Devnote Tuesday: Full Steam Ahead!

[Starwaster]

There is also a stock warp module. It kept showing up when I was messing around in MonoDevelop.

I keep meaning to experiment with that. Did you get it to do anything?

[Superfluous J]

And doesn't really generate effects anything like a real Lagrange point anyway.

Neither does the original proposal

[Bill Phil]

And doesn't really generate effects anything like a real Lagrange point anyway.

You could say the same about patched conics...

It works as an approximation, but it's still not accurate.

[White Owl]

So this stock rotation module has been available for half a year, but only a handful of people ever knew about it? Is that right? Awesome.

Any word on whether this thing is more or less stable than Infernal Robotics? I stopped using IR after the nth vehicle exploded for no reason.

[FlowerChild]

So this stock rotation module has been available for half a year, but only a handful of people ever knew about it? Is that right? Awesome.

Any word on whether this thing is more or less stable than Infernal Robotics? I stopped using IR after the nth vehicle exploded for no reason.

Myself and a few others are running a bunch of tests on it and discussing it over in the BTSM thread if you're interested, as I'm planning on trying to turn this into a balanced part of career mode.

http://forum.kerbalspaceprogram.com/threads/61632

The only real problem we discovered with it so far is that parts of a vehicle don't collide with parts of the same vehicle, which is also true of other parts like landing gear, but it's far more noticeable when parts start spinning around. Otherwise it seems very solid so far.

And yeah, it's got to have been like 6 months or so that this has been in the game, and I heard squat about it before today or I would have jumped right on that thing

[tntristan12]

You could say the same about patched conics...

It works as an approximation, but it's still not accurate.

Here's the thing though. It's *very* accurate! Unlike the "invisible third planet" idea that keeps popping up, patched conics are actually based on real-world physics. I'm not sure a lot of N-body physics activists (or "hacktivists" in the case of those who *really* want this invisible third planet) realize that KSP didn't invent patched conics. The approximation is set up by treating the gravitational force from the "smaller" body as a disturbance until the point where its magnitude is greater than that of the "larger body." Doing the math on it yields a sphere of influence inside which the smaller body is dominant.

You can create a MatLab simulation to see for yourself just how close the approximation is! You have to get to pretty large timesteps in your numerical integrator before you see a difference. Patched Conics is such a good approximation that it's often used as a starting point for more complex force models when planning interplanetary transfers (which I'm sure you know require insane amounts of precision). For KSP, Patched Conics is the best we can get without implementing a numerical integrator that runs at higher time warps (which would be very unstable in Unity). We're not talking "off by a few meters." This is more like "the Mun is now on a solar escape trajectory" kinda thing.

[Starwhip]

Full Steam Ahead!

Keep the wheel rolling, Dev Team!

[Bill Phil]

Here's the thing though. It's *very* accurate! Unlike the "invisible third planet" idea that keeps popping up, patched conics are actually based on real-world physics. I'm not sure a lot of N-body physics activists (or "hacktivists" in the case of those who *really* want this invisible third planet) realize that KSP didn't invent patched conics. The approximation is set up by treating the gravitational force from the "smaller" body as a disturbance until the point where its magnitude is greater than that of the "larger body." Doing the math on it yields a sphere of influence inside which the smaller body is dominant.

You can create a MatLab simulation to see for yourself just how close the approximation is! You have to get to pretty large timesteps in your numerical integrator before you see a difference. Patched Conics is such a good approximation that it's often used as a starting point for more complex force models when planning interplanetary transfers (which I'm sure you know require insane amounts of precision). For KSP, Patched Conics is the best we can get without implementing a numerical integrator that runs at higher time warps (which would be very unstable in Unity). We're not talking "off by a few meters." This is more like "the Mun is now on a solar escape trajectory" kinda thing.

L points act as gravitational bodies. ACT. That's the key word. Patched conics is nice because you can predict where everything will be very easily. But it's still not accurate. Gravity from many bodies acts on an object. That's not the case in patched conic approximations.

[tntristan12]

L points act as gravitational bodies. ACT. That's the key word. Patched conics is nice because you can predict where everything will be very easily. But it's still not accurate. Gravity from many bodies acts on an object. That's not the case in patched conic approximations.

L points absolutely do NOT act like gravitating bodies. You are never pulled towards an L point. In fact, for 3 put of the 5 possible L points you are actively being pulled away. The other cases are only stable because the forces from the two bodies pull you in that direction. There is simply no way to simulate or approximate L points using an invisible gravitating body. It's a kludge - a hack. It's a dreadfully inefficient and problematic one. It is not even remotely close to what is physically happening, whereas patched conics have a basis in reality. It is accurate enough to simulate what happens in stable orbits around either body, and what happens when you go between. L points are a perturbation - an exception rather than the rule. They are not going to happen, so let's not get this thread more off topic than it already is.

[Rune]

I'm gonna go ahead and guess that the secret feature is robotic parts. We already have a stock rotatron module, just hidden away and not applied to any stock parts, but it's in the game without any plugins. Maybe it's an expansion of that? Would be great for space shuttles and stuff.

