03:04 <Tone> Do they still have hill spheres in principia?

03:07 <UmbralRaptor> No, but yes.

03:07 <UmbralRaptor> That is, not as a formal thing, but they remain a useful rule of thumb for orbit stability.

06:03 <GH> [Principia] scimas opened issue #1453: Plotting frame changes to body centered inertial when using Revert to Launch https://git.io/vQzE7

07:57 <GH> [Principia] scimas opened issue #1454: Random crash when timewarping on Mun https://git.io/vQzoO

17:42 <GH> [Principia] eggrobin commented on issue #1447: The full FATAL failure:... https://git.io/vQgaW

17:42 <GH> [Principia] eggrobin commented on issue #1454: The FATAL failure is a trapped C♯ exception;... https://git.io/vQgal

20:16 <lamont> egg: what APIs would break in “porting” MechJeb to principia

20:17 <lamont> like… don’t think too hard about calculations that MJ does which will be wrong, or the ways it does guidance incorrectly and such

20:18 <lamont> but what fundamental APIs change?

20:18 <GH> [Principia] pleroy opened pull request #1455: An API for segment hiding (master...Segment) https://git.io/vQgMe

20:19 <egg> lamont: what do you mean by porting

20:19 <egg> if you don't care about things being wrong, there's no porting to do

20:19 <lamont> yeah

20:19 <lamont> so the APIs are there

20:20 <lamont> but like nodes don’t “Burn down” so the node executor will be a pile of lolwut

20:20 <egg> otoh if you care about the difference between the Keplerian prediction and the n-body one... that's going to be fun

20:20 <lamont> yeah

20:20 <lamont> so all the PEG ascent guidance should work then

20:20 <egg> and since it deals with extended burns the concept of "burn down" makes little sense, we've been thinking about that but we'd need to write optimization code

20:21 <lamont> yeah

20:21 <lamont> so PEG is the first step to having proper closed loop guidance

20:21 <egg> yeah, but if we did guidance we'd likely do it C++-side

20:21 <egg> (if only to benefit from our typed APIs)

20:22 <egg> also because you'd need accurate predictions, which come from our APIs

20:22 <lamont> it evolved into the apollo guidance system, which was also applied to landing, which evolved into the shuttle guidance which also executed burns

20:22 <egg> 1 min, gotta review phl's CL

20:22 <lamont> yeah i gotta fix some code that just broke

20:27 <egg> lamont: I'm bac

20:27 <egg> s/$/k

20:27 <Qboid> egg meant to say: lamont: I'm back

20:27 <lamont> yeah, so i think i get it

20:28 <egg> lamont: get what?

20:29 <lamont> got a sketch in my head of the road forwards to having useful maneuver execution + planning in principia

20:31 <egg> lamont: well, planning is on our side though

20:31 <egg> lamont: and I might be interested in eggsecution too, optimization is fun

20:31 <egg> lamont: so I'm not sure how you'd integrate that with our currently nonexistent eggsternal API

20:32 <egg> lamont: unless you want to contribute things to principia, which is some interesting madness :D

20:32 <egg> lamont: ask UmbralRaptor :D

20:37 <egg> !wpn lamont

20:39 <egg> !wpn ferram4

20:39 <ferram4> ...how about not and say we didn't?

20:41 <egg> ferram4: not what?

20:42 <ferram4> A presumably radioactive element

20:42 <egg> oh

20:43 <egg> ferram4: I thought you were talking about lamont's guidance questions

20:43 <ferram4> Oh. I'm interested in that.

20:43 <ferram4> Mostly from an optimization POV

20:43 <ferram4> See #RO

20:43 <lamont> lol

20:44 <egg> ferram4: yes, but as discussed, so are we :D

20:44 <egg> ferram4: and it's probably a good idea to keep as much of the computation principia-API-side as possible, since it has to be backed by our simulation

20:45 <egg> lamont: ferram4: that way we can all enjoy https://twitter.com/whitequark/status/840988072116273153

20:45 <kmath> <whitequark> there's C++ code and then there's C++ code in Principia https://t.co/7yzqs9v8uL

20:51 <egg> lamont: ferram4: what could work is to have very high-level guidance APIs (close to what the UI will need/do) that call the principia lib (via our adapter or directly via the proto-generated interface code), and then implement guidance logic in C++ below that; we would have to jointly design that high-level API and keep it somewhat stable (and then we can all have fun writing the guidance itself as part of principia, but

20:51 <egg> with a fixed API that should be much easier to develop)

20:52 <ferram4> Make it work in C#, give it an opt pass there, then make it work in C++

20:54 <egg> lamont: ferram4: (continuing) the problem is that we're not really in a state where we can keep a stable API, things change quickly

20:55 <egg> ferram4: that works if you don't care about Principia, it doesn't work if you want to work with Principia's flight plans, or if you want to work with something where the Keplerian approximation is bad; there you *need* the principia libs

20:56 <egg> if you can't predict, you can't optimize (and obviously if you want to work with principia's flight plan, you have to work with principia's flight plan)

20:56 <egg> lamont: ferram4 this seems tricky.

20:58 <egg> s/4/:/

20:58 <Qboid> egg meant to say: lamont: ferram: this seems tricky.

20:58 <egg> uh

20:58 <egg> eh good enough

20:58 <ferram4> Well, for now, I'm mostly interested in PEG

20:58 <egg> ah, there it's not really principia's business :-p

20:59 <ferram4> Outside of that, I'd probably fly it by the seat of my pants.

20:59 <egg> ferram4: well you can still have your pants assisted by principia's flight planner :-p

20:59 <egg> computer-assisted pants

21:19 <GH> [Principia] pleroy closed pull request #1455: An API for segment hiding (master...Segment) https://git.io/vQgMe

21:19 <GH> [Principia] pleroy pushed 5 new commits to master: https://git.io/vQgSp

21:19 <GH> Principia/master caf6e09 pleroy: Segment hiding API.

21:19 <GH> Principia/master 68d53ce pleroy: After egg's review.

21:19 <GH> Principia/master 216aa50 pleroy: Error.

21:24 <egg> lamont: any plans to add the conditions to the Makefile? it would be nice to get things buildable on mac (also that would allow UmbralRaptor to go back to contributing)

21:24 <lamont> yep

21:25 <lamont> i’m just aggressive about only working on one thing (WIP limits)

21:27 <egg> lamont: also see backlog here for thoughts on principia-compatible guidance: it's complicated

21:27 <lamont> i am… not surprised...

21:28 <egg> :-p

21:30 <egg> lamont: this would be a longer-term discussion, maybe if we can come up with a stable API if we can manage to narrow things down to a small set of functions without tricky underlying logic (the computations will be complicated of course, but state/logic shouldn't be or it the API will be flaky as hell)

21:46 <egg> there are too many words in that sentence

21:47 <egg> s/ if (.*) it the (.*)/\1 the \2/

21:47 <Qboid> egg meant to say: lamont: this would be a longer-term discussion, maybe\1 the \2

21:47 <egg> uuuuuh

21:47 <egg> s/ if (.*) it the (.*)/$1 the $2/

21:47 <Qboid> egg meant to say: lamont: this would be a longer-term discussion, maybewe can come up with a stable API if we can manage to narrow things down to a small set of functions without tricky underlying logic (the computations will be complicated of course, but state/logic shouldn't be or the API will be flaky as hell)

21:47 <egg> close enough