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

[rrkss]

Bug reproduction steps:

1. Load KSP with realism overhaul MOD and start career in sandbox mode.

2. Go to the VAB and load my Saturn V Spacecraft.

3. Ensure that the root part is the MK 1-2 Command Pod (press 4 and click on the part below it, then click on the command pod)

4. Ensure that the crew are in the command pod in the VAB (3 crew in the command pod, 0 in the lander).

5. Click the launch craft button.

6. Craft breaks up on launch pad as soon as KJR 3.0.1 stabilizes the physics.

The craft file for the space craft can be downloaded at https://www.dropbox.com/s/v46mtenr5s...20V.craft?dl=0.

A picture of my GameData Folder https://www.dropbox.com/s/jdvoieyckftr06i/KSP.JPG?dl=0

A picture listing all my installed Mods and their version numbers can be found at: https://www.dropbox.com/s/ho6g2blidta67c3/KSP%20MODS.JPG?dl=0

My output_log.txt file can be downloaded at: https://www.dropbox.com/s/cdklh7pko4q9x25/output_log.txt?dl=0

As a side note, if I revert back to KJR 2.4.5 the craft does not break up on the launch pad and can fly normally but I have to make the F1 engine cluster on the first stage the root part otherwise the command pod wobbles excessively during launch and the crew dies. If I make the same engine cluster the root part in 3.0.1, its a 50/50 shot of the bug occurring while a 100% chance if the command pod is the root part.

[steve_v]

Meh, I went through a big debugging session with this not long ago (and there are my logs a few posts back). Maybe later I'll look into it. If it works for you, disregard.

[ferram4]

...I appreciate all the effort going into the mod lists, but given how simple the reproduction steps are, either something is being missed or it is a mod interaction. If it's a mod interaction, you need to simplify it down to only the mods necessary to cause the issue, and no others, because the more mods are involved, the more complicated the troubleshooting becomes. No, it's not much work, and you guys get the better end of it, because you don't have to try and fix the issue, just tell me the actual causes of it.

[lyndonguitar]

KJR still sticks together decoupled parts if the decoupling force isn't too great. One example is decoupling a docking port from another non-docking port. I need to use RCS thrusters or engines on both the vessels in order to make them separate, as you're trying to snap a cable

[ferram4]

Try the dev build. That's already been fixed.

[eddiew]

Hi Ferram,

I noticed your last GitHub commit is a fix for an undocking issue. Does this by any chance involve MK2 clamp-o-trons (the spaceplane ones that open and extrude before docking) never letting go after a scene switch away from the active vessel...?

Thanks!

[ferram4]

Why not try the dll and find out? More data points are always helpful when dealing with bugs.

[eddiew]

I actually did, but I didn't have a ship in the right place to test it right off - so I was lazy and asked

But just for you I launched a guinea pig and... well, tbh, if that was what the KJR update was meant to fix, it hasn't worked for me ^^; Of course, if that wasn't a bug you were tracking, it probably wasn't the right test anyway

[ferram4]

Very interesting, because it fixed the standard shielded clamp-o-trons perfectly.

[Starwaster]

eddiew, make absolutely sure that you properly downloaded and copied the the dll from the github site.

It's this one right here: https://github.com/ferram4/Kerbal-Joint-Reinforcement/blob/master/GameData/KerbalJointReinforcement/Plugin/KerbalJointReinforcement.dll

Delete the old one first to eliminate any chance of accidentally not copying over it. (even I made that mistake just the other day testing to see if 3.0.2 fixed a problem with the SDHI service module getting stuck on decoupling)

(Spoiler alert: It does.)

[Blipser]

I havent checked if this has already been reported, but Im having a bug with this in RSS, whenever I drop my first stage on very heavy launchers the procedural fairings base has a structural failure from the procedural tank of the second stage, even though almost no forces are actually applied. I'm using 0.9 and it happens with both compatible versions of this mod, and I assume this is something to do with the fairings base, which weighs about 20 tonnes, having appropriately large nodes, while the tank which weighs about 1000 tonnes has very small connections.

[eddiew]

Very interesting, because it fixed the standard shielded clamp-o-trons perfectly.

Actually the shielded ones worked fine for me in KJR 3.0.0 - I was using them a few days ago to test scenarios and work out why my spaceplane was bolted to my station.

It's this one right here: https://github.com/ferram4/Kerbal-Joint-Reinforcement/blob/master/GameData/KerbalJointReinforcement/Plugin/KerbalJointReinforcement.dll

Yep, that's the one I got.

Though it may actually be another mod giving grief here in this particular case. I have issues with MK2 clamps when they're attached to a station with a Station Science lab in them. Attaching them to anything else (e.g. a probe/tugboat) didn't give any issues. The only reason I queried my MK2 issue here was because I noticed Ferram's commit message

Think I need me a second game instance with HyperEdit and minimal mods. Would make finding out what's not working with what a bit easier...

[Traches]

I'm having the "big rockets fall apart on the pad" bug as well. I've simplified it down as far as I can; if it turns out this is a problem with my installation or something I've done wrong then I'm terribly sorry for wasting your time!

The only mods I need to cause the bug are KJR and procedural parts. Here's a video demonstrating it:

(Here's the craft file from the vid.)

Text version: With KJR and procedural parts installed, stack 2 mondohuge tanks (I use 10m diameter and ~15m long, which is around 1000 tons) on top of each other, add some clamps to the bottom, and launch. Usually it holds together through the first physics load (sometimes it doesn't), but if you go into timewarp and back it falls apart. "Unbreakable joints" in the cheat menu predictably prevents this (which I've been using as a workaround for my RO save).

Here's the log file for that KSP session (I quit out immediately after I stopped recording), and here's a screenshot of my gamedata folder.

