[RESOLVED/FIXED] Question about 1.1.3 orbit bugfix

[p1t1o]

**EDIT**

Well you don't hardly ever get to do this!:

FIXEDFIXEDFIXEDFIXEDFIXEDFIXEDFIXEDFIXEDFIXEDFIXEDFIXEDFIXEDFIXEDFIXEDFIXED

As of version 1.2 pre-release:

Officially reported as "Closed" on the bug tracker, multiple reports that the orbital stability issue no longer exists!

FIXEDFIXEDFIXEDFIXEDFIXEDFIXEDFIXEDFIXEDFIXEDFIXEDFIXEDFIXEDFIXEDFIXEDFIXED

Congrats to all at Squad, esp. @NathanKell who appears to have taken the lead on the issue.

Thanks to all that took time to report anything to this thread.

 

 

 

**EDIT**

IF YOU ARE STILL EXPERIENCING THIS ISSUE, IN WHATEVER BUILD OF KSP IS CURRENT AT THE TIME, LOG IT ON THE BUG TRACKER

http://bugs.kerbalspaceprogram.com/projects/ksp

Page for [v.1.2 pre-release] issues: http://bugs.kerbalspaceprogram.com/projects/prerelease/issues

**EDIT**

 

Hi all,

first the usual congrats to squad for providing us with another update, thanks guys!

A few quick questions jump to mind though:

re:

Greatly reduce Apoapsis/Periapsis changing with no input, with thanks to ferram4 and eggrobin. option is toggleable in Settings->Gameplay and tunable in Physics.cfg.

What does "greatly reduce" mean?

Toggleable? Why would one want to switch this off?

Being tunable in the cfg implies it isn't necessarily already optimal, how does one manipulate the cfg to make it optimal? 

Is it tweakable so that people can simulate decaying orbits if they so wish?

Edited September 27, 2016 by p1t1o

[magnemoe]

Interested in this too, have problems with it on huge ships who seems to get an trust up to 10 m/s on load for something 30 meter long and 300 ton with most of the weight on one end. 

[Superfluous J]

I would assume it's lack of hubris on the dev's part. "I made this. I think it's rock solid. But I thought that before and was wrong so I've opened up the options so others can try to make it better."

Considering the last fix to orbits was the single biggest bug in 1.1.2 (my opinion, yours may vary) I think this is a great idea.

[p1t1o]

1 hour ago, 5thHorseman said:

I would assume it's lack of hubris on the dev's part. "I made this. I think it's rock solid. But I thought that before and was wrong so I've opened up the options so others can try to make it better."

Considering the last fix to orbits was the single biggest bug in 1.1.2 (my opinion, yours may vary) I think this is a great idea.

Mmmmm, I hope thats not the explanation, it seems like something that is either fixed or not, and when it comes to the stability of orbits in a space sim, it really shouldn't be a "maybe".

I mean, it sounds fixed, and the option to toggle/tweak isn't going to do any harm, but it jumped off the page as "odd" to me. Hey-ho!

[Cairol]

There is some Squadcast where the devs talk about 1.1.3 and especially this bug fix (minute 36).
They pointed out that the orbital decay is practically gone, except for really really high orbits. (like for a sun dive, they said)
That's why they said it's only been "greatly reduced".

[Majorjim!]

1 hour ago, Cairol said:

There is some Squadcast where the devs talk about 1.1.3 and especially this bug fix (minute 36).
They pointed out that the orbital decay is practically gone, except for really really high orbits. (like for a sun dive, they said)
That's why they said it's only been "greatly reduced".

Yeah but why have the option to turn off the fix? It makes no sense whatsoever. Also, many have reported that decaying orbits are still a thing so this all seems very fishy to me. Can one of the devs shed some light on this please?

[Anth]

From one of the KSPTV twitch videos...NathanKell seemed to indicate that he fixed it by when there was no force applied from the ship he put it onto rails. plus there is some movement at times which self corrects.

I would assume that theres a mess of code at the moment and they couldnt find the real cause so this is a way of bandaging it until they find it. They will be cleaning up a lot of the code for 1.2, that might allow them to figure out the real reason for the deviation.

[AlphaMensae]

 

Over on the 1.1.3 Release devblog thread, I replied to this same question (from someone else), and @NathanKell followed up:

On 6/23/2016 at 9:58 AM, Laguna said:

@Arsonide was in EJ_SA's Twitch stream yesterday, and I asked him this very question.  He actually had to run it past @NathanKell, and the explanation was that due to the fix tweaking things deep inside KSP, it was made toggleable in case anything was missed during testing.

In other words, if the orbit decay fix ends up breaking something else in a worse way, it can be turned off in a case of "pick your poison".

