dfc> [Principia] pleroy closed pull request #2281: Avoid a unbounded behavior introduced in #2279 - https://git.io/fj7hb

18:31 <discord-> S​ir Mortimer. — I cooked up a flight plan that put me into a polar eve orbit using 2 manœuvre nodes, using this apparently god awful and totally useless principia flight plan UI. must have been a fluke.

18:31 <discord-> S​ir Mortimer. — from LKO, mind you.

18:44 <discord-> R​urouniDonut. — How do you guys go about finding interplanetary launch windows in rss?

19:00 <discord-> l​ipo842. — I personally just place a random satellite to 400 km circular orbit in lunar plane and use mechjeb maneuver planner, in particular advanced transfer to another planet, plot the lowest dV transfer and add an alarm to that

19:13 <discord-> l​pg. — that's reliable with principia?

19:14 <discord-> R​urouniDonut. — Going into Lunar plane is expensive depending on Precession

19:27 <_whitenotifier-5dfc> [Principia] eggrobin commented on issue #2282: Flight planner is bad - https://git.io/fj5nw

19:27 <_whitenotifier-5dfc> [Principia] eggrobin closed issue #2282: Flight planner is bad - https://git.io/fj5e8

19:28 <discord-> e​gg. — @Sir Mortimer I have replied to the rant

19:28 <discord-> e​gg. — pray I don't reply to the rant again

19:42 <_whitenotifier-5dfc> [Principia] Hubcapp commented on issue #2282: Flight planner is bad - https://git.io/fj5n1

19:47 <discord-> i​tsRyan. — @RurouniDonut It's not worth doglegging if you're very far above plane. Better to just do a simple Hohmann from an AN/DN when you cross the lunar plane. Or if you don't want to calculate that you can do a simple Hohmann at any time and correct plane mid course.

19:57 <discord-> i​tsRyan. — @RurouniDonut It's not worth doglegging if you're very far above plane. Better to just do a simple Hohmann from an AN/DN when you cross the lunar plane. Or if you don't want to calculate/wait for that to line up you can do a simple Hohmann at any time and correct plane mid course for a few hundred m/s or less. (edited)

20:03 <discord-> G​regrox. — @egg why did you have to write your own version of math functions like (I think) cube roots

20:03 <discord-> G​regrox. — unrelatedly, I wonder if the navball in the reference frame that is attached to the celestial sphere, could have a little star map projected onto it

20:04 <discord-> G​regrox. — just with RSS/RO

20:04 <bees> i once wrote my own math functions for speed

20:04 <bees> it was possible for my narrow use case, and with slight loss of accuracy i got 10x speed increase in slowest point

20:05 <discord-> G​regrox. — obviously it (navball thing) wouldn't make sense in general since people have different skyboxes with different stars, but for RSS at least, even if you don't have a real skybox installed, there is a true, canonical star map in the form of our night sky

20:10 <discord-> e​gg. — bees: @Gregrox: reproducibility, and fun

20:10 <_whitenotifier-5dfc> [Principia] pleroy commented on issue #2261: Computation of markers is costly - https://git.io/fj5cI

20:11 <discord-> e​gg. — also it happens that my cbrt is the best one around

20:11 <discord-> G​regrox. — cbrt?

20:11 <discord-> G​regrox. — cube root

20:11 <discord-> e​gg. — cube root

20:11 <_whitenotifier-5dfc> [Principia] pleroy closed issue #2261: Computation of markers is costly - https://git.io/fj9IA

20:11 <discord-> G​regrox. — surely the problem of "how to cube root a number" has been solved by this point in history

20:11 <discord-> e​gg. — but mostly the main goal is that it behaves the same across platforms, otherwise it's whatever the target libm does which is wild

20:12 <discord-> e​gg. — @Gregrox oh yes, many times, in many different ways

20:12 <discord-> e​gg. — https://twitter.com/eggleroy/status/988114872616484866

20:12 <discord-> e​gg. — cc @bofh453 (mofh)

20:12 <discord-> G​regrox. — would the result be the same, but say the time taken to do the calculation would be different in different places?

