Age | Commit message (Collapse) | Author |
|
Bug: 37208805
Test: compiler, on device
PiperOrigin-RevId: 162268272
Change-Id: I98d63d063b9a4dff6a1b1b7462378ef7d0139bd2
|
|
When a user would delete their last contact, the contacts fragment would show
a list with the old contact and the empty contacts background view underneath
it.
Test: ContactsFragmentTest
PiperOrigin-RevId: 162267124
Change-Id: I294f2fe06f08ba6bd31148681647ffe3d536e1bf
|
|
Since we are given the camera permission by default for video calling, we don't
need to request it from the framework. But, for privacy reasons, we still need
to tell the user that we have it.
Since we're changing the preference for call composer, v10.1 users will see this
this toast again if they've never made a video call.
The toast is shown in the following situations:
1. If a user receives a video call
2. If a user makes a video call
3. If the user opens call composer
The toast doesn't show:
1. If the user has already seen the toast anywhere before
2. If the user has revoked the system permission and grants it
with a system dialog
Bug: 36608790,63405063
Test: manual
PiperOrigin-RevId: 162258610
Change-Id: Ie93413c9c2e1f083919e7183eb920100b06fd4a4
|
|
* changes:
Fix Lightbringer call upgrading when Dialer was not in memory before the call
Handle null host for secret code
Dont start a service to cancel missed call notifications, use DialerExecutor instead
Fix NPE when user click on call log.
|
|
|
|
We need to load reachability data if we do not already have it when checking
for upgrade. Previously we only did this for reachability.
Bug: 63757003
Test: LightbringerImplTest
PiperOrigin-RevId: 162246049
Change-Id: I9b4919e938b814ea18dc5107967bb6de7a945c19
|
|
The "host" is the number in the "*#*#NUMBER#*#*" format. It was assumed that it could never be null, but silent crash reports proved it wrong. Repro steps are unknown.
Bug: 63711549
Test: PawSecretCodeListenerTest
PiperOrigin-RevId: 162232903
Change-Id: I15419178099cea7f05c1c9ed5cd595a01ae8db2b
|
|
instead
We cannot start services while in the background on O. Currently, we start a
service to cancel missed call notifications in the onStop of our Activity. This
is when we are going into the background so it is not safe (as we have seen
from the crashes).
Bug: 63633461
Test: CallLogNotificationsServiceTest
PiperOrigin-RevId: 162029424
Change-Id: Ib0f46aed848ba0898af8cbee1c114b1e41f3ae32
|
|
This is caused by null tag of videoCallButtonView when expanding action view.
Bug: 63710739
Test: none
PiperOrigin-RevId: 162011496
Change-Id: If592be8cd9b05c369e2ff8b51610d93af48834f1
|
|
This allow new carriers to test their values or help debug/QA for situations like disabling prefetch.
Bug: 63013427
Test: N/A use at your own risk.
PiperOrigin-RevId: 162003756
Change-Id: Ifb227163bc3f2a94c1c07cc60fdc3a031fd625c8
|
|
This CL makes a number of changes to include temporarily unavailable
statuses (go/ec-temp-unavailable) as part ECCapabilities.
- Adds isTemporarilyUnavailable methods, which will be set by
ECBroadcastReceiver upon getting a capabilities update.
- Adds a builder to ECCapabilities as there are now 4 booleans that are
set upon construction.
- Changes the names of getters on ECCapabilities, which was required to
get autovalue to correctly generate the builder.
- Adds an ALL_CAPABILITIES constant to match the NO_CAPABILITIES
constant.
Bug: 62609419
Test: Existing tests, ECCapabilitiesTest
PiperOrigin-RevId: 161890511
Change-Id: I967a9a14793ac12101ecfed59e5f7af1217faf3b
|
|
This is just a code style clean up.
Test: none
PiperOrigin-RevId: 161885792
Change-Id: I6827f562faeca92f304efbfdb2ff35c918915a26
|
|
Before, we were creating a custom thread pool for spam transport, which isn't necessary. Instead, we can just use an application wide thread pool.
Also fixed some instances where we were not clearing the thread stats tag after opening the socket.
Test: manually launched app with data cleared and observed no crash
PiperOrigin-RevId: 161882076
Change-Id: I39bdd31cf5fa8a974d7535e861ec0716c85986f3
|
|
We now have googledialer, aosp dialer, and voip registries for traffic tags which should always be used to avoid reusing tags.
There are comments in each file about the reserved tag ranges.
Note that this moves some traffic tags from one value to another. Traffic data is cleared across device restarts so that shouldn't be a problem.
Test: unit tests verify no conflicts
PiperOrigin-RevId: 161848949
Change-Id: I5b51e6242494a0bc211748f793a415a5fead5097
|
|
Bubble showed call ended text and didn't dismiss when requesting/receiving video call upgrade. Fix it by showing text only when call is really disconnected and always hiding.
