[WIP] Kerbal Joint Reinforcement - Next

[kurtu5]

Works like a charm.  Thanks for the shiny new charm.

 

[Rudolf Meier]

10 minutes ago, kurtu5 said:

Works like a charm.  Thanks for the shiny new charm.

*phew* that's great... thanks for testing

 

... those screenshots. Do you also instinctively click on them with the right mouse button and try to move the camera?

Edited May 17, 2019 by Rudolf Meier

[kurtu5]

Yes.  If I ever make a game, I think I will add a 3D screenshot feature that uses whatever passes for VRML nowadays.

[kcs123]

4 hours ago, Rudolf Meier said:

I'm now doing what nobody should do... throw out software that's not very well tested... was a great idea

Why not ? SQUAD doing that all time . They do tests, just not well enough.

Anyhow, it is always good idea, regardless of software you develop, to give your product to other people for testing. No matter how effort you made to test for yourself, you will always miss something. Heck, I even found some bugs in my code after several years passed since I last time worked on some pieces of code.

So, no worries, as soon as people notice something they will report and you can have starting point to search for fix.

[Paadwyn]

I just tossed in the 4.0.5 Dev version to test.

Runs smooth, no stuttering.

Vessel switching is good, no more kraken swallowing your ships, no error message.

Still works with Global Construction, crates expand properly.

The only thing I noticed, and I can rectify with design work, is that the joints on some things don't seem to be as rigid. For example, the 5m Toroidal SAS for the starlifter really oscillates. I can fix that by strapping it down. Another satellite tended to wobble more before finally smoothing out. Not severe, just noticeable.

But, I can work around those...the rest, awesome. Working nice.

[Rudolf Meier]

3 minutes ago, Paadwyn said:

The only thing I noticed, and I can rectify with design work, is that the joints on some things don't seem to be as rigid. For example, the 5m Toroidal SAS for the starlifter really oscillates. I can fix that by strapping it down. Another satellite tended to wobble more before finally smoothing out. Not severe, just noticeable.

thanks for the report... these things you mention like the real strength of the joints and an analysis wheter it's too strong or too weak is something that I'm still working on. The rest of the code is now new, remains this one... I'm planning to build some kind of sensor so that we can collect data and see what's wrong and needs improvement. I will inform everyone when those tests can start...

[Paadwyn]

Strapping it down (what you're supposed to do anyways) made a world of difference.

It still has an odd oscillating on the entire craft, needs to perform a constant roll correction, but that's got nothing to do with KJR.

Still, nice to see my beast finally working right. Excellent work, and prompt too.

Thumbs up!

[Rudolf Meier]

3 minutes ago, Paadwyn said:

It still has an odd oscillating on the entire craft, needs to perform a constant roll correction, but that's got nothing to do with KJR.

when accelerating? or just sitting around?

[Paadwyn]

Just sitting idle. SAS slowly rolls the entire vessel side to side like it's overcompensating. Some adjustments to the Wheel Authority help smooth it out.

[Paadwyn]

I think what's happening (After watching it closely) is that it's twisting slightly and the outside nacelles are slightly swaying, and it takes a while for it to finally bleed off the torque. Using more struts would help secure it better.

There could be joints that aren't quite holding tight and twisting slightly.

[kcs123]

I have encontered different issue. Previously working craft created with 4.0.0 and 4.0.3 version of KJRn started to explode when I staged it with KJR 4.0.5. Upper stage craft parts seems to receive a lot of aerodynamic stress that was not happening before. Have yet to inspect to see what is going on. Have to revert back to KJR 4.0. or 4.0.3. and attempt to re-create same craft again from strach with KJR 4.0.5. to narrow down issue. Don't know if I would have time for it today, though.

EDIT:

Yep. Without KJRn craft behave just fine. It is lightweight rocket, though, so issue may not happen on heavier/larger crafts.
I will send craft file soon. Currently downloading DBG version of KJR to see if I can find more about issue.

Edited May 18, 2019 by kcs123

[Rudolf Meier]

1 hour ago, kcs123 said:

I have encontered different issue. Previously working craft created with 4.0.0 and 4.0.3 version of KJRn started to explode when I staged it with KJR 4.0.5. Upper stage craft parts seems to receive a lot of aerodynamic stress that was not happening before. Have yet to inspect to see what is going on. Have to revert back to KJR 4.0. or 4.0.3. and attempt to re-create same craft again from strach with KJR 4.0.5. to narrow down issue. Don't know if I would have time for it today, though.

EDIT:

Yep. Without KJRn craft behave just fine. It is lightweight rocket, though, so issue may not happen on heavier/larger crafts.
I will send craft file soon. Currently downloading DBG version of KJR to see if I can find more about issue.

thanks

I have an idea... but you cannot test this with the current debug version

I will create a new one with two additional options, then you can test this completely without uninstalling KJRn between the tests

and... I will (just now for this debugging) maybe add a third option which would modify the behaviour internally that I think could be the problem

I'd recommend to wait for the "4.0.5 debug plus" (which will only be different in the options of the debug dialog, so, no official new release and only temporary... but... since we all know... I have to upload the code the license and all that crap, because otherwise I'm not allowed to send this version to someone... not even for a 5 minute test... ...)

