Current, I don't think it can crash there, and I see nothing for this issue in my app logs, even with many users.

Do you have a stack trace ? maybe you have a too old SDL version and the bug as been fixed.

If this "skip" value is not set, you'll have a visual glitch / bad rendering (portrait vs landscape rendering)/

I have 3 reports in the play console.

2 of them are from Samsung devices:

Samsung Galaxy S7 (herolte), Android 7.0
Samsung Galaxy S6 Edge (zerolte), Android 7.0

Stacktrace:

java.lang.NullPointerException: 
 
  at org.libsdl.app.SDLSurface.surfaceChanged (SDLActivity.java:1400)
 
  at android.view.SurfaceView.updateWindow (SurfaceView.java:668)
 
  at android.view.SurfaceView.onWindowVisibilityChanged (SurfaceView.java:266)
 
  at android.view.View.dispatchWindowVisibilityChanged (View.java:11077)
 
  at android.view.ViewGroup.dispatchWindowVisibilityChanged (ViewGroup.java:1290)
 
  at android.view.ViewGroup.dispatchWindowVisibilityChanged (ViewGroup.java:1290)
 
  at android.view.ViewGroup.dispatchWindowVisibilityChanged (ViewGroup.java:1290)
 
  at android.view.ViewGroup.dispatchWindowVisibilityChanged (ViewGroup.java:1290)
 
  at android.view.ViewRootImpl.performTraversals (ViewRootImpl.java:1813)
 
  at android.view.ViewRootImpl.doTraversal (ViewRootImpl.java:1515)
 
  at android.view.ViewRootImpl$TraversalRunnable.run (ViewRootImpl.java:7091)
 
  at android.view.Choreographer$CallbackRecord.run (Choreographer.java:927)
 
  at android.view.Choreographer.doCallbacks (Choreographer.java:702)
 
  at android.view.Choreographer.doFrame (Choreographer.java:638)
 
  at android.view.Choreographer$FrameDisplayEventReceiver.run (Choreographer.java:913)
 
  at android.os.Handler.handleCallback (Handler.java:751)
 
  at android.os.Handler.dispatchMessage (Handler.java:95)
 
  at android.os.Looper.loop (Looper.java:154)
 
  at android.app.ActivityThread.main (ActivityThread.java:6682)
 
  at java.lang.reflect.Method.invoke (Native Method)
 
  at com.android.internal.os.ZygoteInit$MethodAndArgsCaller.run (ZygoteInit.java:1520)
 
  at com.android.internal.os.ZygoteInit.main (ZygoteInit.java:1410)




The 3rd stack trace is from a Huawei P9 (HWEVA), Android 7.0




java.lang.NullPointerException: 
 
  at org.libsdl.app.SDLSurface.surfaceChanged (SDLActivity.java:1400)
 
  at android.view.SurfaceView.updateWindow (SurfaceView.java:664)
 
  at android.view.SurfaceView.setFrame (SurfaceView.java:334)
 
  at android.view.View.layout (View.java:17685)
 
  at android.widget.RelativeLayout.onLayout (RelativeLayout.java:1079)
 
  at android.view.View.layout (View.java:17688)
 
  at android.view.ViewGroup.layout (ViewGroup.java:5631)
 
  at android.widget.FrameLayout.layoutChildren (FrameLayout.java:325)
 
  at android.widget.FrameLayout.onLayout (FrameLayout.java:261)
 
  at android.view.View.layout (View.java:17688)
 
  at android.view.ViewGroup.layout (ViewGroup.java:5631)
 
  at android.widget.LinearLayout.setChildFrame (LinearLayout.java:1762)
 
  at android.widget.LinearLayout.layoutVertical (LinearLayout.java:1606)
 
  at android.widget.LinearLayout.onLayout (LinearLayout.java:1515)
 
  at android.view.View.layout (View.java:17688)
 
  at android.view.ViewGroup.layout (ViewGroup.java:5631)
 
  at android.widget.FrameLayout.layoutChildren (FrameLayout.java:325)
 
  at android.widget.FrameLayout.onLayout (FrameLayout.java:261)
 
  at com.android.internal.policy.DecorView.onLayout (DecorView.java:774)
 
  at android.view.View.layout (View.java:17688)
 
  at android.view.ViewGroup.layout (ViewGroup.java:5631)
 
  at android.view.ViewRootImpl.performLayout (ViewRootImpl.java:2513)
 
  at android.view.ViewRootImpl.performTraversals (ViewRootImpl.java:2228)
 
  at android.view.ViewRootImpl.doTraversal (ViewRootImpl.java:1366)
 