Request upgrade, declined and then accepted:
https://drive.google.com/a/google.com/file/d/0Bz1rQbdSCWSKZURkNWVOMnNnZHc
Receive upgrade, declined and then accepted:
https://drive.google.com/a/google.com/file/d/0Bz1rQbdSCWSKWGhJbi1uR004REE
Test: manual
PiperOrigin-RevId: 161750763
Change-Id: Iefb58edfbfffc33ee5231ef4928f31202e2880f4
|
|
The Hebrew translation had an extra single quotation causing an
UnboundQuotationException. This change fixes that temporarily by looking for
that specific translation and adjusting it. This change should be reverted once
the translation is corrected.
Test: release build in hebrew results in no crash, manual test
PiperOrigin-RevId: 161730025
Change-Id: Ifb171e3e795e91a93096493ec770ae339a6ed12f
|
|
A subsequent change will move the ContactPhotoManager related code,
eventually enabling the removal of contacts/common.
Bug: 37208805
Test: TAP
PiperOrigin-RevId: 161724700
Change-Id: Ice0789951ae544f6e27bcdaab0c032b218c83613
|
|
Bug: 37846172
Test: on device
PiperOrigin-RevId: 161720359
Change-Id: I2df2036c6d9941d37511f450f58682cf07e774b0
|
|
This is temporary solution to avoid expansive operation for frequent method calls. In long term we should figure out a better way for such cases.
Bug: 62424638,62426694
Test: manual
PiperOrigin-RevId: 161717598
Change-Id: I75e6589174b81ae611893e5435a94e7f629f500f
|
|
video: https://drive.google.com/open?id=0B2Hce9qilHmvdl8xY1hmcDZ0R3M
Test: manual
PiperOrigin-RevId: 161713747
Change-Id: I8185ca897a82479ef7c552a1ce6891e31c4c5aad
|
|
screenshots:
default sizes:
before: http://screen/u29zyepX0OM
after: http://screen/tXGSUbNfWAE
large sizes:
before: http://screen/NH3OmGDRyBp
after: http://screen/W6DVopCxMUp
Bug: 63156716
Test: screenshots
PiperOrigin-RevId: 161693857
Change-Id: I8310336080ae080dc586dcf21d4966260880a627
|
|
Without this change, all video calls are referred to as just "video call". This
CL uses the Lightbringer interface to allow customizing that text.
Before: https://drive.google.com/open?id=0B7uuA4cyYX0xeVZCTGtMUUtoRVU
After: https://drive.google.com/open?id=0B7uuA4cyYX0xMnFhbTBXMDI2VW8
Bug: 63138393
Test: CallTypeHelperTest
PiperOrigin-RevId: 161692812
Change-Id: I36dc1a1fae96dddee91c5efb8892c4a6c7ef67ca
|
|
This issue occurs when users are shown a post call snackbar and their shared
prefs are updated before pressing "send message". This change resolves the
issue by getting the phone number when the prompt is created rather than when
the button is pressed.
Bug: 62294499
Test: PostCallTest#postCall_UpdatedQuickly_PostCallStartedSuccessfully.
PiperOrigin-RevId: 161689241
Change-Id: Ie2c857f6743aa556f993bc3f8e92c8b2b7357c08
|
|
If a user caused their activity to undergo several state changes rapidly, they
could get the contacts fragment into a state where it wouldn't have any
elements but still call onScrollChange. This would result in calls to
getFirstCompletelyVisibleItemPosition returning NO_POSITION (-1) and throw an
AIOOBE when attempting to get a row's header value.
Bug: 63594129
Test: manual.
PiperOrigin-RevId: 161610423
Change-Id: I0c53587a6499c42abe5e51b6ac2b4108c0971545
|
|
before: http://screen/DsVrM1vK89e
after: http://screen/RzGZvPTXRQb
Bug: 63093275
Test: manual
PiperOrigin-RevId: 161609906
Change-Id: If53681b9414dd79dba16371f42be437f1afa2729
|
|
The Call button may have two lines of text. We were not properly setting the
second to GONE in all cases.
Note: We don't need to require Google Dialer being the default Dialer for the Duo integration to work. I added this check so removing it doesn't go against any previous well-considered decision. It also enables the Espresso test work without needing a flag to override.
Before: https://screenshot.googleplex.com/3YXaZdbQk7k
After: https://drive.google.com/open?id=0B7uuA4cyYX0xNThETTJWdTVQQWM
Bug: 63062360
Test: GoogleDialtactsActivityTest
PiperOrigin-RevId: 161606497
Change-Id: I7526a4fc60b84906cc04563b635eaad9f348415e
|
|
If a user took a picture then quickly closed call composer, when the image was
ready we would try to update our view state. Since our fragment is no longer
attached, this cuases a crash when we check that getContext() is not null.
fragment.getContext is never null in robolectric.
