PR #390 is breaking calculations dependent on part size in the SPH.

[BrettRyland]

What did you do before the bug occurred?

PR #390 (which was added in v1.41.0) is breaking BDArmory HP and armour calculations in the SPH by returning 0-sized parts when they're selected from the menu.
In general, it breaks any part size calculations in the editor performed from OnStart.

This is easily demonstrated using a test class

public class PartSizeTest : PartModule
{
	public override void OnStart(StartState state)
	{
		var bounds = part.GetRendererBounds();
		Debug.Log($"DEBUG {part.partInfo.name} has {bounds.Length} bounds: {string.Join(", ", bounds)}, part size: {CalculatePartSize(bounds)}");
	}

	public Vector3 CalculatePartSize(Bounds[] bounds)
	{
		Bounds result = new Bounds(part.transform.position, Vector3.zero);
		for (int i = 0; i < bounds.Length; i++)
		{
			result.Encapsulate(bounds[i]);
		}
		return result.size;
	}
}

and patch (so that the class is added to all parts for the test)

@PART[*]
{
	%MODULE[PartSizeTest]
	{}
}

then starting with a MK1Fuselage tank and selecting a second one from the menu

With v1.40.1:

[LOG 13:35:23.511] DEBUG MK1Fuselage has 1 bounds: Center: (0.0, 10.0, 0.0), Extents: (0.6, 0.6, 1.0), part size: (1.3, 1.3, 1.9)
[LOG 13:35:24.723] DEBUG MK1Fuselage has 1 bounds: Center: (16.5, 1.2, 15.6), Extents: (0.6, 0.6, 1.0), part size: (1.3, 1.3, 1.9)

With v1.41.1 (or v1.41.0):

[LOG 13:33:34.369] DEBUG MK1Fuselage has 1 bounds: Center: (0.0, 10.0, 0.0), Extents: (0.6, 0.6, 1.0), part size: (1.3, 1.3, 1.9)
[LOG 13:33:36.939] DEBUG MK1Fuselage has 0 bounds: , part size: (0.0, 0.0, 0.0)

What happened?

The bool ignoreTransparentFX = !p.frozen; condition used in the PR is insufficient and should not be used, or at least, it should be optional and default to the previous behaviour.

Note: any such modification that changes a behaviour (other than purely performance behaviours) should default to the status quo to avoid permanently breaking mods that are no longer maintained.

Upload your ksp.log file

KSP.log

A screenshot of the problem

No response

[BrettRyland]

Running a few tests, a better condition to use in PR #390 would be

bool ignoreTransparentFX = !p.frozen && !AllParentsTransparent(p.transform);

where

static bool AllParentsTransparent(Transform t)
{
    if (t.gameObject.layer != LayerMask.NameToLayer("TransparentFX")) return false;
    if (t.parent == null) return true;
    return AllParentsTransparent(t.parent);
}

(feel free to give the function a better name).
This gives ignoreTransparentFX = false while the part is picked up, but ignoreTransparentFX = true otherwise.
Also, at no point in my testing did p.frozen become true as the selected part gets unfrozen when picked up, copied or switching to placement mode. Moving a part that can be placed into an inventory over a part with an inventory caused a part to become frozen.