Game Crashing Bug - Unmanned Tundra Colonization parts vs Inflatable Habs

[GHrynk]

There is a game crashing interaction between the Tundra 2.5m and 3.75m Colonization modules and the inflatable 2.75m, 4.25m, Ring, and Ranger Habs when the vessel is NOT manned. This may not be all, as I haven't exercised every MKS part permutation. My test vehicle is an RC-001S with 4ea Z4K batteries, 2ea Gigantors, 3.75m LS Tank, and launch clamps to hold it up. Colonization and inflatable hab parts are mixed together in iterative evaluations.

Upon Alt-F12 cheat of unmanned vehicle to min orbit above Kerbin, I activate the various module converters. When clicking "Start Living Module" on the Colonization parts, the bug takes place and the vehicle disappears, but altitude goes to 0 meters, even though camera view is still in orbit. Attempting to revert to VAB immediately causes a game crash. This is the least spectacular failure mode - with other mods mixed in (such as Better Time Warp), the craft has exploded and parts separately fly off (though this may be a rigid attachment/auto-strut induced effect). SO FAR, I haven't seen this bug when at least one Kerbal is aboard during the evaluation.

With real time log window, some of the errors I see come up immediately upon the event include:

"computeMassandInertia: Provided mass or density has no valid value" "PxRigidBodyExt::setMassAndUpdateInertia: Mass and Inertia computation failed, setting mass to 1 and Inertia to (1,1,1)" "Infinity or NaN floating point numbers appear when calculating the transform matrix for the Collider. Scene hierarchy path 'largeSolarPanel/mode/SunPivot/panel/panelbase'" (lots of these) Electric charge suddenly goes to NaN out of 18015...

I've done this using a clean install of KSP 1.2.2.1622 Win x64 (with pressure, g force, etc., limits enabled) with USI Constellation 2017.03.17 (with some part packs deleted afterward to reduce loading time) and no other mods.

[nagleaidan]

Can confirm on my test setup (on mac). Putting log files here for storage. https://www.dropbox.com/s/og4m0rh0efhxfq2/Player%28mksbug%29.log?dl=1 https://www.dropbox.com/s/9p59pufurgmabsb/KSP%28mksbug%29.log?dl=1

[GHrynk]

UPDATE: This bug still seems to be active in MKS 0.52.1, KSP 1.3.0.1804 (Win x64). Test craft included unmanned Mk 1-2 Cmd Pod, RC-L01 RGU, 2.5m Tundra Kolonization Module, 2.75m Expanding Hab, Expandable Hab Ring, and 2.5m Kerbitat. Expanding modules and Kerbitat worked without incident after activating, but game crashed as previously described when activating Kolonization module. Verified that crash does not occur if crew are present.

[Gnurfos]

I tried again but still cannot reproduce. Could you attach a craft file or more detailed instructions ? Are the inflatables inflated before launch ? Are all parts filled with machinery. Are there ColonySupplies on board ?

[GHrynk]

There IS a missing piece to the problem! It's the launch clamps that appear to be the catalyst and the action of NOT staging them off before F12 cheating to orbit. It also seems to be limited specifically to the Hab Ring now (previously it was all the inflatables). Try this stack and procedure:

Unmanned Mk1-2 with two Z-4K batts underneath, and on top stack a 2.5 Colonization Module, Hab Ring (in Habitat config, with default 500 machinery), and an RC-L01. Attach two TT18-As to sides of Colonization module, then launch vessel to pad.

If you use the staging function to release the launch clamps and let the vessel fall free to the pad, then F12 cheat to orbit, everything appears to be fine. HOWEVER, if you cheat to orbit, while the clamps were still attached, the bug manifests itself.

[Gnurfos]

A fix will come via USI-LS PR#252 (No idea why the problem only manifests itself under these very particular circumstances. For all that I know, it should be anytime there's an unmanned vessel with any hab part, activating a kolonization module).