Bug: 62687110
Test: manual, cannot test b/c we cannot take photos in Espresso and
PiperOrigin-RevId: 161600278
Change-Id: If9bd98578d221fca4bc99ff17a39f917f3a8bcca
|
|
Test: N/A
PiperOrigin-RevId: 161596844
Change-Id: Id4d8a70fe5690e05dc6bad1aca50a1a53f031a58
|
|
This is caused by setting call state to OUTGOING after it's created but before
it's connecting, which incorrectly switches state as NO_CALL -> OUTGOING ->
PENDING_OUTGOING -> OUTGOING.
The fix is giving it PENDING_OUTGOING state if an outgoing call is initialized
but still not yet getting update from telecom to CONNECTING (in which case the
call will be IDLE state).
Bug: 62965119
Test: manual
PiperOrigin-RevId: 161594187
Change-Id: I02c8b4dcb0bfd5101bc69f48bb99c66d4831b72e
|
|
OEM and carriers need this line to verify.
Bug: 62677283
Test: N/A
PiperOrigin-RevId: 161594185
Change-Id: I00cf649a745a45a14fcea9948f1f8c696fb4b6b1
|
|
It was odd that we were casting and was causing some crashes. This CL makes
things work more as is typical in Dialer.
We previously did not do this because there are some parts of the
VideoShareSession which are not open-sourceable. Turns out we can simplify
those parts out of the interface.
Bug: 63523694
Test: existing tests
PiperOrigin-RevId: 161593028
Change-Id: I8f1379fc46f4e9d41413b731787dbf37e0901da9
|
|
VVM capability can change between dialer versions. For v10 the update is done through system image, so when the user get the new dialer they are guaranteed to received a onCellServiceConnected event() after boot, and activation will be triggered. This does not apply to play store update or side load.
In this CL, activation will also be triggered when the app is updated.
Bug: 35809267
Test: PackageReplacedReceiverTest
PiperOrigin-RevId: 161592369
Change-Id: I6904ef0514f9eec8151cd28be8cf76255640b108
|
|
The dialer app crashes on launch right now when app data is empty, because we fetch the spam list on app start, and we don't tag the network traffic using TrafficStats, which is a strict mode violation.
This CL tags spam network traffic using TrafficStats, by providing a custom transport executor for the GRPC channel. All activity on the channel is tagged as spam.
This unfortunately required creating a new thread pool. I am not sure if this creates more threads than before, because it is not clear what type of transport executor grpc uses by default. Note that the new thread pool will kill unused threads after 60 seconds so won't consume resources when idle.
Bug: 62388129
Test: manually verified app does not crash on launch with empty data. Can add automated tests upon request.
PiperOrigin-RevId: 161568882
Change-Id: I99da6ebb649fe1a63d634871ab314d8d910658f9
|
|
Add additional null checks during deseriailzation.
Bug: 63575857
Test: unit tests, on device
PiperOrigin-RevId: 161564824
Change-Id: I54f52e12397adb4473b523325f8c006ff534fbd9
|
|
Test: ondevice
PiperOrigin-RevId: 161548092
Change-Id: If4cde06669bc33cca69a9b602f54d873b735c730
|
|
This CL attempts to improve many related issues.
The first was that the (post-greenscreen) preview was "squished" when making an outgoing IMS video call:
BEFORE: https://screenshot.googleplex.com/cRcZYmgq1rh
AFTER: https://screenshot.googleplex.com/8cWFQw7Au2U
Another issue is that the preview on the green screen sometimes appears very zoomed:
BEFORE: https://screenshot.googleplex.com/0vyq3U87xVX
AFTER: https://screenshot.googleplex.com/zyhkdATMuUj
These two issues seem to be improved by removing our manual attempts to scale the video via VideoScale. (Note that transforms to fix rotation are still needed in landscape mode and remain.)
Another issue is that when hanging up an unanswered outgoing video call, the preview becomes temporarily extremely zoomed and unblurred:
BEFORE: https://screenshot.googleplex.com/mqwSsTXhwSw
AFTER: https://screenshot.googleplex.com/uKfEpVmd8A2
Another issue is that when rotating the device, the preview would sometimes remain rotated incorrectly.
BEFORE: https://screenshot.googleplex.com/p2mVnPJ7dww
AFTER: https://screenshot.googleplex.com/S8R0FsS0Vsn
I believe that these problems (and possibly other video related flakiness I sometimes observed) happen because we update scaling and views immediately after applying layout changes. Often times, the layout changes haven't taken effect because we're in the middle of a layout pass so the updates are not working correctly. So, I moved most of those updates into specific layout listeners for the preview and remote texture views.
(Note that something similar before was attempted using a global layout listener, but that layout listener removed itself after the first invocation so didn't seem to be working as intended AFAICT.)
The last issue was that when toggling the front/rear camera, when returning to the front camera, the video would appear zoomed. This was fixed by removing the call to setDefaultBufferSize in VideoSurfaceTextureImpl. From the javadoc of that method, it doesn't sound like something that we should need to be doing and it reliably fixes the problem.
