- New core API classes in android.hardware.photography
- android.media.Image and android.media.ImageReader classes for
application access to direct hardware image buffers.
- Additions to android.graphics.ImageFormat to describe new image
types needed by new camera API.
- Some documentation included; very little implementation.
Bug: 9111736
Change-Id: I0680f35944d1cb8845b7dc0c67edc8c0f0864573
Cloning drawables (which happens a lot) was creating copies of NinePatch
objects, which would cause the hardware renderer to generate more meshes
than necessary. Also avoid keeping a String we don't need (it was interned
but still.) Last but not least, remove 1 RectF per NinePatch in the system.
Change-Id: If4dbfa0c30892c9b00d68875e334fd5c2bde8b94
The three arrays were being reconstructed and copied for each add
and each remove. By replacing them with ArrayList only three
constructions total are required. Also, the number of
System.arraycopy() calls is halved.
Change-Id: I0f8def1b517eb1bc5f930fcd5d3d1e0394071f0e
Conflicts:
core/java/android/view/WindowManagerGlobal.java
Small fixes in CrossFade to handle more general situations.
Also, added API to CrossFade to enable different behaviors during
the transition.
Change-Id: I12d17ab8f8f96be370c070373112ad8a62df9115
When relaunching activities the window manager clears out all windows
by calling a ViewRootImpl.die() in a deferred fashion. Then it
immediately deletes the ViewRootImpl and its view from its list
of windows. When the die() is eventually executed it calls
dispatchDetachedFromWindow() which tries to remove the previously
removed windows causing an Exception to be thrown.
This change waits to remove the windows until after die() has been
completed. Fixes bug 8253030.
Change-Id: I5b41be1c6b776e32128c064267653db98bd95292
When a View is added to a ViewGroupOverlay, it must not be parented
in another container. If it is, it will automatically be removed.
This works in general, but if there is a LayoutTransition on the
view's parent, then the actual removal will be delayed until the
transition completes. This means that the call to add(view) in
the ViewGroupOverlay's container will fail and throw an exception because
the view is still parented for the duration of the LayoutTransition.
This fix cancels the running transition to ensure that the view is not
parented prior to adding it to the overlay's container.
Issue #9303245 Crash combo: add(View) and animateLayoutChanges
Change-Id: Id4eaddccf64c353fc77e020882b978cdeb14c0ef
1. Before we were firing an accessibility event from the common
predecessor of views with accessibility related state changes
every X amount of time. These events designate that the tree
rooted at the source is invalid and should not be cached.
However, some of the state changes do not affect the view tree
structure and we can just refresh the node instead of evicting
and recaching nodes infos for views that did not change. Hence,
we need a way to distinguish between a subtree changed over a
node changed.
Adding a new event type will not work since if say two siblings
have local changes and their predecessor fires a window state
change event, the client will drop the subtree rooted at the
parent including the two views with changes. Subsequent, more
specialized events emitted from the two changed siblings will
be useless since the parent which did not changed is already
evicted from the cache. Conversely, if the specialized events
are fired from the two siblings with local changes and they
are refreshed in the cache the subsequent window state change
event from the common predecessor will force the refreshed
nodes to be evicted.
Hence, to enable distinction between node being changed and
a subtree baing changed while not changing existing behavior,
we will fire only window content change event with an additional
argument specifying what changed - node or a subtree for now.
Also if the changes are local to a view we fire the window
content changed event from the view. So, the two siblings will
fire such an event independently and the client will know that
these are local changes and can just refresh the node. If the
changes are structural, then we fire the window state change
event from the common predecessor.
2. Added the input type of a text view as one of the properties
reported by an AccessibilityNodeInfo. It is nice to prompt the
user what input is expected.
3. Added a bundle for optional information to AccessiiblityNodeInfo.
For example, it will be used for putting web specific properties
that do not map cleanly to Android specific ones in WebView.
4. AccessibilityInteractionController was not taking into account
whether the current accessibility focused node is shown before
returing it. Hence, a disconnected node would be returned and
caching it puts our cahche in an inconsistent state.
Change-Id: I8ed19cfb4a70bdd7597c3f105487f1651cffd9e0
The logic that canceled underlying animations caused a problem
when there were more than one animation in the list (due to removing
items from a list that was currently being processed elsewhere).
