assertion failure in jme3test.games.CubeField

[stephengold]

Currently (in master branch at 3.3-6680) the CubeField app crashes with an AssertionError the 1st time ENTER is pressed. Here is a typical stack trace:

START
Dec 30, 2018 10:27:33 AM com.jme3.app.LegacyApplication handleError
SEVERE: Uncaught exception thrown in Thread[jME3 Main,5,main]
java.lang.AssertionError
	at com.jme3.scene.Spatial.updateMatParamOverrides(Spatial.java:596)
	at com.jme3.scene.Spatial.updateGeometricState(Spatial.java:911)
	at jme3test.games.CubeField.gameLogic(CubeField.java:284)
	at jme3test.games.CubeField.simpleUpdate(CubeField.java:145)
	at com.jme3.app.SimpleApplication.update(SimpleApplication.java:239)
	at com.jme3.system.lwjgl.LwjglAbstractDisplay.runLoop(LwjglAbstractDisplay.java:151)
	at com.jme3.system.lwjgl.LwjglDisplay.runLoop(LwjglDisplay.java:197)
	at com.jme3.system.lwjgl.LwjglAbstractDisplay.run(LwjglAbstractDisplay.java:232)
	at java.lang.Thread.run(Thread.java:748)

[stephengold]

Commenting out line 284 seems to resolve the issue, though I don't fully understand why:

                Geometry playerModel = (Geometry) player.getChild(0);
                Geometry cubeModel = cubeField.get(i);
                //cubeModel.updateGeometricState();

                BoundingVolume pVol = playerModel.getWorldBound();
                BoundingVolume vVol = cubeModel.getWorldBound();

[pspeed42]

a) wouldn't it be nice if JME assertions included a message.
b) calling updateGeometricState() manually is generally incorrect anyway. Not sure why it was being done.

[pspeed42]

I guess the assertion is making sure that the parent's overrides were applied first?

    protected void updateMatParamOverrides() {
        refreshFlags &= ~RF_MATPARAM_OVERRIDE;

        worldOverrides.clear();
        if (parent == null) {
            worldOverrides.addAll(localOverrides);
        } else {
            assert (parent.refreshFlags & RF_MATPARAM_OVERRIDE) == 0;
            worldOverrides.addAll(parent.worldOverrides);
            worldOverrides.addAll(localOverrides);
        }
}

This implies that it is incorrect to ever call updateGeometricState() directly on a child. (which is new... was just ill advised before)

[stephengold]

Thanks for the analysis. I suppose we should remove line 284 at least, since examples should ... set a good example. I'm unsure whether we should remove the assert as well.

[pspeed42]

Well the assert is confirming that the parent has been updated before applying its overrides. If we want to allow direct updateGeometricState calling on children then a lot more work would have to be done other than just removing that line.

It would be nice if we would go through and add messages to our asserts so that they mean something, though.

[stephengold]

Fixed in master branch at SHA id=44801da70687f8fbf9ee609cc1ece7f9144d9341