Bug 863103 - Add BackgroundService runIntentInService() and WakeLock

[cpeterso]

@rnewman r?

After reviewing Google's Cloud Messaging (GCM) implementation, here are two BackgroundService additions that may be useful:

1. Add runIntentInService() utility method to ensure the original Intent's extras are copied to rebroadcast Intent:

   https://code.google.com/p/gcm/source/browse/gcm-client/src/com/google/android/gcm/GCMBaseIntentService.java?r=3f8285f108caecf9ee040cdadda3a024b81f7e3e#268

2. Add WakeLock to ensure BackgroundService receives the Intent after the BroadcastReceiver returns.

GCM releases its WakeLock at the end of onHandleIntent():

https://code.google.com/p/gcm/source/browse/gcm-client/src/com/google/android/gcm/GCMBaseIntentService.java?r=3f8285f108caecf9ee040cdadda3a024b81f7e3e#241

Can we safely assume that Android will always call our onDestroy() method? If not, then we risk leaking a WakeLock.

This change is tricky because it relies on every subclass calling super.onHandleIntent() and super.onDestroy() (if the subclass has overridden those methods). Also, we might be able to remove the ugly EXTRA_RELEASE_WAKE_LOCK extra if we assume BackgroundServices are always launched from runIntentForService() and thus called acquireWakeLock().

[rnewman]

onDestroy is documented as being called once a Service is no longer being consumed. So there should be no risk of leaking a WakeLock.

[rnewman]

r+ on the runIntent stuff. f+ on the WakeLock changes, but I think there are some implementation improvements that can be made. Still, hurrah for rigor improvements…

[rnewman]

Chris, want to spend two minutes rebasing this, even if you don't want to wrestle with the WakeLock stuff yet?

[cpeterso]

Richard, here is a simpler proposal that uses (less efficient, but less error-prone) timed WakeLocks. Comments in the .java file explain more. Note that this code is not well-tested yet! <:)