ferram4

Alright, so for the people who are willing to help test and those who are impatient but fully understand the meaning of "it's a dev build, everything might break, don't complain about the Kraken taking your save" there is a 1.2-compatible dev build available here. Main things to test are that nothing spontaneously explodes during loading from timewarp, nothing is magically indestructible in crashes, and that the weldable construction docking ports from RoverDude's Konstruction Mod don't break KJR (if they do, that's my problem; say so here, not over there). Also, as this is a dev build, don't try to get it on CKAN or anything because it isn't confirmed stable yet. Also, as an added bonus, because gravity-based physics easing is now a part of the stock game KJR uses that system. This means it can be shut off if you want. Probably better not to though.

@WildLynx: Sounds like it's the same issue that @Shadowmage made a PR for. It should be fixed for the next release, but when I get a functional dev build up you should try that and see how it goes. @Everyone Else: FAR dev build will be out soon-ish, as soon as I'm sure that everything functions properly enough to put up for testing and once KJR is in a working dev state so I can remove the effects of wobble from my testing. Sorry about the delay, but patience. I'd like to avoid having to do half a dozen patches to fix all the NaNs slipping into the aero calcs like the last few times.

Just a heads up, but it might take slightly longer than normal for me to push out an updated version of KJR. 1.2 has its own version of physics easing and some changes regarding how vehicles spawn on the ground so I want to do some extensive testing once I have it actually working prior to releasing an official version. These systems are just too delicate to not be crazy about. As usual, no promises about when it'll be out though. This could be incredibly simple or it could require bashing my head into a wall.

Well, I just tried, and I cannot see any issues. It flies straight unless it goes far off prograde. Some funky voxelization related to the second stage engine, but I don't think that's the cause here. I'll still have to look at that though.

@nablabla: Standard RO install? What part packs?

@nablabla: Check in the VAB if the cross-section graph is properly aligned with the vehicle. If it is, then I concur with everyone else that it's a general instability / too much TWR combination. If it isn't, I need the craft file.

The way to get more lift out of flaps is to realize that they are plain flaps and cannot be deflected that much before they will stall. At very high deflections their main benefit is actually drag to help slow the plane down for landing. Odds are what's happening in both these pictures is that you've got the flaps deflected pretty heavily with very little in the way of actual wing surface ahead of them to work with, and so a shape that cambered is stalling out really suddenly. Larger wings, smaller flaps relative to the wings, less deflection should be fine. The Firehound example craft does a good job of showing how flaps work.

@AccidentalDisassembly: Yes, FAR does have some effects on buoyancy as a side result of having to rip out stock's drag code. The stock buoyancy model uses drag cubes (partially, IIRC) to model partial submersion of some things and FAR's replacement ends up with a slightly better buoyancy model. As for water drag, yes, it's very very high. Water is ~800x as dense as air, so the forces will be ~800 times as high, and on something as light as aircraft that tends to add up. You need something that can skim on the top of the water to get the drag down. And no, there currently isn't a way to turn off FAR's changing of wing mass, because then I would have to remove the mass-strength multiplier as well and that would be a complicated bit of code to do. @LordOfMinecraft99: Well, you should have all the dependencies with that. It's worth noting that FAR doesn't add any info to the "More Info" tab in the editor, and it removes the stock item, so that is correct behavior. And the FAR removes the arrow from the CoL marker because it's just a location, not a vector of any kind, so if the CoL exists somewhere on the vehicle rather than the center of the floor, then that too is correct behavior. And if that's the case, and the F12 aero overlay shows arrows coming out of your wings when they're at an angle of attack, then the problem is that your planes are too heavy, which if you just installed FAR and are used to building stock planes, that is almost certainly the case. @PTGFlyer: Last I heard no one had any idea why that happens, and given that it involves Kerbals I don't know how much I can do. I can look more into it, but the problem is that unless whatever method KIS uses to attach the part can trigger the stock game's vessel-update methods (considering that Kerbals count as vessels just like regular craft do), I can't do anything at all with it.

