NativeActivity is destroyed when a USB device is attached or detached.

[IllusionMan1212]

As the title suggests, is there any solution or workaround for when this happens?

[DanAlbert]

Maybe another user listening here knows an answer, but if you want a fix you probably need to file an OS bug. NativeActivity is part of the OS.

One thing you might try would be switching to GameActivity: https://developer.android.com/games/agdk/game-activity. If the OS is sending the app destroy event when it shouldn't be though, there's nothing that can be done about this from the app side.

[IllusionMan1212]

I was hesitant to file a bug because I wasn't sure if this is intended behavior or not. When the USB device is attached/detached the application receives PAUSE, TERM_WINDOW, STOP, and INPUT_CHANGED commands which is normal from my understanding because a USB (de)attachment is a configuration change, but what's abnormal in my opinion is that destroyRequested is set to 1 and no RESUME, START, or INIT_WINDOW commands are received.

I think GameActivity will suffer from the same problem since it extends NativeActivity.

Can you point me to the correct place where I should file the OS bug?

[IllusionMan1212]

So I decided to try the ndk samples to see if they were affected too, turns out they weren't so it was clear I was doing something wrong. It turns out I was confused on what destroyRequested is supposed to signal and thought the android_main function will only be called once by the NativeActivity, so I never let the app thread exit when it should have. In hindsight it makes total sense that android_main will be called every time the NativeActivity is destroyed and remade.

Thank you for taking the time to reply.

[DanAlbert]

Ah, yeah, that'll do it.

I think GameActivity will suffer from the same problem since it extends NativeActivity.

Actually it doesn't, it extends AppCompatActivity: https://android.googlesource.com/platform/frameworks/opt/gamesdk/+/refs/heads/main/game-activity/src/main/java/com/google/androidgamesdk/GameActivity.java. Even though it's not relevant to this problem, that can help in a lot of other ways.

Can you point me to the correct place where I should file the OS bug?

In case you need to know in the future, http://b.android.com, and then one of the components in the section titled "platform" (https://source.android.com/docs/setup/contribute/report-bugs#platform). When in doubt, Framework.

[enh-google]

do you have a suggested improvement to the destroyRequested doc comment?

(i'm assuming that the fix you'd suggest to the android_main doc comments is to make it clear that it's not a one-off on initial creation, it's every time.)

[IllusionMan1212]

Nothing concrete but maybe append The android_main function must return to its caller if this is non-zero. to the existing comment.

I think because I was only writing native code I wasn't thinking about the usual activity lifecycle. I expected STOP/PAUSE and START/RESUME to be at play but it completely slipped my mind that the main function will have to be called more than once when the OS recreates the activity, so maybe making it clear somewhere that the main function is subject to being called multiple times is good too.

[enh-google]

does https://android-review.googlesource.com/c/platform/ndk/+/3218918 look better?

[IllusionMan1212]

Looks great. Thank you