Make sure the appearance or the behavior controlled by the new APIs
won't be affected by setting new LayoutParams.
This CL also makes ViewRootImpl#setLayoutParams won't modify the
caller's LayoutParams.
Fix: 175169461
Test: atest ViewRootImplTest
Change-Id: I4cdd846e8638f9813146e6461134e4dd1606bd50
Since SurfaceView no longer enables blast sync transactions, a lot of
the code is no longer needed.
BlastSync is only set when ViewRootImpl calls relayout that returns the
blast sync flag. This means, we are guaranteed to pause the UI thread
since we don't want to allow overlapping relayouts. Because of this, we
can remove a lot of the complex code and variables.
1. We no longer need both mSendNextFrameToWm and
mNextDrawUseBLASTSyncTransaction. This is because if mSendNextFrameToWm
is true, we will pause the traverse and never set
mNextDrawUseBLASTSyncTransaction until mSendNextFrameToWm is set to
false.
2. Similarly, we don't need mRtNextFrameReportedConsumeWithBlast. This
was added to help with overlapping requests. Since that's no longer
possible, we can just remove this and use mSendNextFrameToWm as the
check
3. We also don't have to be thread safe with mRtBLASTSyncTransaction
since we only update it when the UI thread is paused.
4. We no longer need to call setNextTransaction with null since that was
only when SurfaceView was calling useBlastSyncTransaction and the main
window wasn't drawing. That no longer happens and we are guaranteed
that setNextTransaction is only set when a relayout occurs.
Test: Enable blast sync and use split/pip
Bug: 175229445
Change-Id: I0bebd15804ea3b960e85c20235ae5ce98bcc3f00
This change adds the default brightness to the ddc, and a fallback to
config.xml when the ddc doesn't exist.
It adds the minimum and maximum brightness constrtaints to the display
device config - which are currently sourced from config.xml.
Bug: 147415200
Test: manual
Change-Id: Ibbfbbbd495048114befb3f867bd5f4e26916ca9e
Since the display cutout is a type of insets and the display cutout can
be obtained from WindowInsets, it makes sense that InsetsState has the
display cutout instance. In this way, we can send the display cutout to
client via W#insetsChanged instead of W#resized.
This can be a step to remove the class of ClientWindowFrames, and can
also be a step to make client compute its window frame locally.
Fix: 175858810
Bug: 161810301
Test: atest WindowAddRemovePerfTest ImeInsetsSourceConsumerTest
InsetsControllerTest InsetsStateTest ViewRootImplTest
WindowInsetsControllerTests ActivityRecordTests
DisplayPolicyLayoutTests LaunchParamsControllerTests
TaskSnapshotSurfaceTest WindowMetricsActivityTests
WindowMetricsWindowContextTests WindowMetricsTest
WindowFrameTests WindowStateTests WmDisplayCutoutTest
Change-Id: I9a930b1d2f7df3cea2b29629b767a4a5f31bca17
The blur code in ViewRootImpl had a few issues
1. The transaction was applied or merged in frameCompleteCallback. This
is too late since the frame could have already been sent to
SurfaceFlinger and possibly latched already. We need to make sure we
apply or merge the transaction before latching. Therefore, moved the
blur request code into the frameCallback, which is invoked before the
buffers are swapped.
2. The blur code was using the blast sync transaction. This means the
blur would only get applied if a request to sync transactions was
called, which isn't very often. Instead, use mergeWithNextTransaction
which will send the transaction to the BBQ. The BBQ will either apply
it when the buffers are swapped or merge with a pending transaction that
is applied later to work with blast sync. This ensures the transaction
is synchronized with the correct frame
Test: manual
Bug: 174145445
Change-Id: Id36719c03f18739513e3198602c8a7ec4fd8da69
ag/12396996 introduces mInputToken, but it's only added to parcel
reading. It has thus caused a bug (parcel reading misalignment) in
the Wallpaper app as in our preview feature we passes a
SurfacePackage over from Launcher.
Test: manual
Fixes: 176248059
Change-Id: Ife46d6d67991964d63513c946318cffbaf4ffae9
Currently SV with blast sync requests a sync transaction from VRI and
merges its transactions with VRI's sync transaction. This can break if
we are not able to set the request in time.
Instead this change will merge SV transaction from the renderthread to
the VRI BlastBufferQueue specifying the framenumber which the
transaction should be merged with. The transaction will be queued until
we submit a VRI buffer with the requested frame number.
