This allows CookieManager subclasses to correctly override all behvaior
in the base class.
Bug: 6379925
Change-Id: Ife02db24bc192bc1eabf09e0d64668716786d237
When used in a `ViewPager`, fragments that are present on the adjacent,
cached pages will receive context selection dispatches which, depending
on your fragment contents, can be difficult to determine whether or not
the event truly originated from your view.
By using the visible hint we restrict dispatching to only those fragments
which are marked as being visible. Since the fragment pager adapter
updates this setting properly most implementations will be afforded this
fix without any change required. If the user is implementing their own
adapter they likely already understand the implications of these cached
fragments and the reponsibility for updating the boolean falls to them.
Mirrors support library change Ie6a72c1c82c2784774373670007b6f5948fe16da
Change-Id: I1bc6f74e26660ef4a3c20419c077e37992412e94
1. The accessibility focus directions are not needed since an
accessibility service just get the root, first child, next
sibling, previous sibling and call execute the action to
give it accessibility focus. Now the accessibility node
info tree is properly ordered taking into account layout
manager directions for both layout manager that we report
and ones that we have determined as not important for
accessibility. Also the position of a node info are ordered
properly based on their coordinates after all transformations
as opposed to child index.
bug:5932640
Change-Id: I994a8297cb1e57c829ecbac73a937c2bcbe0bac7
Bug #6365056
WebView enqueues a functor in the hardware renderer to handle
animations and this functor is called at a later time by the
hardware renderer. However, the functor was not removed from
the queue when WebView was removed from the window. This could
cause the hardware renderer to attempt to execute an invalid
functor and lead to a crash.
Change-Id: I9d38e80f3fdc5e29d4d0cdfa1e893c251a954508
If a fragment's saved view state is null and the user
visible hint is true then the `result` bundle will have
never been initialized to a value resulting in a
`NullPointerException`.
Mirrors support library change I8ba585bc6b9298841490d64bc22a8219cd261adb.
Change-Id: Iabd5ac293d2ece3771da9ef257479eca0dcd523c
Will also fix Bug 6344997
The previous TextWatcher mechanism was inneficient. It require an
expensive getSpans() call to retrieve all the spans and then search
for the one we're interested in in case it has been changed.
The SpanWatcher is faster, it will broadcast the add/changed/removed
events we're interested in.
Now that we can rely on SpanWatcher, use it to directly track
addition and removals of EasyEditSpans.
No unit test for this feature which require an integration with
the voice IME. Easy to test manually though.
Change-Id: Idabcacc48c479bf9868d5204c0b0ca709207ede2
This is a feature used for NFC-to-Bluetooth handover:
we want to enable BT for file transfer, and disconnect
it when we're done. During this period we don't want
to auto-connect other devices - it should be transparent
to the user that Bluetooth is used. Also, don't allow
A2DP/HSP incoming connections.
Change-Id: I0a03e8084c439b1271b6a80f4d9da5aacfe19c45
The C++ API for scheduling groups is going away, so use the C API.
Renumber THREAD_GROUP_* constants to have the same values as SP_*.
They are @hide so this should be OK.
The old THREAD_GROUP_FG_BOOST constant is removed, and replaced by a
(private) THREAD_GROUP_FOREGROUND constant that is just a placeholder
to correspond to SP_FOREGROUND. It is not usable at this level.
Improved performance of android_os_Process_setProcessGroup:
avoid unnecessary syscall to getpriority for each tid.
Fixed error handling in android_os_Process_setThreadGroup.
It raises an exception if set_sched_policy() returns an error.
Add comments and fix typos in existing comments.
Change-Id: Ib922e92c945985c695ef0addbd2b83b8e82865f8
Let TaskStackBuilder discover a parent activity stack by ComponentName
in addition to explicit Activity classes.
Change-Id: I18b8378548ed1d6ef033800e6a3e11ab965d07e5
This flag was still hanging around pending any need to disable
DisplayList properties. But things seem stable, so it's time to clean up
and simplify the code.
At the same time, I reduced redundance in DisplayList dimensions. We
used to call drawDisplayList() with width/height parameters that were
used to do a clip reject. This is redundant with the DisplayList properties
that set the bounds of the DisplayList; the left/right and top/bottom properties
represent the same width/height properties formerly used in drawDisplayList().
The new approach is to not pass dimensions to drawDisplayList(), but to
instead pull those dimensions directly from the DisplayList when needed.
Change-Id: I8871beff03b1d4be95f7c6e079c31a71d31e0c56
This is a quick workaround to get tests running again, will follow up with full fix tomorrow.
Bug: 6379925
Change-Id: I96d6e27bfb8f8cd41ec08845ab0fb1e584dbc9da
There are times when we deliberately put the WebCore thread
to sleep (i.e. waiting for a user to complete a JS prompt or
dismiss an alert). In this case, we don't want the WebCore thread
watchdog to fire. Pause the watchdog before we put the WebCore thread
to sleep and resume it again when the thread wakes up.
Factor out the repeated send/pause/wait/resume code into it's own
function.
Bug: 6366520
Change-Id: I17ecf9d466ce21b25a9e5cb8ff4cb0e5fab7605b
GeolocationPermissionsClassic
CookieManagerClassic
WebIconDatabaseClassic
WebStorageClassic
Also creats a WebViewFactory top level class - this remains hidden
for now, as it's currently only used implicitly by the other
public WebView classes to create the provider instances.
Bug: 5626244
Change-Id: Id0ca1c16d8058f31a86414bbc0e8a55db4b907ba
