Physics Debug Shape Update on CompoundCollisionShape

[Gaius0815]

There seems to be an issue with updating the physics debug shapes of CompoundCollisionShape.
I tried to remove a child shape of a CompoundCollisionShape at runtime. The child shape is correctly removed but the debug shape is still visible.
Please see the attached example code to reproduce the issue. You can remove the child shape with the DEL key and test the collision with the INSERT key.
Main.zip

This was tested with v3.2.4-stable-sdk1.
Updates: jMonkeyEngine SDK is updated to version NetBeans 8.2 Patch 2
Java: 1.8.0_212; OpenJDK 64-Bit Server VM 25.212-b04
Runtime: OpenJDK Runtime Environment 1.8.0_212-b04
System: Windows 10 version 10.0 running on amd64; Cp1252; de_DE (jmonkeyplatform)

This might be related to:
https://hub.jmonkeyengine.org/t/physics-debug-shape-doesnt-update/25455

[stephengold]

Thank you for reporting your issue.

I'm having difficulty opening the ZIP file you attached.

Which physics library were you using? The 3.2.4 SDK includes both jme3-bullet and jme3-jbullet.

[Gaius0815]

Here is the example as a text file:
Main.txt
I was using jme3-bullet-3.2.4-stable

[Gaius0815]

As far as I can tell the issue is within BulletDebugAppState line 248, private void updateRigidBodies(). Debug shapes are only updated when the number of PhysicsRigidBodys change. And child collision shapes are not PhysicsRigidBodys...

[stephengold]

Note: this issue was fixed in Minie v0.5.1 .

[Gaius0815]

Switching to Minie resolved the issue for me.
Thank you for investigating!

[MeFisto94]

Could we backport this fix to the engine?
In general, if possible, anything that is not a "deliberate choice/significant feature addition"?
I'd be up for the work to do it

[pspeed42]

I'm for it. I don't think there is any reason not to do it other than "who will do the work"... and so if you are already ready to do the work then it should be fine.

[stephengold]

Here's where I fixed it in Minie, apparently: stephengold/Minie@5e0c47a
You can ignore changes related to debug-mesh normals and resolution, which don't exist in jme3-bullet.