The setup waited for a ROTATION_0, but this test starts at ROTATION_90
Bug: 219773854
Bug: 220977327
Test: atest WMShellFlickerTests:SetRequestedOrientationWhilePinnedTest
Change-Id: Ib2bde1f66f248adbe69acf5c344322e32b364eb2
Changed the resource type extra to be an int value to indicate the
resource type instead of having individual boolean extras for each
resource type.
Bug: 218875965
Test: N/A
Change-Id: Ic734c4d0e0a769dd981ef214e1be41f62d74705c
When the windowing mode is changed during fixed rotation, we don't want
to animate to the fixed rotation anymore, because the fixed rotation
will be cancelled later.
Bug: 214324061
Test: atest WMShellFlickerTests:SetRequestedOrientationWhilePinnedTest
Change-Id: I24b5389ec2b6ea28ce3ba7abf38fbd269566edf3
All the values / things being done in PIP are exactly the same as what
DismissView does. This change updates PIP to use DismissView directly
so PIP will always pick up visual changes to dismiss without having
to copy them over manually.
Test: manual - have a PIP, drag it around and observe dismiss, drag
PIP to dismiss, observe
Bug: 217211205
Change-Id: I430df28bdb7980ff5a0316c4aab728fe54e95e6b
1 - larger gradient (same for both phone & tablet)
2 - gradient is now based on theme color
Test: manual - have a bubble, drag to dismiss it, observe the gradient
Bug: 217211205
Change-Id: Ibef23992089f0ae33dd93f9528a3ea2887aae465
* Previously on large screens the expanded view was centered on the
screen now it will be anchored to whichever edge it was expanded on
ONLY when in landscape on a large large screen. In portrait, and on
"small tablets" it will continue to be centered.
* Updates the width of the expanded view on portrait and on landscape
for large screens.
* Adds the notion of small tablet
Test: manual - Open a bubble in various configurations:
1) on phone:
* landscape => bubbles show vertically on the side they were
expanded on and expanded view is centered horizontally
* portrait => bubbles show on top, expanded view centered
horizontally
2) on tablet that isn't small:
* landscape => bubbles show vertically on the side they were
expanded on and the expanded view is anchored to that edge
with some padding
* portrait => bubbles show vertically on the side they were
expanded on and expanded view is centered horizontally
3) on small tablet:
* portrait & landscape => bubbles show vertically on the side
they were expanded on and the expanded view is centered
horizontally and takes up more space %-wise compared to a
large tablet
Bug: 217211205
Change-Id: I7259e03aab35e0039c9b7bd855c12afe38373ace
This makes triaging issues unrelated to the test (e.g., transition crash) easier since all failing assertions are clustered with the same stack trace
Bug: 219773854
Test: atest FlickerTests WMShellFlickerTests
Change-Id: I65bb1b5e87997c856ce080b24d549ba1c978955a
Merged-In: I65bb1b5e87997c856ce080b24d549ba1c978955a
Use parent based coordinate instead of screen based coordinate when
indicating split surfaces' position so they won't be offset unexpectedly
when its parent is not at (0, 0).
Fix: 210696145
Test: check split UI when enabled "render apps below cutout area"
Change-Id: I65399f5ef2975e7ef376d161c41de2cfb7ed29dc
The transition is to click on the pip menu, then click on the dismiss
button. This will cause the focus change.
Fix: 215869110
Fix: 216306753
Test: atest WMShellFlickerTests:ExitPipWithDismissButtonTest
Change-Id: I50dab84f588c3bedaa7d65c02144054d4eefdcd5
- Add simple fade of the drag surface when dropping and when canceling a
drag from the shade (since the shade has closed, the default animation
of animating to the start position is wrong)
- Add case for dragging over home/recents, show the highlighted drop zone
as fullscreen since there is nothing to split with
- Fix issue with drag listener not being removed after a notification
drag
- For now, add a toast when attempting to drag a notification that does
not have a valid intent
- Also don't wait for DRAG_START to start collapsing the shade if the
call to startDrag() returns true
Bug: 217972792
Bug: 218791595
Bug: 219757067
Test: atest WMShellUnitTests
Test: atest SystemUITests
Change-Id: Ic53de0d8a999d4ae39dc207bfc2927466ad2ba7c
Test: atest BubblesTest NewNotifPipelineBubblesTest
Test: manual - force notification permission dialog to always show
- open a bubble
=> it doesn't immediately collapse
- swipe up on gesture nav
=> the bubble collapses
Bug: 218517365
Change-Id: I4271b955cac51b85e0d92e738fa2d0e38afeb154
When asking PipTaskOrganizer to animate the PiP to a different position
while it's still in the appearing transition, the PiP could get stuck in
a bad state where the PiP contents and menu items weren't drawn.