BEFORE: https://screenshot.googleplex.com/6j0XDfLGAzk
AFTER: https://screenshot.googleplex.com/Hs7zsbtyjwc
Bug: 62437862
Test: manually placed and received IMS calls and observed improved preview scaling on multiple devices when calling from O to N, O to O, and N to O. Observed that basic features (mute/umute video, swapping cameras, greenscreens, answer screens, hanging up, etc) still behave reasonably. Open to suggestions for automated tests.
PiperOrigin-RevId: 161444534
Change-Id: I4be348875de11b8517feba86da07fe41a3e5351d
|
|
Although we don't know the reason why v.getParent() can be null, we can avoid calling it.
Test: manual
PiperOrigin-RevId: 161442696
Change-Id: I07af0da9b64fb3fa77b01c0b619837a79d593b67
|
|
Methods such as getAccountLabel() is called on different threads which may cause
race condition issue.
This change also delete CallLogCacheLollopopMr1 since it's not necessary that Dialer is targeting M+.
Bug: 63415147,63524435
Test: none
PiperOrigin-RevId: 161440757
Change-Id: Ia609c52e53dabdce78ffb4320f4cd66e38112e47
|
|
OC preview devices are hitting the assert and skewing crash rate numbers.
Bug: 62338925
Test: LegacyVoicemailNotificaitonReceiverTest
PiperOrigin-RevId: 161438516
Change-Id: Ib533947d2cd9e9a87ffd9fb629f09f877f683026
|
|
In b/63143138 we discovered an issue with the LetterTileDrawable code.
The class has static fields that are shared across both the main and
background threads, which aren't threadsafe. We attempted to make the
code threadsafe in cl/160696979, but that ended up causing issues as
well.
This CL takes a different approach. Rather than trying to make the code
threadsafe, it instead removes the shared fields and makes them final
and initialized in the constructor. This is inherently threadsafe since
threads cannot interact with the object until it is constructed.
Now that the objects aren't shared, we would potentially use more
memory, since each LetterTileDrawable will now hold onto an instance of
the formally shared object. This was not determined to be an issue
in practice, see the data and conclusion in
(go/lettertiledrawable-caching).
Bug: 63143138
Test: LetterTileDrawableTest (can't verify everything :/ )
PiperOrigin-RevId: 161432711
Change-Id: I443bef01ffac405f5e14e9d981ce1d5a8f3f8fec
|
|
* Don't allow cleartext traffic in the voicemail module. This was being overridden, anyway.
Bug: 62339559
Test: See below.
PiperOrigin-RevId: 161418474
Change-Id: Icd945902ed089bf37b43d93d8fe32983a9f88fea
|
|
This CL bumps the version name of Dialer from Dialer v11 to Dialer v12. This CL
also bumps the version code from 15***** to 1600000.
I've also set 'version_conf_incrementer_max_version_code' at go/dialer-v11
to 1599999.
Test: N/A
PiperOrigin-RevId: 161418347
Change-Id: I961f22481d813d66b950894b3783e13624de733e
|
|
We cannot use ConfigBindings as full backups happen with a base Application, not the app's custom subclass.
Bug: 63104326
Test: N/A
PiperOrigin-RevId: 161398864
Change-Id: I7f559fe7cfd8bfdd65f49efea943f3f75488d1a6
|
|
This will prevent merged conference having participant with name as following:
1. Restrict and non-restriced
"<phone number or name>"
"Restricted 2"
2. Two restricted number
"Restricted 3"
"Restricted 4"
After the change it will become:
case 1:
"<phone number or name>"
"Restricted"
case 2:
"Restricted 1"
"Restricted 2"
Note: Unfortunately there isn't a way to relate the pre-merge restricted calls to the post-merge restricted calls. So "Restricted 2" is not necessary the one pre-merge. This also means reset is safe here for merge.
Bug: 63315372
Test: manual
PiperOrigin-RevId: 161267675
Change-Id: I2df16b6f9802932c06f0bdd62618c49f4e6e87a8
|
|
The dialer voicemail UI will send the broadcast even below O.
Ideally all receiver for system and external broadcasts should be moved out of impl/ since they will not be disabled with the module, but at this point changes should be kept at minimal.
Bug: 63437675
Test: VoicemailClientReceiverTest
PiperOrigin-RevId: 161264508
Change-Id: I99476bb27ce3da9396f3b7205c98cb847bfff9b7
|
|
The tests now call through the click listener to expand instead of just calling
the expanding method directly. This was done after we had a bug come up inside
the click listener that these tests did not catch.
The bug: Lightbringer.getPackageName can return null, so when using a stub there was an NPE inside the ExpandCollapseListener.
Test: only tests
PiperOrigin-RevId: 161121777
Change-Id: Ie58a4c14a32984c183b6159db2076a96cf46b37e
|
|
This is changed in ag/2484808 to be defined by a constant telecom string. It
will be available to public in O MR. We have to hard code it until then.
