[1.3] Kerbal Joint Reinforcement v3.3.3 7/24/17

[Deltac]

This seems true for most everything, but 5m parts will still have difficulties without further stiffening, no?

*reads Mr. woot's post, and has a smile. Presses the "quote" button, and then cracks knuckles to have fresh typing fingers. Looks out the window with another smile and forms the post in his head.*

Struts.

If that doesn't work, you're doing something wrong. End of discussion.

[woot]

*reads Mr. woot's post, and has a smile. Presses the "quote" button, and then cracks knuckles to have fresh typing fingers. Looks out the window with another smile and forms the post in his head.*

Struts.

If that doesn't work, you're doing something wrong. End of discussion.

Obviously that works, but it remains an inferior solution for those of us who like our rockets to be as aerodynamic and low-part count as possible.

I managed to get one of my big rockets into orbit, but it was definitely a lot more wobbly than before (even the size-3 sections). If there's simply no way around it, then it's not a huge deal. But the standard strut-heavy style was never appealing to me.

[pantheis]

Obviously that works, but it remains an inferior solution for those of us who like our rockets to be as aerodynamic and low-part count as possible.

I managed to get one of my big rockets into orbit, but it was definitely a lot more wobbly than before (even the size-3 sections). If there's simply no way around it, then it's not a huge deal. But the standard strut-heavy style was never appealing to me.

I tried to load up my existing save in 0.23.5, using updated mods and the old config for KJR from 0.23, and ships already on missions (one around Moho, one on its way to Dres), which were straight as an arrow in 0.23, had their side boosters flapping around and some bending in the middle now. This was a mixture of KW and KSPI parts mostly, with a bit of stock parts thrown in. Note the side boosters are not attached with decouplers, but just radially attached. It should be a rock solid weld.

For me, I'll pass on 0.23.5 until/unless these things can be fixed. The thing I loved about KJR was it making welds rock solid, and still needing an occasional strut for large heavy decoupler attached parts. That's all missing in 0.23.5.

[WindShieIds]

Problem: Doesn't seem to work with the new update, I got the patch but my rockets still wobble.

[ferram4]

Using mod parts that don't conform to stock standards (i.e. nodes are not sized properly, low-mass parts like decouplers are not made physics insignificant)? If so, then take it up with the mod author, not me; it's out of my control at this point. Stock parts do not wobble any more than they are intended to and are highly rigid with KJR; if the mod parts are built to the same standards, no wobble will occur.

[biohazard15]

Ferram, may I ask you a question: how does Unity handle joints in case of "one model, many resizes via part.cfg" parts? Does Unity stiff them according to node size (which it should do, according to common logic)? Or does Unity takes this info from original model and just ignores new settings (may be a bug in Unity, actually)?

There is a problem with SAS rings from NovaPunch, and they are perfect example - there is base 1.25m ring model, and three resize configs to 2.5, 3,75 and 5m. All nodes are sized correctly, part weight are approx. the same as stock decouplers.

Thing is: if you put this SAS as radiator under 1st stage engine, it starts inclining (see pictures here). Interesting, but stock 2.5m SAS (weight 0.2, size 1 nodes) does not suffer from this problem, new stock 3.75m decoupler (0.8t) also flies fine.

[ferram4]

Unity doesn't do anything here. KSP does all of that. And honestly, I don't know what shenanigans are exactly going on in the new joint code, but my guess would be that the lower-profile SAS is the cause of the issue; that reduces the inertia tensor of the part, which will make it more susceptible to wobbling. Try making the SAS thicker and see if that fixes the issue.

[ANWRocketMan]

@ferram4: Sorry to bother, but what are the guidelines for proper node sizes?

[ferram4]

Node size = part diameter / 1.25m scaling factor. Round to integers.

[CaptainKipard]

Is there any benefit from nodes larger than 3?

[Galane]

Node size = part diameter / 1.25m scaling factor. Round to integers.

Any chance of an option/function to look for and correct improper node sizes? There was a mod which tweaked node position to get rid of visible gaps between parts.

'Course any such node tweaking would need a way to exclude adjusting parts for which the "proper" node size causes problems.

[ferram4]

@Cpt. Kipard: If there isn't, then that's quite short-sighted on Squad's part. I hope that's not the case. I suppose some testing can be done to see if it really matters.

@Galane: That's probably worth trying. I'll have to see when I have to do that to make sure that the parts are affected properly, since I'd rather do it through code than through ModuleManager.

[Yarbrough08]

Can the fix for struts zeroing out jettison/decoupler forces be re-enabled? I did not see anything about it in the read-me or config..

[MrHappyFace]

Is this mod still going to be updated?

[ferram4]

@Yarbrough08: It's part of the extra decoupler stiffening. Enabling that should bring it back.

@MrHappyFace: Yes, it will. As I've answered quite a few times in this thread.

[Yarbrough08]

@ferram4: Thank you for the reply and the outstanding mods. It is unfortunate that this bug is still around.

[InfiniteDice]

Is there any benefit from nodes larger than 3?

There is no benefit from node sizes at all as far as I know. Node size balls are just indicators. This is why you can attach a size 1 to a size 3. It could be that squad put them there to differentiate between parts that mesh well together, as having parts with too much variation attach to one another causes them sip around something silly.

Furthermore I did extensive tests when I made the first BoatParts, didn't see any evidence that a larger node did anything at all.

