[WIP] Infernal Robotics - Next

[Rudolf Meier]

currently I'm testing the editor gui... e.g. does "move to positive" move to positive and such things...

and ok, I see the symmetry problem now too... but it's more than just a graphical problem... internally something is crashing... I will fix that

Edited February 23, 2018 by Rudolf Meier

[Rudolf Meier]

On ‎20‎.‎02‎.‎2018 at 5:40 PM, Ziw said:

Feel free to ask about UI parts, they look intimidating at first, but actually quite simple once you figure them out.

That's something I will do for sure... I'm using them and modified it a little bit (not much, mainly the handling of the drag and drop and... I wanted to change the behaviour of when the window starts open and when closed) ... but, maybe you should take a look at them. I don't have access to the latest changes you made to the gui... I'm right now working on 6 bugs... I will upload new versions when this is done...

[TheKurgan]

This thread... watch intently I will... very interesting it is... mmmmm   

/end Yoda_voice

Edited February 23, 2018 by TheKurgan

[Rudolf Meier]

18 hours ago, AccidentalDisassembly said:

Update on this: I tried it again to figure out what was going wrong, and I am reasonably certain the ghost issue has to do with symmetry.

Yes it is... I found it... when KSP creates an object because of the symmetry settings, then it does not send a "start" or "onEditorAttached" as it would normally, but it does send a "onEditorDetached" event when those objects disappear (e.g. when you move your mouse away from a valid attachment point) ... that's why it tryes to destroy a non created object and ... null reference ... crash ... stuff like that :-) ... pretty strange behaviour of KSP I think

[Rudolf Meier]

I wanted to release something like a "release candidate" today, but now I'm working on a new solution for the uncontrolled joint limits... I forgot one very important rule... sorry for that

(no, it's not the "check your staging" ... it's the "if it doesn't work, add more struts")

 

[Dr_Goddard]

Release or die LOL it's the coder rule.

[Rudolf Meier]

I've fixed everything and the new idea seems to work... I'm now checking some gui stuff...

and you're not correct... the rule is "when it compiles, it's a release!"

 

Edited February 25, 2018 by Rudolf Meier

[Rudolf Meier]

ok... I did upload the latest version... the first true Beta

it contains everything and everything should work

the code needs a cleanup and some functions are not yet optimized in a way I would like it

but... in theory it should run without problems...

[AccidentalDisassembly]

Gave the latest version a shot. I could not reproduce the symmetry issue, which is awesome!

However, there seems to be an issue with scaling. I tried stacking all sizes of an extendatron, and this was the result when extended:

https://www.dropbox.com/s/vfsmake4vfdktvs/NewIRScalingIssue.png?dl=0

I remembered that in the past, IR had issues when scaling parts UP (but not down) past a certain point. Those issues were not exactly similar to what I'm seeing here, but I thought I'd mention it, since it would be really nice if IR parts could successfully function if scaled up really large (via TweakScale, or via IR's own scaling, or however that would work)...

 

EDIT: Second issue (maybe related to KJR, maybe not): if you place something like a rotatron on top of a command pod, then build a helicopter-style structure on top of it (one part going upward, then "arms" going outward), if the arms contain many pieces (I had about 8), the craft will begin flailing wildly a short time after launch. Seems like some kind of force building up somewhere...

If that same structure is placed on to the command pod directly, no flex happens, no flailing. Something seems to be happening when a long-ish chain of parts are placed on an IR piece.

Edited February 25, 2018 by AccidentalDisassembly

[kcs123]

Tweakscaling of IR parts are dealed trough "IR_TweakScale.cfg" file. It also depends on module names "MuMechToggle", "IR_Stack" and "IR_Free", how it will be scaled. Becasue IR Next use different module names it probably created similar issue with tweakscale configs as it is with KJR mod.

Too "springy" craft with long parts are alway being issue on scene loading, this is one of reason why KJR is created in the first place. It is not only restricted to IR parts. As much as I was able to understand trough config files, IR joints strength are already made as strong as possible to overcome such issues. That is one of reason why you would more likely to see that joints of other parts start to drift sooner than joint of IR parts that create force when you start to move IR parts of craft.