(This is a take over of cl/161020406 from wangqi@)
Test: VideoCallNotAvailablePromptTest
PiperOrigin-RevId: 161030078
Change-Id: I0e1d8d98cb1f64b035d9063b28f8ba039c7378e4
|
|
There's a bug in O notification framework that the HUN(heads up notification) will always show when a notification with fullScreenIntent is updated.
Previously NOTIFICATION_INCOMING_CALL_QUIET is only used when the call is already active when the new incoming call is received. NOTIFICATION_INCOMING_CALL_QUIET will post the notification to the ongoing call channel instead of the incoming call, which in theory will have lower importance and the HUN will not show. In this CL, all incoming call notification will become "quiet" if the InCallUi is visible.
Bug: 62458080
Test: StatusBarNotiferTest.java
PiperOrigin-RevId: 161029526
Change-Id: I671105ecfc102beb43e52cf4d91d55a66d31cd0e
|
|
The reason of disconnect cause could be null.
This is regression caused by cl/160563754
Bug: 63156395,63302714
Test: manual
PiperOrigin-RevId: 160987545
Change-Id: I3fd726947601f55d5dc917b11626b64bfc9d0135
|
|
*** Reason for rollback ***
Causes crash on launch (b/63252565)
*** Original change description ***
Fix concurrency issue in constructor.
Before:
sColors served as the branch to initialize all member variables.
Subsequent calls to the constructor after sColors had been initialized,
would result in the use of static members that were not yet initialized.
Now:
We guard each reference with an explicit null check during construct.
Members that require synchronized initialization were put into a small synchronized
method.
This was chosen instead of AtomicReferences, and instead of double check...
***
Bug: 63143138
Test: N/A
PiperOrigin-RevId: 160837963
Change-Id: I2586f7586c8f39182d64c3b28a59886c5ba94789
|
|
Test: cl/160695183
PiperOrigin-RevId: 160697524
Change-Id: I07127554a968d8dc36c02b522b7536438be7fb90
|
|
Before:
sColors served as the branch to initialize all member variables.
Subsequent calls to the constructor after sColors had been initialized,
would result in the use of static members that were not yet initialized.
Now:
We guard each reference with an explicit null check during construct.
Members that require synchronized initialization were put into a small synchronized
method.
This was chosen instead of AtomicReferences, and instead of double checked locking,
because -- although verbose -- we can tolerate a small number of overwrites to
each member variable (they are idempotent within the Application scope), thus avoiding the need for any one thread to wait/acquire a lock, as well as avoiding the need to import the Atomic library (which added no incremental benefit).
It is assumed that the JVM will not garbage collect overwritten references to
member variables that are still in use across instances of LetterTileDrawable.
This shouldn't matter because the references are volatile, anyway.
Initialization-on-demand was not available due to the use of non-static
resources on construct.
Bug: 63143138
Test: No. Old unit tests.
PiperOrigin-RevId: 160696979
Change-Id: Ie17a29e48f91cb3df07d81d29b6428b70fb4408a
|
|
The original code is somehow triggering when the VVM tab is not selected.
Bug: 62517716
Test: VisualVoicemailCallLogFragment is utterly untestable
PiperOrigin-RevId: 160692310
Change-Id: Ia9da8e5aa7dd28f22c46d87dd25ebf6097918a9a
|
|
When telecom disconnect the video call with the code of DisconnectCause.ERROR and the DisconnectCause#getReason() of "IMS_ACCESS_BLOCKED", we should show an error dialog to user indicating that video call is not available and providing a option to make a voice call instead.
Screenshots:
https://screenshot.googleplex.com/YTUGVouoLmj
Bug: 33821263
Test: manual, VideoCallNotAvailablePromptTest
PiperOrigin-RevId: 160563754
Change-Id: Ie2eb735a92c577b5ae5a5e8b7efa2d699fc964bc
|
|
This CL bumps the version name of Dialer from
also bumps the version code from 14***** to 15*****.
go/dialer-v10_1 did not have a version_conf_incrementer_max_version_code set, which let it to share the version code increments between v10.1 and v11. There are version codes in v10.1 which are higher than v11 which will lead to users downgrading when we submit v10.1 to the playstore (http://cl/160476738)
To make sure users are not downgraded from v11 to v10.1 when we push to the play store, I've also set 'version_conf_incrementer_max_version_code' at go/dialer-v10_1
to 1504907.
Test: N/A
PiperOrigin-RevId: 160555974
Change-Id: I4a9e64aa2fe461019e179f544fe959b8389a02cf
|
|
*** Reason for rollback ***
Possible cause of Dialer crash: b/63128348
*** Original change description ***
Add accessibility info for bubble.
***
Bug: 62886248,63128348
Test: N/A
PiperOrigin-RevId: 160531267
Change-Id: Ia9bf5fe29523d6f6280e4ed6fcf9a8d23ee72579
|
|
*** Reason for rollback ***
Causes crash, see b/63127884
*** Original change description ***
All Fragments are now support library Fragments.