20:12 <discord-> e​gg. — The result will vary a bit, and the time too

20:12 <discord-> G​regrox. — and would the difference be in different parts of the code or on different machines or both?

20:13 <discord-> G​regrox. — (forgive me, my understanding of computer science is fairly rudimentary)

20:13 <discord-> e​gg. — correctly-rounded is going to have horrible worst case time cost, so all implementations are only faithful (and some are not even faithful but they're not trying)

20:14 <discord-> G​regrox. — what does faithful mean in this context?

20:14 <discord-> e​gg. — (correctly rounded = result is the representable value nearest the true value, faithful = result is one of the representable values on either side of the true value)

20:15 <discord-> G​regrox. — ok I think I understand. So if the number were 31.5 and we could only do integers, correctly rounded would be 32, while faithful could be either 31 or 32?

20:16 <discord-> e​gg. — yeah, you picked a tie which is a bit of an edge case

20:16 <discord-> e​gg. — but let's say the true result is 31.4999999999999999999999996

20:16 <discord-> e​gg. — in order to correctly decide if it's 31 or 32, you need to very precisely compute the true result

20:16 <discord-> e​gg. — to get past all those nines

20:17 <discord-> e​gg. — otherwise if you don't carry enough precision, you might get 31.50000000001, and misround

20:17 <discord-> e​gg. — so some values are hard to round well

20:20 <discord-> G​regrox. — This whole project is really interesting to learn and read about

20:21 <discord-> G​regrox. — because it's probably the most "academic" thing ever done with this game, if that makes sense

20:21 <discord-> G​regrox. — This whole project is really interesting to learn and read (and talk) about (edited)

20:22 <discord-> G​regrox. — KSP teaches you what an orbital period is

20:22 <discord-> G​regrox. — principia teaches you that not only are orbits more complicated than ellipses, not even orbital period is as easy to define as we expect

20:24 <_whitenotifier-5dfc> [Principia] pleroy commented on issue #2247: Poor performance at max timewarp in Realism Overhaul - https://git.io/fj5cl

20:24 <discord-> G​regrox. — will there ever be any kind of station-keeping options in principia? Like to keep your orbit circular in, say, the Duna-Ike system, or to stay around a lagrange point?

20:30 <discord-> S​ir Mortimer. — orbiting Gilly is fun.

20:30 <discord-> S​ir Mortimer. — factoring in the perturbations caused by the highly elliptic orbit around Eve is even more fun.

20:36 <discord-> e​gg. — @Gregrox station keeping is hard https://forum.kerbalspaceprogram.com/index.php?/topic/162200-wip151-161-17x-principia%E2%80%94version-ferrari-released-2019-08-01%E2%80%94n-body-and-extended-body-gravitation-axial-tilt/page/53/&tab=comments#comment-3630422

20:38 <discord-> G​regrox. — I really do want to make a principia-compatible binary star system some time

20:38 <discord-> G​regrox. — if I do it at KSP scale those problems wouldn't show up I expect

20:38 <discord-> G​regrox. — the problems about that guy's trinary system

20:38 <discord-> G​regrox. — not stationkeeping

20:40 <discord-> G​regrox. — yeah I figured stationkeeping would be difficult

20:40 <discord-> e​gg. — I should go back to writing mean element tests

20:41 <discord-> e​gg. — @lamont did I show you the SP3 parser madness

20:42 <discord-> e​gg. — following a spec is hard https://github.com/mockingbirdnest/Principia/blob/1bf799867cd3fd447a516af19b1ff51a6a3b51b3/astronomy/standard_product_3.hpp#L39-L80

20:44 <discord-> G​regrox. — I need to have an equation for figuring out the j2 of the gravitational field of an oblate body, which for the sake of simplification, is of uniform density

20:45 <discord-> G​regrox. — I would assume the j2 of a body of non-uniform density would just be some compromise on spherical vs max eccentricity

20:51 <discord-> R​urouniDonut. — @itsRyan Correction burns always seem to cost a lot more then 100 DeltaV when I do them

20:53 <UmbralRaptop> 1 DeltaV = 1 m/s?

