FlightIntegratorPerf patch breaks Persistent Rotation during timewarp

[dok377]

Describe your problem with KSPCF : When this patch is active, Persistent Rotation doesn't rotate crafts during timewarp. It resumes rotation after the timewarp stops. When the patch is disabled everything works as it should.

Video of the bug: https://youtu.be/-btIv3-Q63o

KSP version : 1.12.4 with only KSPCF, Persistent Rotation and its dependencies
Link to your KSP.log file : https://drive.google.com/file/d/1gOqtxYFgfzef2Nq5nQnipWrmyoHgvM3G

[gotmachine]

This is because we are fixing a KSP inconsistency where the vessel calculated MOI (and angular momentum) was still available on packed vessels, but only when that vessel has previously been unpacked at least once. Meaning that vessels being initially loaded out of unpacking range would have those values set to zero, but vessels going out of packing range or entering non-phys warp would still have access to (stale) values.

The stock code is trying to reset those values, but fails to do so due to a silly mistake.

For reference, here the KSPCF code where this happens :

KSPCommunityFixes/KSPCommunityFixes/Performance/FlightIntegratorPerf.cs

Lines 121 to 125 in 42b3f23

	if (vessel.angularVelocityD == Vector3d.zero && vessel.packed)
	{
	vessel.MOI.MutateZero();
	vessel.angularMomentum.MutateZero();
	}

and

KSPCommunityFixes/KSPCommunityFixes/Performance/FlightIntegratorPerf.cs

Lines 209 to 210 in 42b3f23

	vessel.MOI.MutateZero();
	vessel.angularMomentum.MutateZero();

And for completeness, do note that we are also fixing the same issue for rb_velocity and angularVelocity. Note that the double backed versions of those were already correctly set to zero, as the Vector3d.Zero() method doesn't suffer from the same issue.

KSPCommunityFixes/KSPCommunityFixes/Performance/FlightIntegratorPerf.cs

Lines 203 to 207 in 42b3f23

	vessel.rb_velocity.MutateZero();
	vessel.rb_velocityD.MutateZero();
	vessel.velocityD.MutateZero();
	vessel.angularVelocity.MutateZero();
	vessel.angularVelocityD.MutateZero();

I'm a bit torn on what to do, TBH. Technically, we are just exposing more consistently a bug in Persistent Rotation, which shouldn't be using those fields while the vessels are packed.

In stock, an easy way to reproduce the same PR issue is to launch and teleport two vessels in orbit at a 250m distance from each other (this is easily done with the stock rendezvous cheat, in the "set orbit" tab of the ALT-F12 menu). Make one of the vessels rotate, switch to the other, quick-save and quick-load : the other vessel won't be rotating anymore (you don't even need to engage timewarp), as long as it doesn't get close enough from the active vessel (unpacking range in stock is 200m).

And for reference, the problematic code in PR :
https://github.com/linuxgurugamer/PersistentRotation/blob/dc1fd8de0bbac625234aad560d9249fa79f065ab/Source/Main.cs#L361-L363

[dok377]

So this is technically not an issue with KSPCF. Maybe I should also notify LGG? He might take a look at it and see if anything can be done.

[dok377]

Created an issue there, maybe there's something that can be done on PR's side. IF LGG is still active, that is.

[linuxgurugamer]

Not sure if it can be fixed, I'll take a look