Waitwaitwait... what? We have a stock rotatron? And we've had it hidden away for the better part of a year?

Ok, then I guess it's time to put my newfound coding skills to work (as if you needed those to add amodule to a stock part), and press-gang the until-now-useless octagonal struts into greatness, and forget about laggy stock contraptions with dozens of parts that can't resist timewarp. I think I'll actually "code" a module manager config file to apply it to any part, that should be even more convenient. Rotating habs and choppers, here we go!

Just the other day I built this awesome monolithic 40m diameter ring station, actually. I figured out how to get rid of the ugly gaps, put enough fuel so it can act as depot, and do all of the above in under 130 parts plus ten (including struts!) for the launcher, now the thing is going to look gloriously dynamic when docked in pairs! And then there's the thread about turboprops that got me thinking about propellers... this is so much awesome. Rear-loading cargo bay ramps, even!

Rune. Wish I had heard of this sooner!

Edited February 20, 2015 by Rune

[Superfluous J]

Wish I had heard of this sooner!

A lot of wish we'd heard of it sooner.

[Majorjim!]

Waitwaitwait... what? We have a stock rotatron? And we've had it hidden away for the better part of a year?

Ok, then I guess it's time to put my newfound coding skills to work (as if you needed those to add amodule to a stock part), and press-gang the until-now-useless octagonal struts into greatness, and forget about laggy stock contraptions with dozens of parts that can't resist timewarp. I think I'll actually "code" a module manager config file to apply it to any part, that should be even more convenient. Rotating habs and choppers, here we go!

Just the other day I built this awesome monolithic 40m diameter ring station, actually. I figured out how to get rid of the ugly gaps, put enough fuel so it can act as depot, and do all of the above in under 130 parts plus ten (including struts!) for the launcher, now the thing is going to look gloriously dynamic when docked in pairs! And then there's the thread about turboprops that got me thinking about propellers... this is so much awesome. Rear-loading cargo bay ramps, even!

Rune. Wish I had heard of this sooner!

Hold on just a cotton picking moment. Useless cubic struts? What planet you on Rune..?

And what's all this talk of a rotation module? What is this? Surely by modifying the code to add rotation to a part is a waste of time? Isn't this what IR does already?

Also, you don't have to use your new 'skills'. Someone has done this already: http://www.curse.com/ksp-mods/kerbal/223357-spin-o-tron-motors

There is a part using this module. And the code to add it to a part is on the page before this..

MJ

Edited February 20, 2015 by Majorjim

[Rune]

Hold on just a cotton picking moment. Useless cubic struts? What planet you on Rune..?

And what's all this talk of a rotation module? What is this? Surely by modifying the code to add rotation to a part is a waste of time? Isn't this what IR does already?

Also, you don't have to use your new 'skills'. Someone has done this already: http://www.curse.com/ksp-mods/kerbal/223357-spin-o-tron-motors

There is a part using this module. And the code to add it to a part is on the page before this..

MJ

Not cubic octagonal struts, just plain octagonal struts. These guys. I've used them a grand total of about two times in two years, and then dropped them since they have pretty much the same form factor as probe cores, but no added benefit. Not that I use the cubics much anyway, 90% of their use is creating surface nodes, and I do that by attaching with Ed Tools. Using large numbers of them is asking for trouble for abusing physic-less parts, too.

As to not installing IR, that's one less mod substituted by a measly config file for module manager, so no added textures and/or scripts. That saves resources, not to mention the plague of bugs IR has had since it's first implementation. I've already had enough mod headaches here! Nowadays I try to run a lean install as close to stock as I can, saves you a lot of trouble. I save all of my excess RAM to visual improvements, really.

Rune. The "coding" part are three crappy lines in a module manager config file I can steal from a gazillion places like KAS, I know. Hence all the ".

Edited February 20, 2015 by Rune

[Majorjim!]

Not cubic ortogonal struts, just plain octagonal struts. These guys. I've used them a grand total of about two times in two years, and then dropped them since they have pretty much the same form factor as probe cores, but no added benefit. Not that I use the cubics much anyway, 90% of their use is creating surface nodes, and I do that by attaching with Ed Tools. Using large numbers of them are asking for trouble for abusing physic-less parts, too.

As to not installing IR, that's one less mod substituted by a measly config file for module manager, so no added textures and/or scripts. That saves resources, not to mention the plague of bugs IR has had since it's first implementation. I've already had enough mod headaches here! Nowadays I try to run a lean install as close to stock as I can, saves you a lot of trouble. I save all of my excess RAM to visual improvements, really.

Rune. The "coding" part are three crappy lines in a module manager config file I can steal from a gazillion places like KAS, I know. Hence all the ".

KK.

I think Squad didn't tell us about the rotation module because it's glitchy and unstable like IR.

They have said before that Unity does not like rotating parts. I will be amazed if they add them.

Pleasantly amazed.

[Jovus]

It just struck me: has Squad asked Valentina Tereshkova about using her name in this fashion? It would at least be polite to do so first.

They probably have. Maxmaps seems pretty on top of things.