Switch a bunch of imports to support library Fragments (and Loaders) instead of framework Fragments (and Loaders). This was needed to use FragmentManager.isStateSaved() instead of manually tracking that state. This should reduce errors around committing after state is saved.
Also adds METADATA presubmit to prevent new usages of framework Fragments or Loaders.
***
Bug: 38241305,63127884
Test: N/A
PiperOrigin-RevId: 160530287
Change-Id: Ie2c4d3561ebea2d0f05097ce89bdd9bc8767c8cf
|
|
I think this was inadvertently (?) disabled.
Test: no
PiperOrigin-RevId: 160455622
Change-Id: I9f00331d7ef53108eb0adb9bec9c91be22422d2f
|
|
The state of a session is more closely tied to the session itself,
rather than the EnrichedCallManager. The enum was originally placed in
the Manager as it was used in the API, but now that's no longer the
case.
This CL moves the Session State declaration into Session, which is a
more appropriate location.
Test: tap
PiperOrigin-RevId: 160448612
Change-Id: I907476140bc174a061d48ee8e58878af64a62b29
|
|
The old Log class uses the calling class name for the tag. Proguard
minifies class names, making it annoying to figure out which class is
logging in a bugreport. You'd see something like:
06-14 08:50:10.942 10022 20233 20233 I Dialer : bhp - Phone switching
state: OUTGOING -> OUTGOING
Now you'll see
06-14 08:50:10.942 10022 20233 20233 I Dialer :
InCallPresenter.onCallListChange - Phone switching state: OUTGOING -> OUTGOING
Test: Code builds, no changes were made to the actual logged strings
PiperOrigin-RevId: 160445433
Change-Id: Id4d07a33b9ec7a9bceca48e71cdcfc5e6291b571
|
|
b/38149414 building with dreamchip regressed startup time by 400ms. This seems to be caused by linking against the library with cc_library, but the true reason is still unknown.
In this CL, dreamchip libraries is stored as an asset instead. During runtime the asset will be copied into the data directory then loaded.
Currently the copying happens on the main thread. Whether if it's acceptable to introduce further risk to handle this is to be discussed. Disk IO could already happen before this change.
Bug: 38149414
Test: N/A
PiperOrigin-RevId: 160444223
Change-Id: I9f098e7b79df113b2992d8590b33f423942716e4
|
|
populated. Will just return null in createAnimation() thus not animating (this will likely only happen if the View hasn't laid out yet, so there's nothing to animate from).
Bug: 63054791
Test: ChangeOnScreenBoundsTest.java
PiperOrigin-RevId: 160433277
Change-Id: Ib87d274fd6f109290ec74f2241e2fb559d6138cd
|
|
Including in:
- expanded call log item
- collapsed call log item
- search
- in call UI
LOG_STORAGE_INCREASE(GB/week): 3.15
5M active block user/day * 10 events/active = 350M event/wk
350M event/wk * (3 * 3 [int32])/event = 3.15GB/wk
Test: GoogleCallLogAdapterTest, ContactListItemViewTest, InCallFragmentTest
PiperOrigin-RevId: 160423797
Change-Id: I6c0ade53caf767ea9b2610a4c0eb68fcc2bcedf7
|
|
Bug: 62886248
Test: Manual with talkback enabled
PiperOrigin-RevId: 160349111
Change-Id: I9eaaac90b93b7e6f3a002288d0d68fbbe3fae56c
|
|
Bug: 63012482
Test: CallLogCacheLollipopMr1Test
PiperOrigin-RevId: 160303077
Change-Id: I00c80a681963c5a7a25fffd04e02db32ca2386dd
|
|
*** Reason for rollback ***
This CL causes a crash when swapping SIMs.
*** Original change description ***
Adding voicemail transcription branding and progress UI
screen shot showing branding (its harder to capture the progress UI):
https://drive.google.com/open?id=0B9o_KvtLkcuIdkkycVo1RFhsaENYV3J2Yi1LWnJzR0FfSHJR
***
Bug: 63059930
Test: N/A
PiperOrigin-RevId: 160296762
Change-Id: Ib8d709d03d6d8a3fee4a0c384c15df00f55cb182
|
|
The request to stop transmission can be in-flight at the time Dialer goes into the background so the pause request will still include Rx. This happens because pausing previously just gets the current video state and adds pause.
Bug: 62990728
Test: ImsVideoTechTest
PiperOrigin-RevId: 160210111
Change-Id: Ie9134c285a73d4348036f6bbb847abae53a1d373
|
|
This prevents accidental logging.
Test: GoogleCallLogAdapterTest, CallLogListItemViewHolderTest
PiperOrigin-RevId: 160207870
Change-Id: Ib0582613979114c24bb0bd38df41ad7d7dadb63a
|
|
Switch a bunch of imports to support library Fragments (and Loaders) instead of framework Fragments (and Loaders). This was needed to use FragmentManager.isStateSaved() instead of manually tracking that state. This should reduce errors around committing after state is saved.