Stock aerodynamics do not model wing sweep at all. A wing will make the same lift and drag regardless of its sweep, so what you're trying to do here won't change anything about the wing's characteristics themselves. However, rotating the wing will move the CoL back, which means that you need more pitch-up authority from the tail, which means more drag from that, so you go slower. You would probably be able to go faster by instead shifting the wings forward to get the CoL closer to the CoM and reduce the forces required from the tail if you wanna stick with stock aero.

The upper bound for "very bad things" heavily depends on the mass ratio between the two parts in question and the number of separate masses that are connected; the more parts with greater the disparity in mass, the lower the value needs to be. Sorry I don't have an actual numbers answer, but I've seen too many situations where even the current values are too high, and I don't know enough about the internals of the physics engine to analyze that and get some details. That will fix the torque issues, but the resulting internal torques might cause the entire thing to fly apart, so it's probably not a smart idea. As for the clipping and the lack of a "linearspring" setting, that's because KJR really doesn't do anything to change the linear springiness of parts, besides setting the cap on the max force possible based on their size. Trying to change that inevitably causes far more problems, and it's only in completely insane situations (tiny things being asked to handle forces well beyond what they should) where issues occur. You could try increasing the breakForce to see if that fixes anything, but I have doubts. In practice, what you really need is to make sure that the masses of the crazy-stupid-OP engine and whatever it's attached to are somewhat near each other, that the engine's mass isn't stupidly low, and that its size is large enough for a decent connection. That should fix everything. Not much I can do for the other situation.

Currently all the wing lit is handled simply as an approximate implementation of lifting line theory, with compressibility corrections and some 2d shock-expansion calculations to adjust lift slopes for handling higher Mach number behavior. Calculate the lift based on angle of attack, apply it, include induced drag, that's pretty much it for wing lift.

Yes. Yes it does. The multiplying is a hold-over from awhile ago when the intake areas were originally set and made no sense, and to keep compatibility with any mods that balanced off of that the base values were never changed. So in order to keep things consistent FAR has to also account for that factor, so even though you've set it to 0.573 m, everything thinks that you actually have intakes the size of an aircraft carrier.

@Verran: Yes, that is certainly more than enough to trip it, because (for some unfathomable reason) the areas in-game are scaled down by a factor of 100. The circular intake part you welded on has, by default, an intake area of 0.006 (uh... dekameters^2?). So... that explains everything then. Welcome to KSP, where sometimes things don't make sense. @Boris-Barboris: So... now redo this analysis, acknowledging that cambered wings can have a substantial non-zero pitching moment even when the lift force is zero and the drag force is small. And the CoP must be placed so that the minuscule lift and drag forces have sufficient moment arm to balance the large, constant pitching moment. Similar effects happen on wings in stall as well. Actually, to get more into the maths, for an ideal, thin, 2d airfoil, the aerodynamic center (force location when allowing for a constant pitching moment) is always located at the quarter chord. Further, assuming you know the magnitude of that zero-lift pitching moment coefficient, you can solve for the forward/back location of the CoP using: X_cp = 0.25*c*(1 + 4 * Cm/Cl), where Cm is pitching moment coeff, Cl is lift coeff, and c is the chord length. Yeah, the CoP moves around a lot.

@baldamundo: By damping I mean damping. It reduces motion, like dampers do. Or tries to anyway. @Ourworldalpha1: Welp, then there's nothing I can do. That requires mesh changes, and I can't fix part meshes in other mods. @Vladokapuh: Well, FAR doesn't affect landing gear at all, so that means the only change is that you're hitting that ground harder with FAR or that you're getting a really high AoA after hitting the ground to get you back off the ground again. There really aren't any other options. @Boris-Barboris: Well, to start off with, the CoP moving off the wing is a thing that happens. When you're looking to put a force center somewhere that allows all pitching moments to be handled by only lift and drag forces, if the lift or drag forces are very small they need a large moment arm to cancel the pitching moments. Resulting in the CoP being placed somewhere completely off the vehicle. Which is why the CoP is absolutely terrible for any discussions of stability and why FAR uses the aerodynamic center instead because at least by allowing a constant moment you get a force center that is somewhere reasonable. The shift is based on the expected aerodynamic center shift for a 2d NACA airfoil in stall, which has it shift backwards. While I suspect that adding adjustments for aspect ratio and sweep angle might be able to band-aid this somewhat, the fact is that fundamental flaws in the current sim mean that it will only get good behavior in very specific cases. And it'll probably introduce bugs. And then instead of spending time working on the already-delayed wing overhaul, I'll be fighting new bugs. Just like the way that bugs have already prevented the wing overhaul from working. I'll accept PRs to add band-aids, but I'm not going to spend lots of time tweaking, adjusting and bugfixing a system I'm planning to rip out. @Verran: Uh... so I can't reproduce this issue myself using a basically identical setup, sans AJE, which leads me to believe AJE is the culprit, considering it has a way that it can cause this. FAR has a bit of code that attempts to model the extra spillage drag from intakes when jet engines are heavily throttled down; normally it's not that much, although for very slippery planes it can double the drag on them. Now, to get the spillage drag up high enough to raise the drag coefficient to 5.85 (!) AJE has to be increasing the "area" value of those intakes dramatically. I'll look into what's going on there, but I suspect that it's AJE in some way.

