Ground vehicles/EVA's don't work correctly near other players.

[godarklight]

JohannesMP from the forums made a very good video demonstrating what happens when trying to use ground vehicles near other players.

Interesting events are listed in his post:
http://forum.kerbalspaceprogram.com/threads/55835-KMP-v0-1-5-0-0-23-wip-alpha?p=910958&viewfull=1#post910958

I replied with relatively-educated guesses on each issue in the following post.

This issue is a combination of many different unrelated bugs. Should we track them here in this post (edit a list of bugs), or track them in separate issues?

[JohannesMP]

Just for clarity, I'm JohannesMP on the forums, in case I need to reply here :)

[JohannesMP]

Looking through your forum reply, I'd tentatively say that we have the following issues (again, some of these might have been reported already, and I might have missed some) :

• When first logging in with a new vehicle, other new vehicles will not be visible (1:47)
• When viewing the tracking station where another user is controlling a vehicle, a vehicle will sometimes show up multiple times in the list, despite only existing once when checking /listships
• Vehicle position data received by a client could be interpolated better to make the slow tick rate (333ms) less obvious. For example using the current velocity/acceleration the vehicles position during the next 'tick' could be approximated, making movement appear more smooth without requiring higher tick rates. As a separate issue, higher tick rates for close vehicles, or vehicles that are accelerating (as opposed to constant velocity) should also be considered.
• When Taking control of a vehicle, other vehicles in the vicinity will sometimes explode
• Modified parts, such as docking shields don't synchronize fast enough, and sometimes not at all. It sounds like the "Cannot board a full module" at 11:05 might be related to this.
• Miscellaneous bugs with EVA'd Kerbals, such as not being able to climb a ladder, or a Kerbal 'poof'ing because it's not syncing him as being on a ladder.

[godarklight]

1:47 seems to be a surface only bug, I know rockets will spawn in correctly when they are being launched, but I think they need to be higher than 22km (otherwise they would be insta-destroyed because they are on rails).

I know we send position and velocity, but I don't think we send acceleration: #307. Kerbals actually walking was discussed in #334.

The tick rate would be difficult to change, because at the moment the player status and the active vessel (primary updates) and all non-active vessels within a certain range (secondary updates) are packaged up and sent in a plugin_update message.

Another bug that I saw, and I'm fairly positive about this (it happened at 7:15), both players look like they are sending secondary updates for the debris. That's why you see it lift in the air, one player saw it dropping and sent that, so the other player saw it dropping and reported that in a bit of a loop.

Taking control of a vehicle: They used to instantly explode, but in your video you can very clearly see it twisting (attempting to barrel roll), as if the rotation was being set incorrectly. Hopefully this is related to them going "poof".

Modified parts: There's either going to be an event that fires on a part change, or we will have to scan the ship for changes. Either case will result in the full ship being sent again (which we do every 45 seconds). However, I'm not quite sure why the rover never opened its docking ports, you left it for longer than that...

[JohannesMP]

1:47 might also be what's causing people to not be able to find one another with planes or other in-atmosphere instances, such as in this case: #599 - one temporary workaround seems to be to have both people spawn their vehicle, move it outside of the bubble, disconnect completely, and then reconnect and resume control of the vehicle via the tracking station.

[TehGimp]

1. 1:47 definitely shouldn't be Vessels are destroyed if they touch the atmoshpere #256 at least (which was my own best guess for the culprit behind Atmospheric vessels are... weird #599)--the fact that reconnecting is enough to fix it up is interesting though. I expect existing debug logging would be illuminating if anyone can capture as much for a similar test (not sure when I'll get a chance next myself). We'll track this issue here.
2. Multiple versions of the piloted vessel in Tracking Station is Ships duplicating. #526 -- current easiest way to reproduce is to have someone sit in a vessel while you connect, most often you'll have dupes once the initial universe sync is finished. I'm not sure if this can occur outside of this test case, and so far a definitive fix has been unusually elusive.
3. Better interpolation, as noted by @godarklight, is requested in Enhancement: Smoother motion using last known acceleration values #307 and shouldn't be particularly challenging (though it likely involves breaking changes to the KMPVesselUpdate object that will require special consideration for old-database upgrades). The current tick rate behaviour isn't likely to be substantially changed in the near future due to various performance-related concerns.
4. Can anyone confirm if this is a near-surface rotation problem or dupe similar behaviour in orbit? If it's a rotation bug let's get a separate issue report for it.
5. Right now this type of structural update is set to take at most 45s to reach other players (which, to be sure, is very high), but the update can be dropped in certain circumstances and it looks like 'something' is preventing it from being applied here. Once someone does an orbital rendezvous test with something similar (e.g. extending solar panels) we'll know whether those updates are failing universally when in physics range or if it's maybe an atmo/surface issue. This one is also probably worth breaking into its own report once someone gets a chance to characterize the underlying issue better.
6. Better ladder behaviour for EVA'd kerbals is something that can be worked on (there's lots of references to ladder-related junk in the API to play around with), but I don't think we're at the point where we should prioritize something so open-ended. I'll create a separate issue report for it so we don't forget, but will leave it un-milestone'd for now.

[godarklight]

Adding reference to #499, they look related.

[JohannesMP]

I've moved this to its own issue for clarity's sake: #692