Report a new location only after receiving a non-null update

[mvglasow]

Prevents stale locations from being re-sent when backends supply null locations, fixes #75

[mar-v-in]

This does not allow to use forced locations (primarily used for debugging currently, but might also be useful for automated testing in the future).

Why don't you just verify that reportLocation is only called after the location was updated (i.e. location to be reported has newer timestamp than last report)?

[mar-v-in]

Additionally, report(null) is now different from report(old location) which should not be the case (no report, null report and old location report should all be treated the same as they all do mean that there is no new data available)

[mvglasow]

3b995b8 modifies BackendHelper to compare timestamps and report locations only if the new location has a more recent timestamp than the previous one.

For forced locations, eead0bd adds code to make them behave in a similar way as a location update from a backend: a forced location which is not more recent than the last forced location is silently discarded. When a new location is forced, reportLocation() gets called, resulting in a single location update. Differences to a backend: forcing a null location will clear the last forced location (which is the same behavior we've had up to now), and forced locations are always asynchronous (there is no equivalent to update() for forced locations).

[mar-v-in]

should be boolean instead of Boolean

[mvglasow]

Fixed in 1093f40

[n76]

Both of my backends are now causing a crash. From the trace back (below) it seems that I don't have the time set properly on the location report, but I am getting this even if I perform a myLocReport.setTime(System.currentTimeMillis()) immediately before calling report(myLocReport).

I've taken a look at your Apple and Mozilla backends to see what you are doing differently and it appears that the setTime(System.currentTimeMillis()) call is what you use.

Any ideas?

E/AndroidRuntime(22629): Process: org.fitchfamily.android.wifi_backend, PID: 22629
E/AndroidRuntime(22629): java.lang.RuntimeException: Error receiving broadcast Intent { act=android.net.wifi.SCAN_RESULTS flg=0x4000010 } in org.fitchfamily.android.wifi_backend.wifi.WifiReceiver@39f248db
E/AndroidRuntime(22629):    at android.app.LoadedApk$ReceiverDispatcher$Args.run(LoadedApk.java:876)
E/AndroidRuntime(22629):    at android.os.Handler.handleCallback(Handler.java:739)
E/AndroidRuntime(22629):    at android.os.Handler.dispatchMessage(Handler.java:95)
E/AndroidRuntime(22629):    at android.os.Looper.loop(Looper.java:135)
E/AndroidRuntime(22629):    at android.app.ActivityThread.main(ActivityThread.java:5254)
E/AndroidRuntime(22629):    at java.lang.reflect.Method.invoke(Native Method)
E/AndroidRuntime(22629):    at java.lang.reflect.Method.invoke(Method.java:372)
E/AndroidRuntime(22629):    at com.android.internal.os.ZygoteInit$MethodAndArgsCaller.run(ZygoteInit.java:903)
E/AndroidRuntime(22629):    at com.android.internal.os.ZygoteInit.main(ZygoteInit.java:698)
E/AndroidRuntime(22629): Caused by: java.lang.NullPointerException: Attempt to invoke virtual method 'long android.location.Location.getTime()' on a null object reference
E/AndroidRuntime(22629):    at android.os.Parcel.readException(Parcel.java:1552)
E/AndroidRuntime(22629):    at android.os.Parcel.readException(Parcel.java:1499)
E/AndroidRuntime(22629):    at org.microg.nlp.api.LocationCallback$Stub$Proxy.report(LocationCallback.java:91)
E/AndroidRuntime(22629):    at org.microg.nlp.api.LocationBackendService.report(LocationBackendService.java:52)
E/AndroidRuntime(22629):    at org.fitchfamily.android.wifi_backend.backend.BackendService.process(BackendService.java:186)
E/AndroidRuntime(22629):    at org.fitchfamily.android.wifi_backend.wifi.WifiReceiver.onReceive(WifiReceiver.java:92)
E/AndroidRuntime(22629):    at android.app.LoadedApk$ReceiverDispatcher$Args.run(LoadedApk.java:866)
E/AndroidRuntime(22629):    ... 8 more

[mar-v-in]

@n76: Use v1.6.4

[n76]

I'm testing against GmsCore v0.2.2, listed as the latest in your downloads and just showing up in your F-Droid repository (I'm using F-Droid to keep up to date with your releases).

FWIW, I am seeing the same crashes and similar stack traces on your Apple and Mozilla backends.

[mar-v-in]

Please try to download the app again (or update the f-droid repo) there was a broken build online for a few minutes.

[n76]

Apparently that was the problem. . . Not crashing now. Thanks!