@RageMode: If it's only happening during staging, then the problem is that your upper stage isn't stable while you're still within the atmosphere. The realistic solution to this is... either wait until you're further out of the atmosphere to stage (like standard orbital launch vehicles) or put fins on the upper stage to stabilize it (like on sounding rockets). Also, make sure that your fairing isn't too much wider than the rest of the rocket, or else that'll produce silly amounts of body lift and make things a lot worse. @LordOfMinecraft99: The first thing I would do is clear out the extra ModuleManager files you have in there. You have multiple copies of the same MM in your save and while I know that MM will disable older copies, I don't know how it reacts to multiple MMs that are the same version. The second thing I would do is provide repro steps for the issue that use as few mods as possible. The third thing I would do is reinstall all of FAR (deleting the FAR folder completely) because the second FAR exception indicates that one of the configs isn't being loaded or something similar. That exception will never happen otherwise. @Ourworldalpha1: There are no FAR bugs with Procedural Parts, so I don't know what you're talking about there. The only FAR-specific Tweakscale bug was wrt the mass of tweakscaled wings. That was fixed, and there are no other Tweakscale-FAR interactions that occur unless something has changed in Tweakscale that prevents it from telling FAR to re-voxelize things. In any case, you need to provide more information before anyone can do anything, because all of the known Tweakscale bugs have been fixed long ago. @baldamundo: The wing leveler levels wings. The pitch damper damps pitch. The yaw damper damps yaw. The AoA limiter limits AoA. The Dynamic Control adjustment reduces control inputs to counter increasing dynamic pressure. At least the last of those doesn't sound quite so obvious. @Bloojay: Well, there is an option in the space center scene's FAR menu, that's generally a good place to start. On the other hand, body lift at high dynamic pressures will tear pretty much everything apart, including space shuttles. Citation: Challenger post-ET disintegration.

@Qwits: Frac Max 1 Drag only affects the linearized wave drag calculation that is used for Mach 1 and a few of the values above it; at the lowest, it is 0.7, at the highest, it is 1. For adjusting lift/drag in the source code, you'll have a lot of trouble doing that. The wing code is set to be overhauled and replaced anyway, and although the symmetrical airfoil is assumed otherwise the lift slope for wings is overly optimistic. Often ends up way too close to the value for a pure airfoil. Vortex lift is not modeled because the current wing model cannot properly determine where the leading edge separation vortices should go on highly swept wings (accounting for the mess of vortex breakdown and what happens on double-deltas, inverted deltas, etc.). Otherwise, the sudden stall is appropriate for a thinner symmetrical airfoil, especially one with as tight a nose radius as the 6A series. @MachTurtle: I try to avoid touching as many stock things as I can because whenever I do it results in numerous new bugs for me to chase down and fix. Since this behavior is handled by stock, and trying to fix it is an invitation to disaster, I won't. Besides, for something like that it should already be handled correctly by stock, so... bug report time for the devs. @RageMode: 20 kN of drag without the size of the rocket involved or the speed you're going doesn't actually tell me anything. Honestly that sounds like an incredibly low amount of drag unless you're getting it on a tiny rocket under 100 m/s. Honestly, that sounds like an incredibly streamlined, low-drag rocket to me. If your side forces are equal to your drag forces, then you aren't going straight up; probably not much angle of attack going on, but you aren't going straight up. Or your rocket is deliberately non-symmetrical, in which case, make it symmetrical. Most of my rocket designs haven't bothered with fins, but they involve lots of engine gimballing, an initial TWR of somewhere between 1.2 and 1.6, no higher, and a pitchover around 100 m/s and then I stick to prograde as much as possible. Moving off prograde for any reason is silly. Spending any significant time coasting between staging is also bad within the atmosphere, because the way Kerbin is set up it aero affects staging far more than on Earth, where staging doesn't happen until aerodynamic forces don't exist anymore (heat does though, but that's different). If your flipping happens exclusively in the upper atmosphere, and only ever when you move off prograde... stop moving off prograde. Seriously, body lift is a strong and powerful thing, wait until your dynamic pressure has dropped more.

