1. Added a check whether the received hover event is in the bounds
if the view before sending accessiiblity hover events. Hence, if
a widnow gets a hover event and it gets to its topmost view the
latter will not fire an accessibility event.
bug:5233588
Change-Id: I61e81fbf0ce59a7f57377262ce066bcf8db98f56
Bug: 5156144
Input channels could leak or simply live longer than they should
in some cases.
1. Monitor channels (used by the pointer location overlay) are never
unregistered, so they would leak.
Added code to handle failures in the receive callback by closing
the input channel.
2. The DragState held onto its input window and application handles
even after the input channel was disposed.
Added code to null these handles out when they are no longer needed.
3. Input channels previously used as input event targets would stick
around until the targets were cleared (usually on the next
event).
Added code to detect when the input dispatcher is in
an idle state and to proactively clear the targets then
to ensure that resources are released promptly.
4. Native input window handles held onto the input channel even
after the input window was removed from the input dispatcher.
Consequently, the input channel would not be disposed until
the input window handle itself was freed. Since the input
window handle is held from managed code, this meant that the
window's input channel could stick around until the next GC.
Refactored the input window handle to separate the properties
(info) and identify (handle) state into different objects.
Then modified the dispatcher to release the properties (info)
when no longer needed, including the input channel.
7. The pointer location overlay does not actually use its
standard input channel, only the monitor input channel.
Added INPUT_FEATURE_NO_INPUT_CHANNEL to allow windows to
request that they not be provided with an input channel
at all.
Improved some of the error handling logic to emit the status
code as part of the exception message.
Change-Id: I01988d4391a70c6678c8b0e936ca051af680b1a5
Added Surface.setPosition(float, float) which allows to set a surface's
position in float.
Bug: 5239859
Change-Id: I903aef4ad5b5999142202fb8ea30fe216d805711
* changes:
MediaDump: use the MediaPlayer#setSurface method
HTML5VideoView: switch to MediaPlayer#setSurface
MediaPlayer: unhide the setSurface method
Surface: unhide the SurfaceTexture ctor
There was a bug in an InputMethod app, where popups for the keys
would not pop-down again. The problem was that they were being
marked INVISIBLE, but the new invalidation logic noop'd the
invalidate() call that used to take place. Adding to that was logic
in setFlags() that only invalidated a parent for parents that are
instanceof ViewGroup. In this case, the parent is a ViewRootImpl.
Fix is to call invalidateChild() on the parent if it's not a ViewGroup.
Change-Id: I2c2352072d383cee1367ea7ee6c2207077721fd5
This change unhides the Surface constructor that creates a Surface that
pushes frames to a SurfaceTexture.
Change-Id: I2e587afc52c9838d25b8681cae7421734b3e7b6e
This change moves the Surface ctor that takes a SurfaceTexture to the
portion of the file containing the public APIs.
Change-Id: I192df7bf5c0e6d0d0607bd577d72646312b2f717
A recent fix for invalidation noop'd calls to invalidate() on
GONE/INVISIBLE views. This logic also noop'd views which might
be GONE, but which are in the process of fading in/out via
LayoutTransition animations. These views should invalidate as
usual.
Change-Id: Ie90a340f70290391a3aa4e68df535c6aabf4e5eb
An earlier fix nooop'd invalidate calls on non-visible views. This
caused an issue where changes to the view while it was not visible
(such as changing the text of a TextView) would not get picked up
by the invalidation process.
The fix is to automatically invalidate() a view when it becomes
visible, to account for any changes that may have occurred to its
content while it was not visible.
Change-Id: Ia0f8fb42ca627b5a38f08761fa622b23f2dfa38b
It looks like this is caused by the change in HC to stop activities when
the screen is off. ViewRootImpl (a.k.a. ViewRoot) has special code to
avoid doing work when it is stopped, and it is now stopped when the screen
is off. The problem here is if the window's activity is stopped when the
window is first displayed, then it would never do the initial fitSystemWindows()
with the status bar offsets given by the window manager, and never do this
again until the status bar changes.
Also included here is some re-arranging of the code dealing with the offsets
changing, because it was dealt with in two places and only one had a bunch of
code dealing with HW accelerated drawing and performing the fade transition
between states. Now all of that is unified into one place.
Change-Id: I9828f02664cc622dbf186effb1f685a8aa4456a1
The content retrieval APIs are synchronous from a client's
perspective but internally they are asynchronous. The client thread
calls into the system requesting an action and providing a callback
to receive the result after which it waits up to a timeout for that
result. The system enforces security and then delegates the request
to a given view hierarchy where a message is posted (from a binder
thread) describing what to be performed by the main UI thread the
result of which it delivered via the mentioned callback. However,
the blocked client thread and the main UI thread of the target view
hierarchy can be the same one, for example an accessibility service
and an activity run in the same process, thus they are executed on the
same main thread. In such a case the retrieval will fail since the UI
thread that has to process the message describing the work to be done
is blocked waiting for a result is has to compute! To avoid this scenario
when making a call the client also passes its process and thread ids so
the accessed view hierarchy can detect if the client making the request
is running in its main UI thread. In such a case the view hierarchy,
specifically the binder thread performing the IPC to it, does not post a
message to be run on the UI thread but passes it to the singleton
interaction client through which all interactions occur and the latter is
responsible to execute the message before starting to wait for the
asynchronous result delivered via the callback. In this case the expected
result is already received so no waiting is performed.
bug:5138933
Change-Id: I382e2d8689f5189110226613c2387f553df98bd3
