I want to include some stuff (like android.platform.test.rule.CrashDetector) but I want to land just the dependency part and see if I get postsubmit build problems that sometimes appear in obscure AOSP targets.
Test: presubmit
Bug: 187761685
Change-Id: I334b9f1029edbe8621df940900f43d0c5059373e
Merged-In: I334b9f1029edbe8621df940900f43d0c5059373e
While re-arranging icons the hotseat remains in scale 1.0f, while the workspace reduces it's scale (as defined by SpringLoadedState.java). Previously, the code to aggregate animations was assuming hotseat and workspace always had the same scale.
MultiScaleProperty.get() was being used to set the starting value of the animation. Previously, it was returning the last aggregated value. However, this value was correct only for the workspace, but not for the hotseat. Returning the current view scale makes it always correct.
Bug: 220271046
Test: Dragged icons from hotseat to workspace, and verified animation didn't jump
Change-Id: Ic01776c1d8e3967624626ed7c44d194a06295790
It seems if the dragging target position's X coordinate is 0
then there is a weird bug in portrait mode.
Test: atest TwoPanelWorkspaceTest
Bug: 197631877
Change-Id: I7d8702b06e5ed396814339080f8e814e9f8a6a65
The unfold progresses are mapped to 0.85 - 1 range and set as a scale for launcher.
In case of multiple scale animations for workspace and hotseat, they are combined using MultiScaleProperty (e.g. opening an app while unfolding/going to all apps while unfolding). Note that this is a pretty difficult scenario to be in. If that happens, we multiply all values and bound the result between the max and min values.
Bug: 217368525
Test: atest MultiScalePropertyTest and manually
Change-Id: I6131c39f36deade0b7280c72edda2d72045344e9
TAPL was expecting pilfer upon UP in the swipe-pause-to-overview gesture.
Actual Pilfer happens earlier, which resulted in wrong sequence-mismatch diags if the gesture fails before UP.
Now expecting Pilfer earlier.
Also fixing a problem that sneaked through earlier reviews when even down time was passed incorrectly.
Bug: 187761685
Test: presubmit
Change-Id: If4289a011247bf7801405ecaac8befa8461e8a37
Needed because when running tests in virtual devices is very slow. The previous number of steps was also greater than the default timeout for home gesture.
Fixes 215673732
Test: atest Launcher3Tests:RequestPinItemTest
Change-Id: I9b13d97deb158222ba0ebf3f6a536c5a7ae632cf
This reverts commit 1a9ef83988.
Reason for revert: Remove ScreenRecord to see if it resolves stale input event problem
Bug: 215672979
Change-Id: Ia7b7cd992c79af34b11d17becde98f253ab778ac
The current AllAppsContainerView has been split into a base class and a
subclass for activities. The base class allows using all apps with an
activity context. A-Z and work profile tabs are supported, but search
and the hotseat still require an Activity.
Test: Manual. All apps should work the same.
Bug: 204696617
Change-Id: I3b146714bc11b3b3555d97623aab5d78ac836482
When running on tablets in landscape, the QSB should inline with the hotseat icons. This does that and fixes small bugs that didn't account for the new hotseatBorderSpace.
Bug: 210118169
Test: atest Launcher3Tests:DeviceProfileTest
Test: visual, using HSV and Window
Change-Id: I5a89c5449bf59fde736618151eceaacca443ef67
- OOP test activity has an alertTitle and shouldn't be treated as system anomaly dialog
Fix: 214558538
Test: NexusLauncherOutOfProcTests / MemoryTests
Change-Id: Ia9c330c3c3f48b43fbada40a3410ccc9fb8bfbb8
- Somehow when home gesture's travel distance is above certain number (~2560px in the bug's case), the touch will wrongly go to the 3P launcher instead of staying in quickstep launcehr only. As it's unusal for home gesture to be so long, limiting it to half of screen height.
Bug: 214279686
Test: presbumit and TaplTestsLauncher3#testWidgets on cf_tablet
Change-Id: I1107074556aea3011b186ba34dc830dbb1445517
1. Implement function to uninstall app from appIcon.
2. Add test uninstall to TaplTestsLauncher3.
Bug: 210927656
Test: Launcher3Tests:com.android.launcher3.ui.TaplTestsLauncher3#testTryUninstallFromWorkspace
Change-Id: Iedd3bd3a46bc626fc414fd8c5bd07ebc0fa235bb
- TODO: enable test in portrait
Test: TwoPanelWorkspaceTest
Former CL: http://ag/16357467
Bug: 197631877
Change-Id: I35c326b7a6c612d2fb66c8b94b810f5131fd58d8
The display's x-coordinates for input are in the range:
[0, displayWidth).
Injecting input at the point (displayWidth, 0) is technically outside
of the display. After accounting for insets, the x-coordinate range for
the display is:
[leftInset, displayWidth - rightInset).
An input event injected at x = (displayWidth - rightInset) would be
outside of this range, and would be within the right insets region.
To account for this, use a one pixel buffer when calculating the
right gesture start margin.
A similar one pixel buffer is already used for the bottom margin
calculation.
Test: atest NexusLauncherTests:
com.android.quickstep.TaplTestsQuickstep
Bug: 208849306
Change-Id: I9d5a6429cd55fdfcbf2c849e27e3d3a0a7faaa47
Test is passing correctly but for the wrong reasons. If we attempted to use a different distance it would have failed.
Bug: 197630182
Test: manual
Change-Id: Ib275b89c86cdb9f75d40b9d25c8c498098b543b7
Memory dump can be generated only once, so generating it in the intentional leak test leaves us without a meaningful dump if a real leak happens later, and produces a confusing message that the dump was already generated.
Bug: 195319692
Test: presubmit
Change-Id: I35932deba0c46ddb27362467109daf74d75cbff4
Vadim Tryshev
Thales Lima
Nicolo' Mazzucato
András Klöczl
Schneider Victor-tulias
TreeHugger Robot
Benno Lin
Alex Chau
Bob Badour
Pat Manning
kholoud mohamed
Brian Isganitis
Ling Wo
Prabir Pradhan