Also adds METADATA presubmit to prevent new usages of framework Fragments or Loaders.
Bug: 38241305
Test: Covered by existing tests.
PiperOrigin-RevId: 160202511
Change-Id: I5dc0b467da984ca571e49b4dfa0deea44eb372b2
|
|
The previous fix (cl/159623587) has a regression in this scenario:
Device A: video call Device B
Device B: accept video call
Device B: go into background (press Home)
Device A: go into background (press Home)
Device B: return to foreground (return to call)
Unexpected result: A is transmitting video even though it has not returned to the foreground as well
Bug: 62784036,38359325
Test: manual
PiperOrigin-RevId: 160013824
Change-Id: I6e9c89c8fa52882c49d32e061032b93b1e8552f0
|
|
• moved scrollbar to edge of screen
• changed divider to .5dp and added shadow
• change "DECLINE" to "NO THANKS" for verizon TOS (it already was that way for non-verizon)
• changed "ACCEPT" button to "TURN ON" text
• increased the bottom bar height
• made the google transcription text in the TOS bold
• add an image to the non-verizon TOS screen (just using test image for now)
i didn't add the 'Learn more' link, since its not ready yet.
verizon screen shot:
https://drive.google.com/open?id=0B9o_KvtLkcuIcVk0dHBtWmFfdjJWQVV3OEdVc0JXN01XZHQ4
non-verizon screen shot:
https://drive.google.com/open?id=0B9o_KvtLkcuIci1OMlVXTHVuMGYwMWZVTS01dGVwMUpLdmRr
Bug: 62375681
Test: manual and updated unit tests
PiperOrigin-RevId: 160006810
Change-Id: Id45a3a848bb219a70fddbb5a7ada29cd39e604a9
|
|
screen shot showing branding (its harder to capture the progress UI):
https://drive.google.com/open?id=0B9o_KvtLkcuIdkkycVo1RFhsaENYV3J2Yi1LWnJzR0FfSHJR
Bug: 62376944,62424455
Test: device testing
PiperOrigin-RevId: 159993127
Change-Id: I8355164b5831e85de13915e221f6e0f0163e8c81
|
|
Logging impressions for interactions with the voicemail transcription server.
Bug: 62423554
Test: none
PiperOrigin-RevId: 159985772
Change-Id: I782d7eb4df917ae42940f359d98abc87f7ec0f95
|
|
screenshots:
short message (before cl): http://screen/pMAgfV540Y5
short message (after cl): http://screen/VN3TVTBYjPN
long message (before cl): http://screen/V8jSApXmEC2
long message (after cl): http://screen/Vd7zyf2JGvP
long message zoomed (before cl): http://screen/urWDNGACgOf
long message zoomed (after cl): http://screen/i2xEch8HHoq
PiperOrigin-RevId: 159968371
Change-Id: Icaf0a69949b206efd8d08001065cfb4ccdcb1726
|
|
By not checking CallUtil.isVideoEnabled, we were offering upgrading even after the user had disabled it.
I also tried to make the isAvailable method a bit more readable.
PiperOrigin-RevId: 159966735
Change-Id: Ibf828a561654e0e9521a6d86ad18269da426bba4
|
|
There is still one more violation in hangouts that i will address in a separate cl.
PiperOrigin-RevId: 159955865
Change-Id: I61cb65d5f099baae497dfe2daddb4f7da6866ad6
|
|
Replace the null "client type" value by a more detailed value in the MO SMS.
Create a client type value based on :
- Manufacturer name
- Model name
- Android os version
Test: OmtpConstantsTest.java
Original author:
http://aosp/319963 - Alexis MARCOU <alexis.marcou@orange.com>
PiperOrigin-RevId: 159906976
Change-Id: If759f7c96c9fb9a3895f8d944c842c350b8d2472
|
|
If the user disallowed the "Display over other apps" permission then
incoming calls would no longer show up on the lock screen.
The problem was that if the permission was denied then we would attempt
to ask the user to give us the permission. This causes us to start an
activity while the in-call activity is starting which break things.
Fix was to never prompt the user to give us this permission. This is ok
because not having this functionlity doesn't break anything.
PiperOrigin-RevId: 159901233
Change-Id: I3c06ba327ef265fd7981931f49c44e00d1c25350
|
|
Test: IconFactoryTest.java
PiperOrigin-RevId: 159871437
Change-Id: I8df921f36bde619811129ae0f3013ff77903fc8e
|
|
This fixes two bugs we previously had:
1) If video calling was disabled in Settings, we would still show a video call option in the call log (fixed by checking the setting)
2) If a different app from Dialer inserted a video call into the history, we would incorrectly show a video call button in the call log (fixed by checking the account handle matches the default phone account)
PiperOrigin-RevId: 159870114
Change-Id: I19ef6b50db50f9961a486f48996427a1da8f8813
|
|
PiperOrigin-RevId: 159845977
Change-Id: I5615960a4d007924d3db64fc777c7a40593df612
|
|
|
|
https://googleplex-android-review.git.corp.google.com/#/c/2446830/
Test: TH, make
Change-Id: I39aba7e972bac6e5864e70ed693849d90ecd7e08
|
|
This will better handle the cases where a show/hide action is performed while the Bubble is animating.