On 6/23/2016 at 10:37 PM, NathanKell said:

Technically, running with the option off is exactly equivalent to how things were in 1.0.5 and 1.1.0. And the drift compensation disables itself automatically at a certain threshold of error (i.e. when it would make things worse). But juuuuuust in case, it's fully toggleable too.

 

Edited June 26, 2016 by Laguna

[NathanKell]

It's also worth pointing out that a fix that works great in the Kerbol system might screw things up in OPM or 64K or RSS. And I would prefer not to make modders' lives harder, as I hope you appreciate. The rest is as @5thHorseman says (and the second- and first-person quotes of me). Configurability is good, and is also a safety net.

[kBob]

2 hours ago, NathanKell said:

Configurability is good

An understatement.  I was so happy when I came back to KSP after a forced hiatus and found a lot of the new things like reentry heat had been made user configurable: so nice, I love choice (and now I just need an extended life time to delve into all the different options and nuances).

[Majorjim!]

5 hours ago, NathanKell said:

It's also worth pointing out that a fix that works great in the Kerbol system might screw things up in OPM or 64K or RSS. And I would prefer not to make modders' lives harder, as I hope you appreciate. The rest is as @5thHorseman says (and the second- and first-person quotes of me). Configurability is good, and is also a safety net.

Thanks for the clarity NathanKell. Is it possible for the fix to cause issues in stock KSP? As knowing that it was this fix causing any issue would be almost impossible to deduce.

[Sandworm]

I get the impression that the bug hasn't been fixed.  It looks like they had some code throwing unusual behavior (the orbit decay thing).  Rather than fix the code, it sounds like they have applied a fix to the behavior, a patch that operated to counter the bad code.  That would allow for toggling, but isn't really a fix.  It's a layer of new code fighting older code.  We have seen this approach from Squad before with the SAS "fix" that only applies in certain modes.  I would much rather see the underlying problem dealt with conclusively.

I haven't witnessed the orbital decay bug myself, yet.  In the past I only saw it on occasion and so cannot say it is gone.  ((CTDs from the VAB seem reduced, but are still happening.))

[Majorjim!]

Just now, Sandworm said:

I get the impression that the bug hasn't been fixed.  It looks like they had some code throwing unusual behavior (the orbit decay thing).  Rather than fix the code, it sounds like they have applied a fix to the behavior, a patch that operated to counter the bad code.  That would allow for toggling, but isn't really a fix.  It's a layer of new code fighting older code.  We have seen this approach from Squad before with the SAS "fix" that only applies in certain modes.  I would much rather see the underlying problem dealt with conclusively.

I haven't witnessed the orbital decay bug myself, yet.  In the past I only saw it on occasion and so cannot say it is gone.  ((CTDs from the VAB seem reduced, but are still happening.))

I don't think it can be fixed fully. I think it is the result of a fundamental limit within Unity itself.

[NathanKell]

@Sandworm I'm not sure whether to laugh or cry, but instead I'll ask you to actually read the threads where this is discussed quite fully.

 

@Majorjim It's possible for the fix to cause issues in the same way it is possible for _any_ fix to cause issues. All I can say is that at the moment no one has found any issues related to it.

[Nathair]

16 hours ago, NathanKell said:

Configurability is good, and is also a safety net.

Can't be said too often.

[p1t1o]

@NathanKell 

Thanks for weighing in (and for tackling thankless bugs!) its immeasurably more reassuring to hear things "from the horses mouth". It sounds like the orbital bug is a right JohnSnow

[Kobymaru]

On 26.6.2016 at 10:51 PM, NathanKell said:

 

@Majorjim It's possible for the fix to cause issues in the same way it is possible for _any_ fix to cause issues. All I can say is that at the moment no one has found any issues related to it.

First, thanks for the fix! Second, I don't think that togglable "this fixes a bug but potentially causes other bugs but probably not"-options are a good idea. I mean they could be applied to any bug and fix, really. Are you going to add an option per issue on the tracker? A fix is a fix, and if the fix breaks other things (which will always be a possibility, but that's what QA is for, right?), then it's better to fix the fix instead of adding the option to disable it.

That's just my opinion though.

Edited June 30, 2016 by Kobymaru

[CaptKordite]

Herein is my experience:

I had the orbital decay problem appear with 1.1.2. Everything was fine, orbits were rock solid, then suddenly ships started showing a decay. It was a matter of 0.1 meter per second or so. It affected every ship that was under active physics, the ship I was controlling and a ship I was attempting to dock with (making docking more difficult). If I switched to another ship, that ship had a problem. Craft on rails or under time acceleration did not suffer this issue.

This is Windows 10 and Steam. I deleted all the files in the KSP folder. Had steam revalidate and reinstall everything. Started a new stock game, and the issue was still there.

When the 1.1.3 update arrived it installed and the issue was resolved. I did not attempt loading up any of the old game saves. A pure stock game ran without issue. I added mods, started a new career, and everything continued to function normally until yesterday.

