This reduces the jank in animation by reducing the measure calls, as the
animating views do not change size. This means only using `wrap_content`
or fixed non-zero sizes.
Test: jank suite
Test: atest CombinedShadeHeaderConstraintsTest
Test: manual with different cutouts
Test: perfetto trace
Fixes: 258083936
Change-Id: Ia53411dd02f3fb0019d67a8ee3bc5111d13066f9
Unit tests for radio module and get/set methods in tuner session
of broadcast radio HIDL HAL client are added. Meanwhile, replace
deprecated Matchers package with ArgumentMatchers package.
Bug: 258034192
Test: atest TunerSessionHidlTest RadioTunerHidlTest
Change-Id: I5530c5b42cf3f4763df8d003871196d630ef06cd
TL;DR: Skia should always call releaseProc, and maybe sooner than we thought.
There are multiple scenarios where SkImage:MakeFromTexture will fail,
returning a nullptr and calling releaseProc due to a RefCntedCallback
falling out of scope. Previously this could cause mUsageCount to fall to
0, resulting in the AutoBackendTextureRelease deleting itself even
though DeferredLayerUpdater owned a ref and expected it to still exist.
Also added logging for some reasons that could cause the later call to
MakeFromTexture to fail.
Bug: b/246831853
Test: hwui_unit_tests
Change-Id: I7fd2566b9a85fe286f72b0fc42eba5450cac69b0
Merged-In: I7fd2566b9a85fe286f72b0fc42eba5450cac69b0
if needed.
Bug: 255625888
Test: manual: Verify battery content description in all states
Test: atest BatteryMeterViewTest
Change-Id: I497c14096da2693cc35bf01d76287c021ddc3bf8
necessary.
When writing some tests for the new dock defend icon, I came across
some existing issues when we don't correctly update the text or content
description. This CL adds tests that would've been failing, and updates
BatteryMeterView to make them pass.
Bug: 255625888
Test: atest BatteryMeterViewTest
Change-Id: I3485a50c6471dd75dce64a3daf414208afa8d25e
with the bottom of the other icons, even if the shield is displayed.
Bug: 255625888
Test: manual: Verify battery icon bottom stays aligned with other icon
bottoms both with and without the shield. See video attached to bug.
Change-Id: Ib203e8ce27d515a177e8371958fa43ecf8b20e74
Change-Id: I1d8c53cc2beae90f6dca14cf339733bfb085a56e
The shield will be displayed when the battery is marked as overheated.
Bug: 255625888
Test: manual: issue `adb shell am broadcast -a com.android.systemui.demo
-e command battery -e overheated true` and verify the battery icon
displays a shield
Test: See bug for manual demo video
Test: statusbar.battery tests
Change-Id: I6b3036ae47ad7b5ae8dc78f83d019130935c9b50
For foldables, the power button can appear on three different sides
when rotating. Make sure power button reveal scrim comes from wherever
it is.
Apologies - the whole file was formatted but the only part that
changed was in PowerButtonReveal
Fixes: 204737218
Test: manual, use device power button and keep rotating
Change-Id: I1dbd1fd76456ef2c1a7de2847877e3c029176eba
This enables accessibility services to take a screenshot of
a window without any overlay windows drawn on top, which is
beneficial for ML-based understanding of the visual content
in the window.
Bug: 249830826
Test: atest AccessibilityTakeScreenshotTest
Test: use in an a11y service, observe screenshots
contain only content from the requested windows
and no overlays
Change-Id: Ic886aeace9cb2984a472273858281b4f2c3a0e46
* changes:
[automerged blank] Fix testUsesSdk test to use higher extension version. 2p: 587492489c
Fix testUsesSdk test to use higher extension version.
The callback that home controls observes is only triggered when the
controls services update, which is one of the conditions for whether
home controls is available. When the user adds the first or removes the
last device control, the hasFavorites condition changes but does not
trigger an update.
So to address that, home controls should check availability when the
dream overlay becomes active, parallel to which on the lock screen.
Test: DreamHomeControlsComplicationTest
Test: reboot device with no favorited device controls > add favorite
device controls > start dreaming > see home controls chip
Test: similar to above but start with favorites and remove last
Bug: 257348199
Fix: 257348199
Change-Id: I537a240963e0523c8fc40008dee961c7a434094a
These numbers are useful in adjusting JS policy, so we should persist
them so we can use them after the device reboots.
Bug: 258701271
Test: atest FrameworksServicesTests:JobStoreTest
Change-Id: Iee8fb699ea5b4c01446ce8fd8f806203effbe4ce
Due to the extension of the definition of the ApexInfoGathered atom,
we are changing the params to `FrameworkStatsLog.write()` to match
the atom definition.
Bug: 245944675
Test: Manual.
Change-Id: If5a2df21ecb13d0bc34b70116cc133ef823a642a
This CL introduces
onInputMethodSubtypeChangedForKeyboardLayoutMapping()
to
InputManagerInternal
in a way that is similar to our previous attempt [1].
The two major differences are
A. It also provides InputMethodSubtypeHandle so that
InputManagerService can use it as look up keys, and
B. Subtypes that return false from
isSuitableForPhysicalKeyboardLayoutMapping() [2]
are mapped to null.
Other than those two differences, this CL is the same as our previous
CL [1].
See also another previous CL [3] about how this callback can be used.
[1]: I58e71ffce9ac9131551a00dd35e24235dadfef87
b097b8262b
[2]: Ifc0247041a43ef64f8a76a23832da2ee058c6958
6980a434cd
[3]: Ie88ce1ab77dbfe03ab51d89c1dc9e0a7ddbb3216
d5f7ed9fe9
Bug: 252816846
Test: manually verified
Change-Id: I087f02d06dc5575d2e657c25614d0361ef2ff443