@AccidentalDisassembly, you might want to try other mod that help with KSP stock game issues on loading flight scene:

 

 

[Rudolf Meier]

1 hour ago, kcs123 said:

Tweakscaling of IR parts are dealed trough "IR_TweakScale.cfg" file. It also depends on module names "MuMechToggle", "IR_Stack" and "IR_Free", how it will be scaled. Becasue IR Next use different module names it probably created similar issue with tweakscale configs as it is with KJR mod.

Too "springy" craft with long parts are alway being issue on scene loading, this is one of reason why KJR is created in the first place. It is not only restricted to IR parts. As much as I was able to understand trough config files, IR joints strength are already made as strong as possible to overcome such issues. That is one of reason why you would more likely to see that joints of other parts start to drift sooner than joint of IR parts that create force when you start to move IR parts of craft.

Yeah... those names... I made a proposal on how to change KJR so that we don't have those naming issues anymore (I don't have an answer yet). Maybe I should do the same for TweakScale (but I'm not sure if they probably already have such a mechanism... I will check that).

The joints strengths are new defined in IR Next. We would have to figure out if it's good or not. The reason is, that we have a lot more restrictions inside those joints now and because of that I relaxed the forces on the moving axis a little bit (so that for example a heavy loaded rotatron is bended a little bit). But I don't know if those values work.

 

Edited February 25, 2018 by Rudolf Meier

[kcs123]

I think that both, tweakscale and KJR are made in a way that original author of mod that KJR and tweakscale interact with, does not have to bother to change anything in their mod to make it work. Not all of modders are willing to bother to change their code, (in some cases it might be too complicated) to make it work with some other mod that is made later on.

Therefore, KJR and tweakscale offers support for other mods trough config files. I'm not sure if developers of KJR and tweakscale are more confortable with one aproach of solving issue than other one. What they would think that fit better on the long run.

As for new joint strength, it is hard to predict what kind of value would suit of all possible craft application. All that can be fine tuned later on. It might be good to expose those values in config file, to allow users to experiment with it and report back what kind of value suit better for their craft.

I'm not sure if those values from part config file do that or something else is involved:

Quote

breakingForce = 999
breakingTorque = 999

I was hoping that I would have more time on this weekend to help with testing, but hopes and wishes are one thing, but real life always kicks in with something else to deal with.

 

[Rudolf Meier]

26 minutes ago, kcs123 said:

I think that both, tweakscale and KJR are made in a way that original author of mod that KJR and tweakscale interact with, does not have to bother to change anything in their mod to make it work. Not all of modders are willing to bother to change their code, (in some cases it might be too complicated) to make it work with some other mod that is made later on.

Therefore, KJR and tweakscale offers support for other mods trough config files. I'm not sure if developers of KJR and tweakscale are more confortable with one aproach of solving issue than other one. What they would think that fit better on the long run.

TweakScale is good... they do have a good mechanism to support mods. So, no changes needed here... it's just some names that are wrong in the config files I distributed (because they changed) ... KJR on the other side lacks an easy interface for mods. And the change I proposed was 2 lines of code  plus an empty assembly-project.

Edit: TweakScale is even very good... I wouldn't change anything... that's just great. I only need to modify my module just a little bit.

Edited February 25, 2018 by Rudolf Meier

[Rudolf Meier]

57 minutes ago, kcs123 said:

As for new joint strength, it is hard to predict what kind of value would suit of all possible craft application. All that can be fine tuned later on. It might be good to expose those values in config file, to allow users to experiment with it and report back what kind of value suit better for their craft.

I'm not sure if those values from part config file do that or something else is involved

They should define this... I have to check if they also do define it or if something is wrong here.

[Rudolf Meier]

8 hours ago, AccidentalDisassembly said:

Gave the latest version a shot. I could not reproduce the symmetry issue, which is awesome!

