Return of Physics-Less Behavior

[Geschosskopf]

I'm seeing a return of an old problem with the TR-38-D size3 stack decoupler that first appeared way back in 0.23.5.  The old problem was that if you staged this decoupler while your ship had an orbital velocity of less than about 800m/s (and the parts on either side of the decoupler had enough mass), the dropped stage became physics-less.  Which meant it essentially stopped in place.  If the ship is pointing in any direction other than perfectly prograde, the upper stage would slam into the stage just dropped at orbital velocity, causing ruin.

Back in those ancient days, parts had a "PhysicsSignificance" line in their cfg files.  If this was 0, physics applied.  If this was 1, then the part was physics-less.  The TR-38-D originally had this set to 1, which is what was causing the disasters.  So we had to change that to 0, which solved the problem.  These days, supposedly there are no physics-less parts, and the "PhysicsSignificance" line has been removed from part cfgs. 

HOWEVER.....  I'm getting this exact same problem again today in 1.3.1.  In a clean stock installation (except it has HyperEdit to speed things up), if I stage the TR-38-D in a 100km Munar orbit (velocity 466m/s < 800m/s), the lower stage stops in place and the upper stage slams into it.  And there's no easy fix because there's no longer a PhysicsSignificance value to change.

I asked about this over in the Gameplay Questions topic:

I also posted the test ship I was using, which always has this problem:

https://www.dropbox.com/s/q2zssq8ap50rmai/Test Krakensbane.craft?dl=0

 

And here's the save folder from my stock test game.  This contains both the ship above AND 

https://www.dropbox.com/sh/kq9t8vf3y8nk5v0/AABeT2D7fqMBPdw12pzfSwJla?dl=0

 

Some folks have tried to replicate my problem without success.  This leads me to believe that something got hosed up on my end in the game.  Perhaps some game-level setting?  I've looked for such a thing but nothing stood out.  Any suggestions?

Thanks.

[OrbitalBuzzsaw]

Odd.

[OHara]

5 hours ago, Geschosskopf said:

In a clean stock installation (except it has HyperEdit to speed things up),

That's the problem.  I cannot reproduce in stock, I can reproduce with Hyperedit.  That is, decoupled parts hover stationary to the parent body, assuming that is what you mean by 'krakensbane'.

Hyperedit recently had a bug appear involving coordinates not being transferred correctly to the game engine.  Maybe Hyperdit's 'set orbit' feature is putting the craft in a state that is inconsistent according to the rules of version 1.3.1.

By the way, you can set orbits now with the Alt-F12 menu.  You can  check craft mass in the VAB  engineer's report and 'i' in the map view.

Edited November 19, 2017 by OHara
Only difference I see is that Hyperedit leaves the 'landedAt' set to 'Launch pad' where Alt-F12 set-orbit clears that. After using Hyperedit and decoupling, the debris has LANDED=true, which matches the symptom.

[swjr-swis]

1 hour ago, OHara said:

Only difference I see is that Hyperedit leaves the 'landedAt' set to 'Launch pad' where Alt-F12 set-orbit clears that. After using Hyperedit and decoupling, the debris has LANDED=true, which matches the symptom.

Not just a tiny difference, this here is key in this problem. This needs to be reported in the HyperEdit thread.

[Geschosskopf]

4 hours ago, OHara said:

Hyperedit recently had a bug appear involving coordinates not being transferred correctly to the game engine.  Maybe Hyperdit's 'set orbit' feature is putting the craft in a state that is inconsistent according to the rules of version 1.3.1.

Perhaps....  Come to think of it, I initially encountered this bug in a heavily modded install when I was trying to test a new ship design and used HyperEdit to put it in orbit.

HOWEVER, some folks who've tried to replicate my problem unsuccessfully have also used HyperEdit to get the stock test ship into Munar orbit, and Krakensbane didn't kick in.

Edited November 20, 2017 by Geschosskopf

[OHara]

6 minutes ago, Geschosskopf said:

HOWEVER, some folks who've tried to replicate my problem unsuccessfully have also used HyperEdit to get the stock test ship into Munar orbit, and Krakensbane didn't kick in.

In the thread you linked above, @Plusck says he did not use Hyperedit, and @bewing would probably use the built-in set-orbit instead.  Try yourself without Hyperedit.

[bewing]

On my part, that's correct -- SetOrbit.

[Plusck]

9 hours ago, OHara said:

In the thread you linked above, @Plusck says he did not use Hyperedit, and @bewing would probably use the built-in set-orbit instead.  Try yourself without Hyperedit.

Thanks for picking up on that. My phrasing sometimes could do with being clearer, since I did say the only mod I have is KER and I used the stock cheat to set the orbit, but it got a bit lost in the other things I was saying

So to be clear: I don't have HyperEdit installed and haven't for the last couple of versions of KSP. Last install was fresh 1.3.1 + latest KER + custom flag (I like flying the Saltire).

[Geschosskopf]

OK, thanks for figuring this out.  But wow, it sure looked like the old Krakensbane back from the grave