Vessels change position abruptly when using non-physical timewarp

[scimas]

Get 2 vessels close together, use non-physical timewarp. One or both vessels can be seen changing position abruptly by about 200 m. When you switch back to 1x timewarp, the position again changes abruptly to (presumably) where the vessel should be after the timewarp (basically looks like it subtracts the previous abrupt change.) Attaching save file in txt format where this can be verified.
Bug_report.txt

[SunScream]

Happens at relatively long range and includes both targeted and non-targeted objects.

https://github.com/SunScream/Principia-issues/blob/master/Offset_in_Timewarp.zip

[pleroy]

We were able to reproduce the problem. Tracing the altitude around the time of the warp shows that KSP hesitates as to the altitude of the vessel:

E0605 16:48:35.144068  9980 interface.cpp:630] RescueShip 77664.5017526983
E0605 16:48:35.172071  9980 interface.cpp:630] RescueShip 77664.5127555303
E0605 16:48:35.206074  9980 interface.cpp:630] RescueShip 77485.1164742472
E0605 16:48:35.211076  9980 interface.cpp:630] RescueShip 77664.5356854253
E0605 16:48:35.226076  9980 interface.cpp:630] RescueShip 77485.1321253225
E0605 16:48:35.255079  9980 interface.cpp:630] RescueShip 77485.1399544015
E0605 16:48:35.284082  9980 interface.cpp:630] RescueShip 77485.1477861585
E0605 16:48:35.289083  9980 interface.cpp:630] RescueShip 77664.5797277195
E0605 16:48:35.312085  9980 interface.cpp:630] RescueShip 77485.163457879
E0605 16:48:35.342088  9980 interface.cpp:630] RescueShip 77485.1712975268
E0605 16:48:35.346088  9980 interface.cpp:630] RescueShip 77664.6127787221

The correct altitude determined by Principia is 77664.0921842906391 m. Where the ~180 m jitter comes from, we have no idea...

[scimas]

So I noticed something interesting yesterday. I was trying to rendezvous and dock which wasn't going so well. I started time warping, experienced the sudden shifting of position of the vessels. It looked like the vessels would crash if I didn't come out of the timewarp and decided to see what happens. They didn't crash but did come very close and passed by each other, so I came out of the warp. Now usually when I come out of time warp, both vessels jitter and it at least looks like the amount of jittering is different for both vessels; you can literally see the target vessel jumping around on screen. This time it didn't happen. The jitter was present, I could see the orbit parameters like Ap and Pe jumping by about 200m when going in and out of warp, but it was like both vessels were jumping by exactly the same amount in exactly the same direction; instead it looked like the celestial body (minmus) was jumping around by that amount.
May or may not be relevant to the bug, just seemed a different experience so decided to share.

[ToniStack]

Noticed this issue myself using RSS in conjunction; but also even when not using timewarp noticed after getting under 200 meters the game suddenly flickered the astroid as well as made another copy of one approximately 700 meters away from the original, after witch docking to claw and capture the astroid in earth orbit became impossible

[scimas]

Since Cauchy, when vessels come close together (close enough to switch between them using hotkeys), they change position abruptly. It looks almost exactly like this timewarp issue, so wrote here before opening a new issue.

[eggrobin]

The issue was that the function used to set the position of unpacked vessels relative to celestials was using the degrees of freedom of the celestials at current_time_, but the degrees of freedom of the parts at the time passed to AdvanceParts, which is after current_time_. This resulted in an incorrect position when the vessels were unpacked (out of timewarp).
As far as I can tell (by looking at the code, I haven't tested that) this has been broken since Cardano.
This will be fixed in Cayley, to be released on the new moon, 2017-07-23.

[eggrobin]

Note: that is only part of the problem; this solves the altitude of the active vessel jumping when warping, but not nearby vessels jumping when warping.

[eggrobin]

Hm, it seems that the remaining problem occurs for vessels ~200 m away, but not for vessels right next to the active vessel (and regardless of warp, a jump occurs when approaching those vessels, as described by @scimas). Further investigation is needed...

[eggrobin]

The jump now seems to affect unpacked vessels, and be in the direction of their orbit (KCI frame, in a low orbit). Further, I am testing in a low orbit, whose speed is 2290 m/s, and the jump is roughly 45 m. This seems consistent with an off-by-one-frame error (2290 m/s / 50 Hz = 45.8 m).

[eggrobin]

I think I know what is going on here, but that's going to have to be a separate issue, it's quite tricky.
Opening #1473 to track this separately.