  at android.view.ViewRootImpl$TraversalRunnable.run (ViewRootImpl.java:6768)
 
  at android.view.Choreographer$CallbackRecord.run (Choreographer.java:926)
 
  at android.view.Choreographer.doCallbacks (Choreographer.java:735)
 
  at android.view.Choreographer.doFrame (Choreographer.java:667)
 
  at android.view.Choreographer$FrameDisplayEventReceiver.run (Choreographer.java:912)
 
  at android.os.Handler.handleCallback (Handler.java:761)
 
  at android.os.Handler.dispatchMessage (Handler.java:98)
 
  at android.os.Looper.loop (Looper.java:156)
 
  at android.app.ActivityThread.main (ActivityThread.java:6523)
 
  at java.lang.reflect.Method.invoke (Native Method)
 
  at com.android.internal.os.ZygoteInit$MethodAndArgsCaller.run (ZygoteInit.java:942)
 
  at com.android.internal.os.ZygoteInit.main (ZygoteInit.java:832)

The SDL version I use is not the recent one, but I regularly look at the diff between my version which is around the time 2.0.5 was released, and the recent HG version. So I hope that I'm not missing important bug fixes. But of course I cannot say for sure.

Not being a frequently reported crash, I assume it will be hard to trigger (haven't managed myself). Might be that it only happens in certain situations of the Android app lifecycle, like e.g. resuming the app from the background. Somehow the system is calling onSurfaceChanged() before onCreate() which sounds pretty strange.

Can you also post your SDLActivity.java ?

Created attachment 3268 [details]
Java activity

The modifications to the file are:

* Added some code for in-app-billing (most likely irrelevant)
* Added some code for checking whether to download and/or mount expansion files (most likely irrelevant as well)
* Added one patch to disable system bars/go to sticky immersive mode upon resume (that one might be interesting w.r.t. this bug, although I don't see anything wrong with it yet)

The stack trace tells that mSingleton is null when calling 
"SDLActivity.mSingleton.getRequestedOrientation();"

And you initialized it in your method "onCreateNoDownload()". I don't know when/how it's called, it seems to be modification on your side. Maybe it's not called sometimes.

Yes, now that you mention it, I have overloaded SDLActivity and added all the code from the Android code example how to download missing expansion files. 

In that (unlikely when you've installed the app via Google Play) case, a different screen is shown until the download finishes.

Because I'm only overloading some of the lifecycle callbacks, indeed the surfaceChanged() call will land in SDLActivity, while the downloader is still in progress and SDL not even initalized :-(

Thanks for pointing this out!

As I cannot reproduce the bug itself, I assume that the explanation above is most likely correct and thus the bug not in SDL itself. Thus I'm marking it as resolved/invalid.

ok. maybe there is no real issue, just the mSingleton beeing not set when taking some path.

One thing that I'm wondering though is whether there is still some sort of memory leak in SDLActivity.java (which could also explain the unexpected callback in my case, because my modifications probably make things worse).

What I'm wondering about are these lines:

        mSurface = new SDLSurface(getApplication());

        mLayout = new RelativeLayout(this);
        mLayout.addView(mSurface);

        setContentView(mLayout);
        
mSurface and mLayout are newly created upon every activity start. The resulting layout object is set as current view.

Because I did not find anything related in onDestroy() I'm wondering whether these objects will actually be fully dereferenced once the app is in this stage of the lifecycle. NULLing the objects references in SDLActivity.java alone probably won't help, because the Activity might still have a ref to mLayout, and mLayout definitely has a ref to mSurface. So during onCreate() both instances are recreated, but maybe the old ones are still active at that time.

In principle, subsequent calls to setContentView() should remove the last reference to older instances of mLayout. I'm not sure though the garbage collector will detect these though.

If my assumptions are correct, then it might be possible that onSurfaceChanged() can be called while the SDLActivity's singleton is not initialized (yet), because the callback happens to an older instance of SDLSurface from the previous apps lifecycle.

Nevertheless, with the regular version of SDLActivity I guess its still highly unlikely if not impossible to trigger this behaviour. If you think that all of this is too much of a theoretical issue, you may close the bug again.

not sure, 
here's the life cycle diagram : https://developer.android.com/reference/android/app/Activity

I understand that after onDestroy(), everything should be released.

if it helps, my AndroidManifest.xml has also those lines :

android:alwaysRetainTaskState="true"
android:launchMode="singleInstance"

Mark as fixed, only one instance is available:
https://hg.libsdl.org/SDL/rev/0ef0e4cb7752

Mark as fixed