Ferram, I've got over a thousand hours in this game and without your mods it wouldn't be a tenth of that. I really appreciate all the work you do, thanks.

[ferram4]

@Blipser: KJR doesn't care about node sizes, because experience has shown part packs don't have the nodes sized right. It uses the mesh for joint strengths.

@eddiew: Welp, then I don't know what to tell you, because there's nothing to trigger a disconnect in the new multipart joints for docking ports in 3.0.0.

@Traches: Okay, I can reproduce it. Thanks, no one else mentioned that gigantic mass ratios were necessary for it to be caused; it's like they didn't even care enough to provide reproduction steps.

I think what's happening is something's breaking in KJR's attempt to stabilize physics loading, as well as the issues of launch clamps and floating point errors with those. I'll see what I can do, but I have no idea if I can actually fix any of it. The more and more I look at things, the more apparent it is that the stock attempt at fixing joints just got in KJR's way.

[Virtualgenius]

Tell them to get with program KJR way or no way

[biohazard15]

Just got a weird bug with vessel (unmanned probe) while on Mun orbit. When I've loaded save, it stopped in orbit (50x50km, around 500m\s speed) right after "KJR is doing its stuff" message - just like old "zero orbital speed" bug introduced in 0.23.5. Interesting, but if I was quick enough to activate time warp, it continued to fly normally - right until I go back to 1x. Reverting back to 2.4.5 fixed it.

Sadly, but I haven't save or log (these became overwritten when I've tried to fix it). Probe used parts from RemoteTech, Tantares and Universal Storage, all latest, 0.90-compatible versions, it also has Mechjeb. Nothing unusual in the logs.

[Traches]

@Traches: Okay, I can reproduce it. Thanks, no one else mentioned that gigantic mass ratios were necessary for it to be caused; it's like they didn't even care enough to provide reproduction steps.

I think what's happening is something's breaking in KJR's attempt to stabilize physics loading, as well as the issues of launch clamps and floating point errors with those. I'll see what I can do, but I have no idea if I can actually fix any of it. The more and more I look at things, the more apparent it is that the stock attempt at fixing joints just got in KJR's way.

Heh, as part of my troubleshooting I uninstalled KJR while leaving the rest of RO installed and tried to launch my Saturn V approximation. THAT was something to see-- who makes a rocket game with floppy rockets? Think about that game design decision for a second-- struts are massless and dragless so there is no gameplay related reason not to use as many as you want, and anything heavy will require a lot of them, but the physics engine crumbles under more than a couple hundred parts. Their bad game design compounds the pain from their poorly optimized code.

From here it seems like they doesn't do much to help the modders who are mostly trying to fix their broken & incomplete game. KSP as a whole is an awesome and wonderfully original concept, but the execution is astoundingly poor.

Anyway since it's only really a problem when timewarping rockets with great personalities to their launch windows, the unbreakable joints cheat is a pretty good workaround.

[ModZero]

Just asking if this is known before I start tracking an issue: with the SDHI service module, the decouple action (it's the animated decoupler thingie) doesn't decouple if there are more parts on the SM: the parts do behave like another ship (no right click), but stay attached. This is with yesterday's git KJR, goes away without KJR, but I didn't do more tracking (yet).

Is that something new?

[Starwaster]

Just asking if this is known before I start tracking an issue: with the SDHI service module, the decouple action (it's the animated decoupler thingie) doesn't decouple if there are more parts on the SM: the parts do behave like another ship (no right click), but stay attached. This is with yesterday's git KJR, goes away without KJR, but I didn't do more tracking (yet).

Is that something new?

Yesterday? Last one I see is four days ago. Where do you see one from yesterday?

Anyway, I tested Animated Decoupler with the latest KJR on the Github repository and it decouples just fine. No longer able to reproduce the issue.

[ModZero]

Downloaded yesterday, yes, it was actually 4 days old. I'll try to reproduce the issue more precisely later then.

[ModZero]

Okay, so this is a minimal install: SDHI, its dependencies (not the optional ones), and KJR. These two work: http://i.imgur.com/7LNMy3n.png, http://i.imgur.com/xErkyMn.png (second one has an engine on the bottom). This one doesn't: http://i.imgur.com/rWDrSdI.png (like third one, but with random stuff radially attached). There is nothing particularly interesting in the log. I'm using 3.0.1 with the dll replaced with one pulled from git.

EDIT: GAH, I hate picture embedding :-(

[Starwaster]

Okay, so this is a minimal install: SDHI, its dependencies (not the optional ones), and KJR. These two work: http://i.imgur.com/7LNMy3n.png, http://i.imgur.com/xErkyMn.png (second one has an engine on the bottom). This one doesn't: http://i.imgur.com/rWDrSdI.png (like third one, but with random stuff radially attached). There is nothing particularly interesting in the log. I'm using 3.0.1 with the dll replaced with one pulled from git.

EDIT: GAH, I hate picture embedding :-(

Forgive me if this is a stupid question, but you are decoupling that by right clicking or by action group right?

[ModZero]

Forgive me if this is a stupid question, but you are decoupling that by right clicking or by action group right?

Yes - right clicking. I think I tried the staging thing yesterday, but it shouldn't do anything with SDHI anyway. Should have mentioned that, it's one way the SM is different from most decouplers.

[Virtualgenius]

@ModZero if you upload the image to imgur and select the the fifth entry down called BBCode (message boards & forums) and copy that link into your forum post that will embed the pic for you

[ModZero]

@ModZero if you upload the image to imgur and select the the fifth entry down called BBCode (message boards & forums) and copy that link into your forum post that will embed the pic for you

Yeah, my edit removed embedding, b/c I hated how much space on the page it took up.