Users can enter arabic phone number or click arabic phone number
to send MMS. Works for generic Unicode digits (full-width, etc.).
bug:5615791
Change-Id: Ieec8c5c6c3736ee2b4ac8ddf17f8c41b2001460e
Signed-off-by: Jake Hamby <jhamby@google.com>
Modify PhoneNumberUtils to automatically convert non-ASCII digits,
such as Arabic-Indic numbers, CJK full-width digits, etc., to ASCII
in normalizeNumber(), extractNetworkPortion(), and stripSeparators().
This enables the SMS application to support sending SMS's to phone
numbers written with Arabic, or other non-ASCII digits. The number will
be converted to ASCII digits and formatted for the user according to the
country formatting rules.
Bug: 5615791
Change-Id: I42039285db5795b1dda22e4251f54af302e27f13
Signed-off-by: Jake Hamby <jhamby@google.com>
Changed the SensorManager class so that it only contains API-related
bits including what's needed to support legacy sensors. Mostly just
moved stuff around. Making the class abstract is safe because
it does not have a visible constructor in the API.
One minor change is that the cache of sensor type to sensor lists
is now per instance of SensorManager instead of being static.
We can fix this if desired.
Another small change is that we bail out early from registerListener
if the listener has already been registered for the particular
sensor. This happened for both legacy and standard listeners.
The problem is that the ListenerDelegate maintains two lists of
sensors, one is a Map and the other is a List. Adding a sensor
twice causes one entry to be added to the Map and two entries to be
added to the List, but when the sensor is removed the next time, only
one entry is removed from the List, leaving it in an inconsistent
state.
Removed Sensor.getLegacyType() since the value it provides is only
needed in LegacyListener and we don't really save any significant
computation by caching it. Removing the field makes support for
legacy sensors a little more self-contained.
Bug: 6339552
Change-Id: I50d41ac97cf535924f2bfa2026d28547a4d00286
Report to developers if active network is "metered" and define it
as the user being sensitive to heavy data usage.
Bug: 3001465
Change-Id: I855ca3cd3eb1de3c4814148d70ccf24957af898a
Before, RecognitionManagerService just cleared the recognizer
setting, which the Settings app really doesn't like.
Bug: 6332933
Change-Id: If4f9b583c304c5ea99021dddda50fca55e3ac541
Fixes b/6341858 AdapterView does not properly check for null before checking empty view accessibility info
Change-Id: Ia19fdef2c7c5f3e6c3053ebc754efe6a664f9d66
The unit tests were eating failure reasons. Remove some assertions so
the useful messages percolate up to the JUnit report.
Change-Id: Ica8efe403ca3eb4f0c4f2a18939a22085f74a08f
All the APKs in the FrameworkCoreTests's res/raw directory should be
built by source.
This change makes one more built from source that was causing test
failures with mismatched certificates.
Change-Id: I1baed15477ee39c45d3d77f9b02b81f771a16ca2
Forward-locked apps aren't very prevalent, but it needed to be
restructured to make sure both streams and ZipFile objects are closed.
Change-Id: I41f863224fecd24069e525e9ce3738de8237bd5e
It's difficult to see in bugreports when this situation arises. Add a
small log so we can easily determine installation failure reason.
Change-Id: Ie59c205cf731cad7b3d04ceb995e58a093c62455
Animation steps are now executed on a Thread launched from the
Choreographer rather than being called at the end of the WindowManager
layout process. Animations and layout are still tightly coupled in
that they share considerable state information and neither can be
executed without holding a lock on WindowServiceManager.mWindowMap.
Change-Id: Ie17d693706971507b50aa473da1b7258e9e67764