PiperOrigin-RevId: 159771812
Change-Id: I87779da8c0360906587355279b3cc9674289e30f
|
|
Reports when a user sucesfully reports a number is inaccurate caller id and
when their report fails.
PiperOrigin-RevId: 159762210
Change-Id: I8507b4d2a3d4b6f3004e8145f4ba0f9e8065e5d7
|
|
Log which tab is presented when app first launched, including:
- speed dial
- call log
- contacts
- voicemail (if not available, we move to contacts tab immediately and log the
move)
Test: manual
PiperOrigin-RevId: 159754375
Change-Id: Ice5be6c0f2c0463204fe3c1c7cf83c4dfc7df5d1
|
|
When a user uses call composer for the first time, it isn't obvious that their
messages are being stored in the Android Messages app. So we display this toast
to them on the first time they make a enriched call to let them know for
privacy purposes.
screenshots:
http://screen/pcuu4GG037Z
http://screen/MLRJiY9V7cH
PiperOrigin-RevId: 159749893
Change-Id: I0cee7b49321f1d2302ba8ed8f6e1c19c44feb19a
|
|
Current auto fill content is usually the device's own number thus not useful
and it covers "1" key.
PiperOrigin-RevId: 159742824
Change-Id: I753832120bf9436506bf5394b47c94f94899a92b
|
|
b/62666193 is one of the top crashers, the assert that the VVM config is valid has failed. This is likely caused by the PhoneAccountHandle being invalid. While the PhoneAccountHandle must be valid when the broadcast is made, the SIM might have been removed and invalidated the PhoneAccountHandle when the broadcast is still in flight.
In this CL the validity is checked first.
PiperOrigin-RevId: 159737846
Change-Id: Iae5a65978879fd1bfdf726008e8b6463499c99e8
|
|
An empty string isn't as clear and can lead to incorrectly matching other
intents when just comparing package names.
PiperOrigin-RevId: 159725546
Change-Id: I2943b3569fb6409f5c69deb2032f3d8248f6c14d
|
|
TEST=N/A
Video: https://drive.google.com/a/google.com/file/d/0ByIs9UmqOFZFZjBsUS1HQVBxUUE/view?usp=sharing
PiperOrigin-RevId: 159637501
Change-Id: If1b088f7abff2ec52fd54fbbb36d6b7c6d54a926
|
|
We need to check that we have a Context before passing it to
RenderScript.create. getContext can return null if the fragment has been
removed.
I switched this to an early return approach to avoid the deep nesting and added tests.
It is safe to ignore the IfThisThenThat
PiperOrigin-RevId: 159633511
Change-Id: If2286a545c2e514c328dff22c48a99b9ed232e88
|
|
When a user deletes a call from the call log, it might not be clear that they
aren't deleting the attachments they made with the call. So we present them
with a snackbar letting them know to open android messages to delete
attachments.
screenshot: http://screen/TSoWmNhjcn4
screenshot: http://screen/vUXShdYpXem
PiperOrigin-RevId: 159631807
Change-Id: Iaad903258160dc789e0c6a524adba56299523636
|
|
This fixes a bug where we would remember being paused after receiving a second
call incorrectly after we swap back to it.
@tgunn, adding you here to double-check I am interpreting those call session events correctly.
PiperOrigin-RevId: 159623587
Change-Id: Ic52e3e9046f767321a48e22274b8a1c27b4dfa86
|
|
PiperOrigin-RevId: 159622902
Change-Id: Idabc3122601e442a07b31c38287af767651e9970
|
|
The WindowRoot now provides a listener for when the Configuration state changes. Bubble will then reinitialize the MoveHandler (to account for potential changes in screen bounds) and ensure the window is within bounds.
PiperOrigin-RevId: 159614123
Change-Id: Icace5b52299825836fb398ce1e6234ad0bb212f3
|
|
|
|
e4de90c6d9
am: cccd0a8597
Change-Id: I56fa753c4191d6969b75f4e22b2013eb56655a9f
|
|
|
|
*** Reason for rollback ***
It's not necessary anymore after we verified it could be controlled from server side.
*** Original change description ***
Flip flag to enable ReturnToCall bubble.
Enables ReturnToCall bubble and changes the name of the flag so that it can be toggled independently from previous versions.
***
Also in the change: Restore TODOs
Test: TH
PiperOrigin-RevId: 159587541
Change-Id: I88f01550f07a7859f41b4f0fc843e92b38f5a85b
|