However, there seems to be an issue with scaling. I tried stacking all sizes of an extendatron, and this was the result when extended:

https://www.dropbox.com/s/vfsmake4vfdktvs/NewIRScalingIssue.png?dl=0

I remembered that in the past, IR had issues when scaling parts UP (but not down) past a certain point. Those issues were not exactly similar to what I'm seeing here, but I thought I'd mention it, since it would be really nice if IR parts could successfully function if scaled up really large (via TweakScale, or via IR's own scaling, or however that would work)...

 

EDIT: Second issue (maybe related to KJR, maybe not): if you place something like a rotatron on top of a command pod, then build a helicopter-style structure on top of it (one part going upward, then "arms" going outward), if the arms contain many pieces (I had about 8), the craft will begin flailing wildly a short time after launch. Seems like some kind of force building up somewhere...

If that same structure is placed on to the command pod directly, no flex happens, no flailing. Something seems to be happening when a long-ish chain of parts are placed on an IR piece.

Thanks for the feedback... yes, scaling was a bug... that's fixed now (not yet uploaded) but... the breaking of the stuff... can you give me more information on that? I've no such problems... is this only when you spin the rotatron very fast? in this case none of my parts ever stayed on top of that construction... do you have a picture?

[AccidentalDisassembly]

3 hours ago, Rudolf Meier said:

Thanks for the feedback... yes, scaling was a bug... that's fixed now (not yet uploaded) but... the breaking of the stuff... can you give me more information on that? I've no such problems... is this only when you spin the rotatron very fast? in this case none of my parts ever stayed on top of that construction... do you have a picture?

It began shaking violently slowly at first, immediately upon entering the flight scene - wasn't connected to rotation, I never actually moved the part. Sadly I don't have a picture. I had KJR installed, in case that matters.

Try this: put a rotatron (controlled) on top of a stack of stuff, pointing upward. I had one on top of a cupola, which was on top ofseveral batteries and a fuel tank (to act as weights).

On top of the rotatron, stack-node mount one of the robotic truss structural pieces vertically, as if the shaft of a helicopter's blade assembly.

On the side of that vertical piece, use another robo truss to start building horizontal "arms" in 4x symmetry and at 90 degrees / parallel to the ground. These are like the helicopter's blades, if it helps visualize. Surface attach the first part, then continue adding trusses by attaching to the arms' stack nodes until each arm is 8 or 9 parts long. 

Launch the craft - in my case, after launching, the arms wiggled a bit, forces built up, and they started flailing wildly after a short time - I never moved the rotatron. When I took that entire "helicopter blade" assembly off the rotatron and placed it on top of the pod's stack node instead of the rotatron, it was immobile.

Hope that gives you an idea (ish) of what I was doing.. otherwise I'll see if I can reproduce and get a picture some time later.

[Rudolf Meier]

7 minutes ago, AccidentalDisassembly said:

...

Hope that gives you an idea (ish) of what I was doing.. otherwise I'll see if I can reproduce and get a picture some time later.

That's what I did. In 1.3.1 and 1.2.2 with and without KJR. I was even able to spin this at a speed of ... don't know exactly... around 400 to 600. It only broke when I went too fast.

[AccidentalDisassembly]

27 minutes ago, Rudolf Meier said:

That's what I did. In 1.3.1 and 1.2.2 with and without KJR. I was even able to spin this at a speed of ... don't know exactly... around 400 to 600. It only broke when I went too fast.

Huh - happens reliably for me, here's a pic - tried to catch it mid-wobble:

https://www.dropbox.com/s/58t0w9zju9xqn02/IRWobble.png?dl=0

 

[Rudolf Meier]

ok... you made them longer than I did... I will try to make those arms as long as yours and see what happens... I didn't see the "8" and the "9" in your last posting

 

Edited February 25, 2018 by Rudolf Meier

[Rudolf Meier]

ok, this problem does on my computer also happen with the original IR... so, it's not a new one... can you confirm this?