Change-Id: Ie207f340b5d7de4ffcf56a26c05ec394abb80224
The new ArrayMap class is more efficient for small collections.
Transitions use maps all over the place to collect/use property
values in setting up the transition animations. Changing to ArrayMap
should be more efficient, especially in terms of memory allocations
and GCs.
Issue #9276256 Transitions: Reduce memory allocations
Change-Id: I07b7d4ba68d5a207808eca11943aa3578fa90e3e
The behavior of running a transition is janky and unpredictable,
when there is already a transition running on the same scene root.
Usually, the new transition simply jumps to the end values, or
jumps to the start values for that transition and animates from
there.
A better approach is to cancel any running transition first, the
start the new transition from that point.
Even better would be to blend old/new transitions, or at least adjust
the animation timing according to where/when the previous transition
stopped. In the meantime, this fix is at least better than the
previous approach of ignoring running transitions.
Change-Id: I4f5fabb55f6454f1e9d66589a9a7c36f9fc013fb
setDuration() wasn't handled correctly for TransitionGroup; it should
propagate the value to its children.
Also, videos with no ids were not being handled correctly. The transition code was
using the default id on those views (-1) to store start/end data about the view,
causing multiple non-id views to clobber values in the hashmaps. The correct approach
should be to ignore default id values - only store information about the view
instances, not about the unset ids.
Also, added a new test InterruptTest to be used to fix the current behavior of
not handling situations where new transitions start while old ones are still taking place.
Change-Id: I4e880bdbb33cc26d487bceb0d56e463e72f7621f
Remove some abstraction-breaking magic in ActionBarView and replace it
with proper resolution of the icon/logo when creating a window. The
old implementation relied on the ActionBarView's context being an
Activity.
Bug 9171554
Change-Id: Idbbb1942622195dcb55e8119f2d64287b07bb509
- Add many details to most methods.
- Add comments specific to IME authors and to editor authors.
- Add missing return value docs.
- Straighten out single-spacing vs double spacing.
Bug: 8843866
Change-Id: If1f6dcf0457d5332a7ebb1ebfb1967c6ff0df722
Bug #9191438
When running under valgrind, the ppid will be different from the ppid
of the system server (which always gets forked from zygote.)
Change-Id: I42cbf99fd0084aeab76c30de9beb7c49ed1fc7d8
This is a new kind of key/value mapping that stores its data
as an array, so it doesn't need to create an extra Entry object
for every mapping placed in to it. It is also optimized to reduce
memory overhead in other ways, by keeping the base object small,
being fairly aggressive about keeping the array data structures
small, etc.
There are some unit and performance tests dropped in to some
random places; they will need to be put somewhere else once I
decided what we are going to do with this for the next release
(for example if we make it public the unit tests should go in
to CTS).
Switch IntentResolver to using ArrayMap instead of HashMap.
Also get rid of a bunch of duplicate implementations of binarySearch,
and add an optimization to the various sparse arrays where you can
supply an explicit 0 capacity to prevent it from doing an initial
array allocation; use this new optimization in a few places where it
makes sense.
Change-Id: I01ef2764680f8ae49938e2a2ed40dc01606a056b
The eglGetSystemTimeNV extension can be used to enable profiling
in PerfHUD ES. When the delta of two calls to eglGetSystemTimeNV
equals 0, we now cancels display lists updates. This allows the
tool to redraw the same frame several times in a row to run its
analysis.
For better results profiling should only be attempted after
setting viewroot.profile_rendering to true using adb shell
setprop.
Change-Id: I02e3c237418004cff8d6cb0b9a37126efae44c90
Bug #8667873
windowShouldResize means we need to layout the window, it doesn't mean
the dimensions of the surface have changed. We should only check the
width and the height. With this fix we can avoid a surface allocation
every time the window shade is opened or closed.
Change-Id: I8afe97b820a865723f2aab7a5aa4ddc8eaaec6e1
Bug #8833153
Display lists ops might now keep references to GL resources so they
must be destroyed when the EGL context goes away.
Change-Id: I0feb18f5539b345234a58dafa6f0775d7d7460dc