This change avoids this by only calling #scheduleAnimateResizePip once
the PiP has finished appearing.
Test: manual
Bug: 220110011
Change-Id: Ie62adea6f2cb5a4abdd6c4b8f9ff23c040077046
In addition, this CL does additional polishing to the code itself.
Bug: 215316431
Test: atest WMShellUnitTests:LetterboxEduDialogLayoutTest
Test: atest WMShellUnitTests:LetterboxEduWindowManagerTest
Test: atest WMShellUnitTests:CompatUIControllerTest
Change-Id: I3cb58f89a52f0d427545e3959dd9082e8b8b5cc7
- Previously, mWMComponent.init() would make a blocking call to WMShell
to initialize on the shell main thread. We're expanding that to
include the construction of the shell objects as well to provide
a little more consistency and predictability on the shell side
(ie. fetching some statics may seem safe, but actually aren't).
We're continuing to recommending that ctors don't include any
initialization logic though
Bug: 181979527
Test: Verify with trace that shell init happens on the
shell main thread
Test: atest SystemUITests
Test: atest WMShellUnitTests
Change-Id: I60559a47e00e91d34be7d4e94f4f3e041fdc590d
Signed-off-by: Winson Chung <winsonc@google.com>
Entering PIP rotation with orientation is changed to use fixed rotation,
so we need to update the animation as well.
Will update exiting PIP and swipe to enter in next cl
Bug: 219411429
Test: manual test enter PIP from different orientation
Change-Id: Id30f15c72b7d59ae63012d9f62e829d73ec4f335
I made it so that the overflow expanded view could return null and
didn't protect this access of it. Fix is to check if the view is
null before doing something to it.
Test: treehugger
Bug: 219491953
Change-Id: I380584223ad56c754cf59c63e3eb2131c3275154
Terms
- Host activity, the activity that the content-pip is originated from.
There should be some UI affordance on Host activity user can tap to
enter content-pip
- Container activity, the activity that carries the content from Host
Activity and transits to PiP mode
Workflow in brief
- Container activity is defined in the same application as Host activity
- A new ActivityOptions#makeLaunchIntoPip API is added, Host activity
uses this API to construct ActivityOptions to launch the Container
activity directly into pip mode.
- ActivityStarter sets a reference on Container activity pointing to
Host activity upon creation and moves the activity to pinned task
- WMShell transits Container activity to pinned mode
- When user taps on `Expand` button in PiP menu, WMShell calls
WCT#startTask with the task id to bring the Host task to front
Content hand-off
- Since Host activity is the one starts Container activity, it can
replace its content with some placeholder right after startActivity
- This CL also makes the assumption that Host activity can establish a
communication channel with the Container activity, such as using the
ResultReceiver parcelable. Therefore it's out of scope to provide
extra callback to Host activity on lifecycle events of Container activity.
See also the design doc go/content-pip-v2 and the end-to-end demo in
ApiDemos `App > Activity > Picture in Picture`
Bug: 165793661
Video: http://recall/-/aaaaaabFQoRHlzixHdtY/e9Mp5rvxYBrNOdGSUoSXHO
Test: manual, see Video
Test: atest WmTests:ActivityOptionsTest \
WmTests:ActivityRecordTests \
WmTests:ActivityStarterTests \
PinnedStackTests
Change-Id: I2bebfea8d1930b0b7e495d91cfd5ddc7c347da68
Using compressed trace files speeds up trace collection and the benefits of using async trace fetch become negligible.
To simplify the code, remove the functionality
Bug: 209843622
Test: atest FlickerTests WMShellFlickerTests
Change-Id: Id21d58af5393eb7f364b353db619c55a5a5b616c