[kcs123]

No problem, I have send you link for craft file in PM just now. I can be patient for next release, don't have much time to play KSP today anyway.

[Rudolf Meier]

13 hours ago, Paadwyn said:

There could be joints that aren't quite holding tight and twisting slightly.

I think (and also thought yesterday to be honest), that you have the same problem like kcs123 talks about.

21 minutes ago, kcs123 said:

No problem, I have send you link for craft file in PM just now. I can be patient for next release, don't have much time to play KSP today anyway.

and I thought IR development is difficult... but this game engines... in general... wow... incredible

I'm so happy that I can work on something else in my day job (also software development and mostly of type "that can never been done... how would you do this?", but still... here sometimes I wonder if it's possible at all, while in my dayjob we do things many say "you can never solve this! because it cannot be solved at all!")  ... anyway, I will continue and maybe... one day... far in the future... it could be solved... who nows?

Edited May 18, 2019 by Rudolf Meier

[Paadwyn]

Bug swatting, fix one problem and create two more

Modding things like this can be rough because you're restricted on what you can do. Trying to develop something inside a set of rules.

I'm still designing my fleet, so I'm not really progressing anyways. I can most certainly help get this solved in the meantime.

[Rudolf Meier]

it is not KJR itself... it's just the AutoStrut-Cycling-Detection I added for DLC compatibility... this one is a problem... it's a part that does have NO physics and still affects it... great, hm? ... I've no idea how to solve this... really none at the moment... for hours I tried everything... that's just stupid...

[Paadwyn]

I don't know how it all affects one another, like I said, I don't know code.

But, it's basically an invisible part that gets added, has no physics. Where does it get added? Does it get added to the snap points? Maybe a non-physic part has no joint strength.

I really have no clue, I'm just tossing out ideas.

[Rudolf Meier]

the easiest way to solve that would be if Squad would add a GameEvent to indicate an autostrut cycling

the only thing I know is, that the new DLC will also bring a ne KSP version... together with robotics the current autostrut solution would never work... cannot, it's impossible... but, I don't think they would add such an event because of that

I think they're not very much interested in mods

anyway... I think, when I cannot find a solution for this, then I simply stop here...

[Rudolf Meier]

let's see... 4.0.6 is online

it's... a constant trial and error... this combination here seems to be more stable in my tests than the others... why? no idea... it shouldn't work at all like this... but hey... it's Unity and KSP ... who knows how it works??

[Paadwyn]

I will check it out when I get home later.

Unlike you, I don't get to play KSP at work

Though I could probably get away with it...

[Paadwyn]

Ok, test results show a massive improvement.

On the one vessel with the toroidal SAS unsupported, it didn't move at all, not like before where it was shaking pretty violently.

The next vessel that was slowly rolling back and forth, it still did, but it bled off the torque much much faster. Definitely workable.

I checked one of my other rockets that are floating around, it wobbled a bit when the SAS is turned on, but only for a half second before stabilizing and everything is good.

I haven't checked anything else out yet, but so far, that update really helped me. I haven't checked out staging yet.

[Paadwyn]

I just tried a few launches to see how it handled, seems to do very well. Even on one of my more...complicated launches. It behaved pretty close to how it did before.

Unless anybody else has issues, I'd say you've nailed it.

[kcs123]

My test rocket with issues behave much better with KJRn 4.0.6. No explosions this time, but still have some kraken forces on staging where you don't expect any. Observing behaviour, it seems that on staging KJRn release too late joins between stages. At least I got feeling that it is happening from in game observation.

Second issue with debug verion is that clicking on KJR toolbar button produce nullreference object exception.
I tried to catch all issues trough screenshots in this album.

EDIT:

Have to be fair to say, that second stage after separation is slightly unstable without KJRn installed too. But, it only deviate heading from prograde just for tiny amount and quickly put rocket back on track. I might be able to solve this by altering ascent path.

Btw, just idea, if it is possible trough available KSP API, to force postponing staging just for one physical tick(frame) or two ?
Meaning, when player execute stage command (trough manual key staging or from any other autopilot), first to release KJR joints between craft parts on different stages, wait for one or two physical ticks and then execute staging command. That 1-2 frame will not be noticable to player, but might help with solving unwanted kraken forces.

Edited May 19, 2019 by kcs123

[Rudolf Meier]

ok, I tried something... see version 4.0.7

it should now decouple faster (and I hope my new algorithm works that I've added to detect which joints need to be destroyed) ... and, the non-existing mass-less objects are now oriented along the x axis of the root part to even improve their physical influence (which they shouldn't have at all... but... hey... who does understand those game engines??)

[dave1904]

I am having an issue with the fasa launch tower completely breaking the craft on the launchpad when staging. I downloaded 4.0.7 and tested it again and the same thing happens. Its a space shuttle build I have used for a long time now in many versions of KSP and KJR.  I will remove the launch tower for now from the craft since the launch clamps seem to be working and see if I find anymore staging issues. I am thinking of disabling launch clamps in KJR config if thats even possible and test it again. I was never a fan of how even the old versions of KJR dealt with launch clamps. What do you think about disabling lauchclamps atleast for me personally?

Edited May 19, 2019 by dave1904