@Qwits: There is no universal drag scaling setting within FAR. Especially not one that will affect subsonic flight. The only one that exists is solely for the sensitivity of the Mach 1 wave drag calculations, and that's only because the stock game's parts alone are too difficult to easily shape into the ideal shapes, so there is some leniency by default to account for the idea of, "this is close, but in reality it's shaped better than this." If you aren't noticing any changes to drag at Mach 1, then apparently your craft have very low wave drag overall if you can fly at Mach 1 and not notice any changes. The viscosity settings are only there because the stock game lacks values for viscosity for each of the planets and is only relevant for heat transfer and skin friction drag. Outside that, there are no other drag settings and never will be. I don't particularly care about popularity and considering the effect that too many users tends to have on the usefulness of bug reports I actually think it's a bad thing. As for "lack of lift," there isn't for the shapes that exist. FAR assumes symmetrical airfoils, based roughly on a somewhat rough-surfaced NACA 64010 for subsonic flight, but with a rather higher maximum lift coefficient of 1.6 rather than the realistic 1.2 (which technically should be even lower because roughness decreases max lift). Any problems actually getting planes to take off with that at low speeds are almost inevitably a case of, "not enough flaps," "not enough slats," "too much weight with not enough wing," or "not enough rotation on the runway." Hell, there's an example FAR plane (the Firehound) that lifts off the ground at 60 m/s if you really want to; it's sluggish as hell, but it does. @MachTurtle: If it's close behind the asteroid it should, to some extent. Conduction will still matter. If it's too far back, it will get heating anyway because the flow will have time to come back together behind the asteroid. That's actually stock behavior, not FAR though. @WildLynx & @Phineas Freak: Well unfortunately I don't exactly have the means to determine when a mesh is completely broken. People need to update their stuff, I can't exactly do it for them.

@eberkain: Alright, I'll look into it. Thanks for the repro steps, they're exactly as detailed as I need them. Should be easy enough to reproduce and fix when I get the free time to do so.

@eberkain: Uh... so then could you provide complete and full reproduction steps so that I can cause the issue and fix it?

@Tiberius K: If you can reduce the problem-craft to the smallest number of mods possible and then provide the craft, I'll look at it. Not looking at bugs at all is how they never get fixed. @MCSquared: Well, if you're using an out-of-date version of KJR, then the obvious solution is to updated KJR. I can't cause any issue like that in my game.

@eberkain: Well, I don't get that when I launch a vessel, so that means that you're missing some crucial reproduction steps. Narrow it down to the smallest number of steps possible with the smallest number of mods installed and I'll look at it.

No, it just renamed the class. And all the references to the class, so nothing broke. It's just a silly name change for now.

@rndomguy: That is why Lanchester was released, it's my post right above the last one.

...of course that would happen. Welp, I need to fix that then.

@rndomguy: That would be a minor problem, wouldn't it? I managed to find what it is, but it should have affected more than just v0.15.7.1, it should also have affected v0.15.7. Anyway, to fix that FAR v0.15.7.2 "Lanchester" has been released with the fix to that. Lanchester is rather notable in that he came up with the concepts behind the circulation theory of lift and the vortex-based lifting line theory for wings before Ludwig Prandtl would, (though it was Prandtl who would mathematically verify it) and also contributed a fair bit to the analysis of airplane stability. Unfortunately, he was enough ahead of his time that his contributions weren't recognized until after he died.