Test: open bubbles with sv blast
Test: atest SurfaceViewSyncTest
Bug: 175594838
Change-Id: I87ba346d782f615bab2e8a692adb2b3c49ef4021
- Add hasImeWindowSurface in TaskSnapShot
- Add InsetsSourceControl#getAndClearSkipAnimationOnce for skiping
IME showing animation once when starting window with IME surface
Bug: 166736352
Bug: 153145997
Test: manual as below steps
0) Enabling developer options -> Quick settings developer tiles ->
Window animation scale to slow down transition animation.
1) Launch an app with focusing an editor to show soft-input
2) Swipe out app task to back to launcher
3) Using quick switch or taping shortcut to bring back the app task
4) Verify that should be no IME showing animation happens during
task transition.
Change-Id: I83ffc03119e01da71ad12f8ad8043cf0730dfd50
1. Create InputMethodInfoList result callback and apply it to :
-. getInputMethodList
-. getEnabledInputMethodList
2. Apply IntResultCallback to getInputMethodWindowVisibleHeight
Bug: 163453493
Test: Manual test with keyboard
Test: atest CtsInputMethodTestCases
Change-Id: Iaa63e01453da4ff0e3f446eac036b3be3180cb73
Create InputMethodSubtype and InputMethodSubtypeList
result callback and apply it to :
-. getLastInputMethodSubtype
-. getCurrentInputMethodSubtype
-. getEnabledInputMethodSubtypeList
Bug: 163453493
Test: Manual test with keyboard
Test: atest CtsInputMethodTestCases
Change-Id: I646ef4ae0570aae1812ea267f309441fdec6938d
The issue is the following:
1. Call relayout from VRI. drawsNeededToReport = 1
2. SV also will need to be updated. drawsNeedToReport = 2
3. VRI is finished rendering, but SV is not. drawsNeedToReport = 1
4. VRI calls relayout again. drawsNeedToReport = 2
5. SV finishes drawing. drawsNeedToReport = 1
BBQ will get stuck since it's waiting indefinitey for a transaction
callback. The previous transaction can never get applied since we're
still waiting until drawsNeededToReport = 0.
This prevents situations like this from happening since it will not
allow a relayout to get called until the previous transaction has been
applied.
Test: Split screen with chrome
Bug: 167202096
Change-Id: I167a90ef00fa6677b3c55239e74eefb6a1384baf
When an app is running in portrait orientation, regardless of
what orientation display is in, IME shouldn't use fullscreen-mode.
Setting IME_FLAG_NO_FULLSCREEN in EditorInfo makes sure IME doesn't go
fullscreen.
Bug: 157870379
Test: Manually using steps in bug
Change-Id: I5a5e73e1dec776665f28a7e2eb091b555198001b
Add support of sensor manager in InputDevice and allow user to obtain
SensorManager object to query supported sensor types and register sensor
event listeners.
Bug: 161634265
Test: atest InputDeviceSensorManagerTest
Change-Id: Ia7a04a0fcf7e026c49a4fb1a9009b132ff84cf7c
After the pipeline changes to Pointer Capture, this CL cleans up the
following:
- The old way of notifying windows of Pointer Capture
(IWindow::dispatchPointerCaptureChanged) is removed.
- Tracking Pointer Capture in WindowManagerService is removed.
Bug: 141749603
Test: manual, flash crosshatch, pointer capture works.
Change-Id: I5bbcef42fe8d0d93a64a4a411b1d493fcb800f92
Slow prefetch requests would block user interactive requests, creating
noticeable sluggishness and unresponsiveness in accessibility services,
especially on the web.
Let's make it so a user interactive requests stops prefetching.
We can't interupt an API call, but we can stop in between API calls.
On the service side, we have to seprate the prefetch callbacks from the
find callback. And we have to make it asynchrnous. It does dispatch into
the main thread, so the AccessibilityCache can remain single threaded.
When the calls are interupted on the application side, returnPendingFindAccessibilityNodeInfosInPrefetch checks the find requests that are waiting in the queue, to see if they can be addressed by the prefetch results. If they can be, we don't have to call into potentially imperformance application code.
Bug:30969887
Test: Performance measurements, tried it out by hand to see if their are any bugs. CTSAccessibility*
Change-Id: Ia8f1152afa3987f262f37ed4583775acdd32db43