All I know is that there is something seriously different/wrong with the new code in ARM. All my previously strong crafts are pulling themselves apart. I'm getting fluctuation between parts which are causing aircraft on the deck to actually fling off the deck

I'll have to look for a way to roll back those changes they made.

Edited April 8, 2014 by InfiniteDice

[GavinZac]

There is no benefit from node sizes at all as far as I know. Node size balls are just indicators. This is why you can attach a size 1 to a size 3. It could be that squad put them there to differentiate between parts that mesh well together, as having parts with too much variation attach to one another causes them sip around something silly.

Furthermore I did extensive tests when I made the first BoatParts, didn't see any evidence that a larger node did anything at all.

All I know is that there is something seriously different/wrong with the new code in ARM. All my previously strong crafts are pulling themselves apart. I'm getting fluctuation between parts which are causing aircraft on the deck to actually fling off the deck

I'll have to look for a way to roll back those changes they made.

Node size determines the strength of connection between parts. Two 3m parts of the same weight as two 1.25m parts will flex a lot less when connected by appropriately sized nodes.

[Starwaster]

There is no benefit from node sizes at all as far as I know. Node size balls are just indicators. This is why you can attach a size 1 to a size 3. It could be that squad put them there to differentiate between parts that mesh well together, as having parts with too much variation attach to one another causes them sip around something silly.

Furthermore I did extensive tests when I made the first BoatParts, didn't see any evidence that a larger node did anything at all.

All I know is that there is something seriously different/wrong with the new code in ARM. All my previously strong crafts are pulling themselves apart. I'm getting fluctuation between parts which are causing aircraft on the deck to actually fling off the deck

I'll have to look for a way to roll back those changes they made.

As of 0.23.5 this is no longer true. Larger nodes mean more connection joints and strength. KJR has also made use of node size as a determining factor.

[coldblade2000]

0.23.5 isn't as good as I hoped. I still get noodle rocket syndrome. What settings should I change in KJR if I'm going to use huge rockets for RealSolarSystem mod

[illectro]

So, I'm having to install and uninstall KJR a lot right now - here's a couple of videos showing the symptoms:

I'm still going to be stuck on 0.23 until Infernal Robotics is fixed, so I don't expect changes to be backported or anything, but maybe you know of a magic config tweak that will stop the panel oscillations.

[ferram4]

@coldblade2000: I've got something in the works, but there's not much you can do for now. Just make sure that the parts you're using have the appropriate node sizes, which helps a lot.

@illectro: Dammit Scott, stop breaking my mods!

Alright, the first one, I have no idea what you managed to do there, but it's quite entertaining. I hope you saved a copy of the output_log.txt from that one, since I don't know what broke and I can't even guess what went wrong; I've never seen that myself. I'm guessing that something interfered with whatever code was used for creating the vessel that held the KerbalEVA, or else the KerbalEVA broke on load, but I dunno what would cause that. Or else something went wrong determining the initial orbit, since that was messed up as well.

Edit: Derp; KJR v2.2 is not back-compatible with KSP 0.23. Make sure you're only using KJR v2.1 with KSP 0.23.

The second one is much simpler: you've got a nice limit cycle going on, likely caused by the spring and damping constants for the parts being a little too high (creating the initial instability) but the angular velocity limit preventing the forces from growing past a certain point. So the thing to do there is to head into the config.xml and look at the "angular drive" parameters. Reducing the spring constant there should help with reducing the initial disturbance, and reducing or increasing the damper constant should affect the stability of everything; the effect really depends on how the frequency and physics timestep play with each other. The constants are currently ridiculously high to get around the angular velocity limit, so reducing both should help, though it will make a lot of things a bit more wobbly. Try dropping the spring constant one or two orders of magnitude and see if that helps.

Edited April 12, 2014 by ferram4

[Starwaster]

@coldblade2000:

Alright, the first one, I have no idea what you managed to do there, but it's quite entertaining. I hope you saved a copy of the output_log.txt from that one, since I don't know what broke and I can't even guess what went wrong; I've never seen that myself. I'm guessing that something interfered with whatever code was used for creating the vessel that held the KerbalEVA, or else the KerbalEVA broke on load, but I dunno what would cause that. Or else something went wrong determining the initial orbit, since that was messed up as well.

Coincidentally enough I just ran into the KerbalEVA thing myself. At first I thought there might have been an out of date KJR in that install but unfortunately it's the latest one.

output_log.txt here

[ferram4]

Alright, then I'm going to need detailed reproduction instructions. There's nothing in there that indicates the error starts with KJR; it may be a stock bug for all I know, but the first error that occurs with anything involving the EVA is trying to get the rigidbody's velocity (called by the regular game, so there's nothing I can do), which ends up nulling out somehow. So either it's breaking right before that or it's a stock bug.

[Starwaster]

Alright, then I'm going to need detailed reproduction instructions. There's nothing in there that indicates the error starts with KJR; it may be a stock bug for all I know, but the first error that occurs with anything involving the EVA is trying to get the rigidbody's velocity (called by the regular game, so there's nothing I can do), which ends up nulling out somehow. So either it's breaking right before that or it's a stock bug.

Try any of the following to reproduce:

1. Go EVA then F5 to save. Then F9
   
2. If that doesn't work, EVA as before, F5. Then kill the Kerbal. Then F9
   
3. If that doesn't work... hmmmm. Will try to reproduce without KJR but it's also RSS so I'm not sure I can get by without it.