1. AccessibilityNodeInfos for visible views should always
be reported and the clients should be able to check
whether that node info is shown to the user. For example,
focus search may return a node that is not on the screen
and the accessibility layer may decide to give it focus
which scroll the source view in the screen.
bug:6421991
Change-Id: Idc1fd8512dda767abe802aacedb0c69582e6fc2a
1. Since we are using a stateless proxy accessibility service to
perform default accessibility gesture handling it shuld not
operate against not important views.
bug:6422069
Change-Id: I74df9c2415ab3b164d9ac5873f7004c0459e2bfa
Propagate the return value from onNavigateUp as the result of
onMenuItemSelected. This fixes a bug where the action bar Up nav
button clicks would not be propagated to support lib FragmentActivity
or other activity subclasses that rely on processing otherwise
unhandled onMenuItemSelected events.
Change-Id: Id879dd1756ed06b8a7b720ebf0eae2a8ddc66ca8
1. Implementing text content navigation at various granularities.
For views that have content description but no text the
content description is the traversed at character and word
granularities. For views that inherit from TextView the
supported granularities are character, word, line, and page.
bug:5932640
Conflicts:
core/java/android/view/View.java
Conflicts:
core/java/android/view/View.java
Change-Id: I66d1e16ce9ac5d6b49f036b17c087b2a7075e4c0
Bug 5731093
Replace TextView by EditText to make it cleaner.
Had to force the text color, although the TextAppearance should
handle it. Maybe something is wrong with the theme.
Also added a hint for consistency.
Change-Id: Icd7170a0ef3757d0313e0799873f2dd375b02f2f
Detect when a vsync message was significantly delayed which may
indicate that a frame was skipped. When this happens, update
the frame time to reflect the approximate start time of the
current frame instead of the start time of the frame that was
skipped a long time ago.
Removed an unnecessary call to getCurrentPlayTime() in the
animator framework. The result was always zero and the call
just made the code confusing.
Bug: 6443611
Change-Id: I92b24f7ffd74c59b75a727b6bfc0bb51fc92a73a
Handle the edge cases a bit better. Deal with padding. If a smooth
scroll begins, the list is empty, but a data change is pending, try
again on the next looper pass.
Bug 6453837
Bug 6434713
Change-Id: I53f22ebacdcbc5d981a6c8055c4fd3fc1ef140f6
This change replaces CopyOnWriteArrayList with a custom CopyOnWriteArray.
The new CopyOnWriteArray only allocates when a concurrent modification
is detected. Only one loop can iterate over CopyOnWriteArray at any
given time and the class is not thread safe.
Change-Id: Ie67a1ec20ff095350bf3c5d4f87cdb231ad57221
1. The model is using a package monitor to observe changes in
packages (adding/memoving/etc) to update its internal state
for which it then notifies its clients. However, the monitor
is called from a binder thread and the work has to be off
handed on the thread that created the model. Hence, clients
do not need to worry about using the the model from the UI
thread since the change callback may touch the UI from
another thread.
bug:6386152
Change-Id: I882a0f4104907f64de64a95ece585052bba457d6
Also don't retain the source bounds in recent tasks, since it
has no meaning there and it would be better when relaunching an
activity to have a new bounds set based on wherever it is now
being launched from.
Change-Id: Ia90c04ee98a888a7f725b038abe23d71e2b12800
...without onPause() in between
There was a bug in the handling of "always finish activities" where we
would go through destroying activities while in the middle of updating
the activity stack. This would result in the activity behind the
non-full-screen activity being created and then immediately destroyed,
which things were not expecting.
Change-Id: Idaa89089f7b1af7eb747d7b8f9f394beeb2d23fa
This fixes a bug where the incoming call screen sometimes had
misplaced icons. The code previously relied on an additional layout
pass that doesn't happen in the incoming call screen in some cases.
Change-Id: If7c39994c1bdadbd3e97f4b9ebb45f68893fd9bd
We don't get link status info if it's down. If the device has been configured
for ethernet and we see the interface appear, we should make it active.
bug:6444395
Change-Id: Ibc233ab12942b069d7db9c0671936798c5161659
This is because we do HotOff -> PowerOff -> HotOff for USER_TURN_OFF
Donot do poweroff for perprocess turn off.
Add small time delays for calls into stack to reset dbus timing.
bug 6438556
Change-Id: Iae8bdd24e41824ed4bc34aca96d082f9284f169c