[cantab]

Regarding Lagrange points, is it possible to simulate them with an SOI if the potential in that SOI is something other than 1/r²? That way you don't have silly singularities and you can even create the effect of being pushed away from the point. Or is this unworkable for some reason?

[FishInferno]

It just struck me: has Squad asked Valentina Tereshkova about using her name in this fashion? It would at least be polite to do so first.

They probably have. Maxmaps seems pretty on top of things.

I believe that the Kerbal's name will be "Valentina Kerman", not "Valentina Tereshkova". I do not believe that the name Valentina is copyrighted.

[Hcube]

Jovus :

"It just struck me: has Squad asked Valentina Tereshkova about using her name in this fashion? It would at least be polite to do so first.

They probably have. Maxmaps seems pretty on top of things."

---->I guess it's perfectly fine as long as it's only the first name

[Starwaster]

Regarding Lagrange points, is it possible to simulate them with an SOI if the potential in that SOI is something other than 1/r²? That way you don't have silly singularities and you can even create the effect of being pushed away from the point. Or is this unworkable for some reason?

There are a number of possible workable solutions of simulating Lagrange points, or indeed gravity from bodies external to the current SOI. And you don't have to go with something that 'breaks the bank' in computational resources.

I do think though that a pre-requisite to even starting such a project is to be willing to accept that whatever solution you go with is not 'the right solution'.

You only have to implement something that produces an effect in-game that approximates the effect of 'being within a Lagrange point'.

That's not going to be a popular opinion with anyone unhappy with certain aspects of the game that are unrealistic but I guess it depends on whether you want something that leans more towards a game or more towards being an accurate physical simulation.

I've always preferred my KSP to be 'more realistic' but I'm willing to accept certain levels of unrealism. (otherwise I wouldn't have been able to make something like Stock Drag Fix, which is totally unrealistic but provides approximately correct results and doesn't break compatibility with stock aero control surfaces or lifting surfaces or the stock SAS system)

[Jovus]

I'm not concerned about copyright, 'cause you're right; this is at the very least covered under Fair Use very easily. I just mean for politeness' sake. If I had to guess I'd say she'd probably be flattered anyway.

Totally unrelated cool bit that most of you probably already know: she volunteered for Mars One.

[John FX]

There are a number of possible workable solutions of simulating Lagrange points, or indeed gravity from bodies external to the current SOI. And you don't have to go with something that 'breaks the bank' in computational resources.

I do think though that a pre-requisite to even starting such a project is to be willing to accept that whatever solution you go with is not 'the right solution'.

You only have to implement something that produces an effect in-game that approximates the effect of 'being within a Lagrange point'.

That's not going to be a popular opinion with anyone unhappy with certain aspects of the game that are unrealistic but I guess it depends on whether you want something that leans more towards a game or more towards being an accurate physical simulation.

I've always preferred my KSP to be 'more realistic' but I'm willing to accept certain levels of unrealism. (otherwise I wouldn't have been able to make something like Stock Drag Fix, which is totally unrealistic but provides approximately correct results and doesn't break compatibility with stock aero control surfaces or lifting surfaces or the stock SAS system)

Exactly. Most people would just be happy being able to put a sat in a solar orbit at the right point and for it to remain there through gravitational attraction, which is what happens in RL although it's a much more complex gravitational interaction.

As to how useful that would be, well I'm not sure I'd bother to do it myself.

[Starwaster]

I'm not concerned about copyright, 'cause you're right; this is at the very least covered under Fair Use very easily. I just mean for politeness' sake. If I had to guess I'd say she'd probably be flattered anyway.

Totally unrelated cool bit that most of you probably already know: she volunteered for Mars One.

Whaaat???? Well did she make it? She better have....

[FlowerChild]

KK.

I think Squad didn't tell us about the rotation module because it's glitchy and unstable like IR.

Yup, I think that's absolutely right.

Further testing on it shows that when you're dealing with a chain of parts with multiple motors attached (like you would for a robot arm), parts become displaced during the save/load cycle and you wind up with stuff like this:

That segment is just floating in mid-air still fully functional, but obviously not where it's supposed to be.

So, in combination with some other more minor issues (like rotated parts passing through other parts in the same vessel), it looks like there's very good reason it has yet to be released as a stock part.

Edited February 21, 2015 by FlowerChild

[tntristan12]

Yup, I think that's absolutely right.

Further testing on it shows that when you're dealing with a chain of parts with multiple motors attached (like you would for a robot arm), parts become displaced during the save/load cycle and you wind up with stuff like this:

http://i.imgur.com/glK7SE9.png

That segment is just floating in mid-air still fully functional, but obviously not where it's supposed to be.

So, in combination with some other more minor issues (like rotated parts passing through other parts in the same vessel), it looks like there's very good reason it has yet to be released as a stock part.

This bug literally ruined my kethane base way back when. My miner got so bowed in the middle that it could no longer drive, so I nuked the save because it got so frustrating

[White Owl]

Ahh... so it is like Infernal Robotics after all. Well, thanks for testing anyway.