1573/Turn the screen back off if it's known to turn on by an action in scrcpy

[brunoais]

On my phone (cheap phone running Android 6), this hack works.
I was unable to find a decent way to achieve this otherwise.

What do you think? Any better ideas?

closes #1573

[brunoais]

@rom1v Any comments or sth?

[rom1v]

@brunoais Sorry I didn't have the time to test yet.

However, as I said in #1573 (comment), I still think this is too "hacky" and may cause problems. For example if 2 scrcpy are started for the same device (this is already a problem for cleanup, but here it's only a one-time problem), or it does not behave the same way if we press power physically or via Ctrl+p, and probably other problems…

[brunoais]

@brunoais Sorry I didn't have the time to test yet.

Ah OK. I didn't know whether you didn't have time or it was de-prioritized or forgotten.

You are right. The behavior is different between acting directly on the device and on scrcpy. Howeverr if there are more than one instance of scrcpy connected to the same device, I don't see what kind of problem this may really create.
However, I do understand the inconsistency this can create.

What about we get this feature opt-in by running it behind a command line argument, with (hopefully) proper explanation about it? Would that be a good compromise?

[brunoais]

I made this feature opt-in.
Even though it's execution is still not so good, maybe with some decent docs about it, it it works acceptably well.

[rom1v]

I quickly tested before my holidays, it worked quite well. I might consider merging with this behavior. But I will do more tests before. But probably not before few weeks.

[brunoais]

Sure. Thank you for keeping me updated.
I'll wait and let me know how to update the README to be more clear to users. I'll try to get some ideas myself.

[rom1v]

I acknowledge that it is convenient especially on right-click to turn the screen on (keeping the physical device screen off), even if --stay-awake (while plugged in) already provided a (partial) solution.

I took some of your commits and squashed them (I didn't make it optional, so I didn't include Create keep_screen_off command line option): 2a7f3d7

• I replaced magic numbers by KeyEvent constants (like KeyEvent.KEYCODE_POWER).
• I moved the test screenStayOff to the callers
• I also moved the calls to scheduleScreenOff() to the injection methods (as a side-effect, this avoids to call it on right-click when it actually injects BACK).

I also suggest to replace Timer by Executor: https://stackoverflow.com/questions/409932/java-timer-vs-executorservice
as follow: e31b13b

The branch is pr1577. Are you ok with these changes?

_Also, I will probably rename "screen off" to "power mode off" (for example schedulePowerModeOff()), to avoid confusion with the existing "turn screen on" (in particular, pressBackOrTurnScreenOn(), when it turns the screen on, schedules "screen off", which is confusing). _

(The behavior is a bit weird after pressing the physical POWER button: the screen is turned on, but then pressing POWER via scrcpy turns it back off.)

I don't know whether I will merge it yet.
More opinions are welcome :)

[brunoais]

@rom1v I approve all the changes you did.

[rom1v]

Merged cf9d449