20:58 <discord-> e​gg. — UmbralRaptop: I think so; this use of Δv for m/s of Δv is deeply irksome >_<

21:08 <discord-> i​tsRyan. — @RurouniDonut Of course it depends on how far off plane you are and how far off the AN/DN you do your Hohmann transfer. But If you're let's say 8 degrees lattitude high on your launch but otherwise time it right to line up with the moon, and somewhat close to AN or DN on the burn it should be right around 100 m/s for a mid course intercept burn. Up to around 400 m/s if you're 90 degress off A

21:11 <discord-> l​amont. — @Gregrox think its ultimately just an integral around the whole volume of the body where the density at every point is inside the integral

23:06 <discord-> G​regrox. — sure I believe you I guess but that isn't helpful to me

23:20 <discord-> l​amont. — yeah i can't find any good article on wikipedia about going from a mass distribution to generating the coefficients of the spherical harmonics

23:22 <discord-> e​gg. — Fukushima wrote something https://www.researchgate.net/publication/328476138_New_approach_to_compute_gravitational_field_of_general_object_accurately

23:23 <discord-> l​amont. — best i've got is some 15 year old handwavey recollections of how its similar to fourier transforms only the basis is the laplace functions rather than sin/cos and mumble mumble hilbert spaces and such...

23:24 <discord-> e​gg. — @lamont yeah but that's the easy part; that's taking the spherical harmonics transform of some surface

23:25 <discord-> e​gg. — but you're not given a geoid, you're given a topography

23:25 <discord-> e​gg. — makes the thing more fun, see the past few forum posts

23:25 <discord-> l​amont. — ah

23:29 <discord-> l​amont. — well i see some volume integrals here with a density function inside of them....

23:29 <discord-> l​amont. — and finite differences and double exponential quadrature and other things that make sense to be in this paper...

23:32 <discord-> l​amont. — https://iopscience.iop.org/article/10.3847/1538-3881/aa88b8

23:32 <discord-> l​amont. — that paper looks slightly more accessible and probably applicable?

23:33 <discord-> l​amont. — but it also contains a large literature review of "polyhedron" models, which apparently are easier but hot garbage for accuracy -- but i wonder if they're not perfect for KSP?

23:34 <discord-> l​amont. — like this one:

23:34 <discord-> l​amont. — https://link.springer.com/article/10.1007/BF00053511

23:35 <discord-> e​gg. — Toshio Fukushima absolutely loves writing papers whose title starts with "Precise and Fast Computation of"

23:35 <discord-> e​gg. — https://github.com/mockingbirdnest/Principia/blob/1bf799867cd3fd447a516af19b1ff51a6a3b51b3/numerics/elliptic_integrals.cpp#L20-L27

23:36 <discord-> l​amont. — yeah i see that, he likes to cite himself a lot as well

23:36 <discord-> l​amont. — > The classical manner of representing arbitrary gravitational fields is by expanding the gravitational potential into a harmonic series and then explicitly computing the series coefficients (MacMillan, 1936; Kaula, 1966; Heiskanen et al., 1967). Series coefficients are evaluated as integrals over the volume of the body, and for a constant density body may be reduced to integrals over the surface of

23:36 <discord-> l​amont. — that 1936 paper is probably more the first-year-grad-student approach

23:36 <discord-> e​gg. — he's really into elliptic functions and elliptic integrals

23:38 <UmbralRaptop> aaaaaa

23:38 <discord-> e​gg. — we use a translation of his implementations of those

23:39 <discord-> P​rojectGemini. — Honestly such a power move citing yourself constantly

23:39 <discord-> e​gg. — also we sent him some errata to his FORTRAN implementation so he just published a shiny new version

23:41 <discord-> e​gg. — https://www.researchgate.net/publication/335109652_xelbdj2txt_Fortran_test_driver_for_elbdj2_upgraded_version_of_subroutine_to_compute_the_double_precision_general_incomplete_elliptic_integrals_of_all_three_kinds_which_is_now_including_celbdj_new_subr

23:42 <oeuf> cc mofh

23:44 <discord-> l​amont. — the werner and scheeres paper seems more accessible for anyone with only college undergrad calculus

23:44 <discord-> e​gg. — @lamont you might want to mention that to eriksonn in the forum thread