if yes... then it's what we already know... it's a unity problem... if you put some small fuel tanks to the end of those tubes and make them heavier, then this shaking goes away... there's nothing we can do about this... except activating KJR for the servos

but this is a project on its own... adding the same constraints as KJR does on a moving joint...  ... not easy

Edited February 25, 2018 by Rudolf Meier

[AccidentalDisassembly]

3 hours ago, Rudolf Meier said:

ok, this problem does on my computer also happen with the original IR... so, it's not a new one... can you confirm this?

if yes... then it's what we already know... it's a unity problem... if you put some small fuel tanks to the end of those tubes and make them heavier, then this shaking goes away... there's nothing we can do about this... except activating KJR for the servos

but this is a project on its own... adding the same constraints as KJR does on a moving joint...  ... not easy

Gave it a try with the version of IR that I have in my modded game - I think it's the most recent. YEP: exact same thing happens. Tried sticking fuel tanks on the ends of the arms, it didn't solve the issue, but using a more massive fuel tank as the base (the part attached to the rotatron) did.

I wonder - I remember reading that Unity uses part mass to determine joint strength (which makes only a tiny bit of sense, but oh well). Is there a way that IR can convince Unity that the joint is between two very massive parts without that mass actually showing up on the craft, or something? Or is there a way to actively dampen the wiggle using IR?

I'm guessing the answer is no, of course! Wonder if the 1.4 update will help...

Edited February 25, 2018 by AccidentalDisassembly

[Rudolf Meier]

2 minutes ago, AccidentalDisassembly said:

Or is there a way to actively dampen the wiggle using IR?

I'm guessing the answer is no, of course! Wonder if the 1.4 update will help...

Damping... I'm working on it... might be...

The update I guess wouldn't help, if they don't use another unity version... if they would, then ... who knows? I never worked with newer unity's

[Rudolf Meier]

1 hour ago, Rudolf Meier said:

Damping... I'm working on it...

ok... solved...

I like solving "impossible" problems

it was not my bug... it's a unity problem (as we all know) and KJR does solve this problem, but ... it could do it better... I now modified KJR and this solves the problem... it's now stable... and I guess other ships would profit from this modification too

the IR joints are still a problem, but we could try more if it would be needed... for the moment I think it's ok... the problem now... should I upload this modifyed KJR? I don't have the ok at the moment... that's why I'm first contacting the owner...

[kcs123]

38 minutes ago, Rudolf Meier said:

ok... solved...

I like solving "impossible" problems

it was not my bug... it's a unity problem (as we all know) and KJR does solve this problem, but ... it could do it better... I now modified KJR and this solves the problem... it's now stable... and I guess other ships would profit from this modification too

the IR joints are still a problem, but we could try more if it would be needed... for the moment I think it's ok... the problem now... should I upload this modifyed KJR? I don't have the ok at the moment... that's why I'm first contacting the owner...

KJR licence is GNU General Public License v3.0 that allows modification and redustribution as long as you keep same licence. It might be OK to distribute modified version for testing purposese, but inthe long run it will make a confusion in community, what KJR version is required and what not.

It might be better if you create pull request for code change. And just as I was looking at KJR github code, I seee that you already did that. We all have to be patient until ferram got free time to merge this in main release.

Until changes are made from ferram side, you might distribute your forked/changed version of KJR with disclaimer that should only be used temporary for testing purposes. For full release of IR next, it will be better for community to be patient about it.

[AccidentalDisassembly]

Testing with scaling. The latest version continues to have an issue with TweakScale that older versions of IR also had. When parts are scaled DOWN, there is no problem. However, when parts are scaled UP past a certain point (I can't remember what point), this happens:

https://www.dropbox.com/s/8q0luncz3sxvnhk/NewIRScaling2.png?dl=0

Note the different proportion of the extensible part actually gets extended.

Otherwise, scaling is working all right - EC scales like I've told it to, mass seems to scale like I've told it to, but there's an issue with extension in particular. I remember trying to get the original IR devs and the TweakScale devs to care about it, but they didn't. =(

No clue if it's a TweakScale issue or IR or what, sadly, but it's a blast from the past for me.