egg changed the topic of #principia to: Logs: https://esper.irclog.whitequark.org/principia | <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…
UmbralRaptor has quit [Ping timeout: 189 seconds]
UmbralRaptop has joined #principia
UmbralRaptop has quit [Ping timeout: 189 seconds]
UmbralRaptop has joined #principia
<queqiao->
⟨Al2Me6⟩ Bug report: (I presume this has to do with the 3390 fix) when you move your cursor off the reference frame selector after switching frames, the camera will jump.
raptop has joined #principia
<queqiao->
⟨scimas⟩ Feels weird asking for features when I myself don't play KSP these days, but since you're talking about UI changes anyway and I wanted this when I used to play: Could you add small circles at the end of the trajectory predictions of celestials? So that you can know what their relative positions are going to be visually.
<queqiao->
⟨egg⟩ ⟪scimas⟫ Feels weird asking for features when I […] ⮪ Hm. Probably not a good idea, because the end is pretty much arbitrary (driven by your computation budget). But some way to peer into future positions might be nice.
<queqiao->
If we could detect hovering over the noodle things would be easier, but I don’t know how to approach _that_.
<queqiao->
⟨Butcher⟩ If you know mouse position and noodle position you can raycast, but I imagine raycasting to the noodles would be expensive.
<queqiao->
⟨egg⟩ ⟪Butcher⟫ If you know mouse position and noodle […] ⮪ yeah, I don’t see making colliders for them being a good idea.
<queqiao->
⟨egg⟩ (and then you’d need to work backward to figure out when the hovered point is, which is only really doable in the C++)
<queqiao->
⟨Butcher⟩ One could do it old school without explicit colliers, but it's still a ton of points.
<queqiao->
⟨Butcher⟩ * colliders,
<queqiao->
⟨egg⟩ ⟪Butcher⟫ One could do it old school without […] ⮪ The points are a bit of a distraction if you are doing it sufficiently old-school, because the internal representation is completely different.
<queqiao->
⟨Butcher⟩ And you can't even short cut it with bounding as most of the noodles will have huge bounding volumes.
<queqiao->
⟨egg⟩ So if I were doing that on the C++ side I could just walk the trajectory to find out which point is closest to the mouse in projection.
<queqiao->
⟨Butcher⟩ How many points in a typical noodle? I guess potentially a lot.
<queqiao->
⟨Butcher⟩ In other news, have a cat and cake.
<queqiao->
⟨siimav⟩ Cat looks really excited about the cake
<queqiao->
⟨egg⟩ ⟪Butcher⟫ How many points in a typical noodle? I […] ⮪ in the drawn noodle, probably; in the underlying computed noodle, perhaps quite a bit fewer.
<queqiao->
⟨egg⟩ (in fact the number of points in the computed fuchsia noodle is a setting you directly control)
<queqiao->
⟨Butcher⟩ Hmm, testing the underlying noodle might not be too bad then, at least you just test the points then you have a single line segment test (or would you need a curve?)
<queqiao->
⟨scimas⟩ ⟪egg⟫ Hm. Probably not a good idea, because […] ⮪ The problem with hovering is dealing with overlapping trajectories. For celestials they're going to be very close together (in cases of long flight plans).
<queqiao->
⟨Butcher⟩ ⟪siimav⟫ Cat looks really excited about the cake ⮪ Very excited. He's not allowed cake though, because it's poison to cats.
<queqiao->
⟨egg⟩ ⟪Butcher⟫ Hmm, testing the underlying noodle […] ⮪ I would probably go over the points to find the segments where I think it is closest, and then use Brent to actually find the closest point on the curve
<queqiao->
⟨egg⟩ This is, in fact, how the apsis markers are computed
<queqiao->
⟨egg⟩ ⟪scimas⟫ The problem with hovering is dealing […] ⮪ yeah, but I was thinking of hovering over the prediction or flight plan, and then hinting the positions of other bodies at that time.
<queqiao->
⟨Butcher⟩ That's quite a nice feature, maybe not worth the perf though.
<queqiao->
⟨egg⟩ well, once you have hover over the flight plan you have another nice thing which would be placing the manœuvres
<queqiao->
⟨egg⟩ and _that_ is definitely something a lot of people have been clamouring for.
<queqiao->
⟨scimas⟩ Fair enough, but think about the use case where, in stock, people just launch into LKO whenever and then do transfer burns during the transfer window. You can again have tightly packed flight plan there. Even a couple of orbits in low altitudes can get very close to each other.
<queqiao->
⟨egg⟩ ⟪scimas⟫ Fair enough, but think about the use […] ⮪ Yeah, it’s not a panacea.
<queqiao->
⟨scimas⟩ That will work for adding manoeuvres to the flight plan, I was talking about detecting hover for showing positions of other bodies.
<queqiao->
⟨Butcher⟩ For 2517 even just bumping the orbit by one period wild be useful even if it didn't keep the exact ejection angle, it would get you pretty in one click.
<queqiao->
⟨Butcher⟩ * would
<queqiao->
⟨egg⟩ ⟪scimas⟫ That will work for adding manoeuvres to […] ⮪ How does stock behave in that respect?
<queqiao->
⟨Butcher⟩ +close
<queqiao->
⟨egg⟩ Note that to an extent the proper way to think about the position of other things is a reference frame change; such hinting may be a nice convenience, but I don’t see it as a central feature.
<queqiao->
⟨scimas⟩ Stock does hover on the celestial's trajectory. And since everything is on rails.. there is only one ellipse to deal with. It doesn't give you the ability to see the positions at an arbitrary time in the future - only within one orbit of the celestial, future and past.
<queqiao->
You only get closest approaches to the celestial _after_ you make a manoeuvre that gets close to the celestial.
<queqiao->
⟨egg⟩ right; so in order to get similar behaviour we should look for the first locally closest point to the mouse ray that’s close enough when hovering, rather than the globally closest point.
<queqiao->
⟨egg⟩ ⟪Butcher⟫ For 2517 even just bumping the orbit by […] ⮪ Yeah, probably the nodal period would be a good general-purpose default; and then someday we could change that depending on the plotting frame for constant ejection angle if you are in HCI or GSE.
<queqiao->
⟨scimas⟩ ⟪egg⟫ right; so in order to get similar […] ⮪ Up to you of course, but I am not asking to recreate the stock behavior. The hover positions in stock don't have anything to do with relative positions of the planets after all. The point, from my perspective, is to visually understand what the state of the solar system is going to be - irrespective of whether you are going to any particular planet or not. Seems like it would be...
<queqiao->
... a good exploratory tool for planning long missions involving visits to multiple planets.
<queqiao->
⟨egg⟩ ⟪scimas⟫ Up to you of course, but I am not […] ⮪ Ah, that makes more sense.
<queqiao->
We had some discussion a looong time ago of being able to explicitly peer into the past or the future (we have your trajectory since launch, but after thousands of revolutions we cannot show your launch trajectory—besides performance concerns, it is hidden behind the ball of yarn).
<queqiao->
Perhaps that is
<queqiao->
⟨egg⟩ -Perhaps that is
<queqiao->
⟨egg⟩ Perhaps that should be approached in a different way entirely though (just show the map at a different time, instead of overlaying the distant future or distant past atop the present?)
<queqiao->
⟨Al2Me6⟩ A (likely unfeasible) suggestion re. node markers: when they are clustered and partly overlap, perhaps they can be indicated by drawing a single mesh segment of some kind? That would also help illustrate precession, perhaps.