The symptoms are slightly different. Most of the craft already in orbit are fine. Their orbits remain steady. All the craft in orbit around Kerbin are fine. Their orbits remain steady. A space station and a lander recently put in Munar orbit are having an issue. Other satellites already in orbit around the Mun and Minmus remain fine. A new satellite in orbit around the Mun has the issue. The apoapsis and periapsis fluctuate quickly, perhaps 3 or 4 meters per second up and down, but they tend downward. With Kerbal Engineer you see the orbital period declining at a rate of about 0.5 second per second. This also occurs with a space station around Minmus. There is a space station around Kerbin, launched after the other two but derived from the same craft file, that does not have the problem.

In a Sandbox save, I launched the same craft file space station that was having problems around the Mun. It was stable in orbit around Kerbin but had the issue once it entered Munar orbit.

As before, I deleted the installation and allowed Steam to reinstall. The issue persisted on a new stock save. I copied the save file from my Windows 10 install over to my Ubuntu partition (which has graphics issues and runs really slow). The craft in Munar orbit still had the issue.

What I don't understand is how I can delete all the files and reinstall and still have the problem. Especially in light of the issue not being there when I first updated to 1.1.3. Is there something stored in the registry? If so, how does it also carry over in the save file to be a problem on Linux?

[Tyko]

I'm seeing this same issue as CaptKordite. Orbits around Kerbin are fine. Orbits around Minmus fluctuate wildly and slowly trend down. I noticed this over the past couple of days. 

I can send/upload any files if that would help

[Majorjim!]

10 minutes ago, tjt said:

I'm seeing this same issue as CaptKordite. Orbits around Kerbin are fine. Orbits around Minmus fluctuate wildly and slowly trend down. I noticed this over the past couple of days. 

I can send/upload any files if that would help

didn't they say it only works over 100km orbits? @NathanKell can you confirm this? does the 'fix' only work above 100km on other planets too?

[NathanKell]

No, no one said anything about 100km orbits in this regard.

 

What I have said is that when the drift compensation would make things worse rather than better, it automatically disables. Because, and let me repeat this one more time, there are basic, unavoidable limitations in using the integration method PhysX uses, and the floats PhysX uses. We have band-aided it as much as we can, but I dare say if you have survived three years of KSP where this issue was worse, you can survive it when it's better. Since when the drift compensation is off, the application of the force of gravity is exactly the same as it always was.

[FullMetalMachinist]

47 minutes ago, NathanKell said:

What I have said is that when the drift compensation would make things worse rather than better, it automatically disables.

47 minutes ago, NathanKell said:

Since when the drift compensation is off, the application of the force of gravity is exactly the same as it always was.

Sorry to sound like a simpleton, but let me make sure I've got this right. 

In pre-1.1 the drift was there, but hidden because of the jittery lines from CoM issues. In 1.1 the "fix" got rid of the jitters, but actually made the drift worse. In 1.1.3 the drift is mostly fixed, but some situations the fix makes things worse again, so it turns itself off. 

So what we have now is either A) the drift is fixed or B) the drift is the same as pre-1.1 (just more visible than it was because the jitters are gone). That all correct? 

And thanks for taking the time to explain the situation time and time again. You guys are awesome. 

[NathanKell]

@FullMetalMachinist yup! I did do a bit of replacement in krakensbane and floatingorigin as well to preserve doubles, so drift should be marginally better than it was even without the drift compensation. But yeah, the fact that you couldn't even read what the meters/tens/hundreds places were on orbits masked the drift often enough.

[CaptKordite]

"if you have survived three years of KSP where this issue was worse, you can survive it when it's better."

When The bug showed up for me, I stopped playing the game because it made docking too difficult to be fun. I waited for the 1.1.3 patch that promised to fix the problem. It did, for a time, but now that the problem has returned, I am again not having fun. I had reached the point in a career (another restarted career, by the way) that docking was going to become much more important. Those problems make the effort not fun. To my mind, KSP is a game. I am supposed to enjoy this entertainment, not merely survive.

[FullMetalMachinist]

2 hours ago, CaptKordite said:

When The bug showed up for me, I stopped playing the game because it made docking too difficult to be fun.

Maybe I'm missing something, but I don't see how the orbital drift is enough to complicate docking. The worst I ever saw was a few dozen meters per orbit. And the drift only happens to ships that are not "on rails".  So for docking it shouldn't be a problem until both ships are in the physics bubble and not on warp, which is just the last few hundred meters at most. 

Are you saying it's worse than that for you? Like I said, I could be missing something. But I've docked several times pre-1.1, in 1.1 & 1.1.2, and in 1.1.3. I didn't notice the drift enough for it to interfere in any way with docking.