When generating delegates, LayoutLib.Create support renaming
inner classes. Only one level of inner class is supported.
The method Outer$Inner#get(...) generates a call to:
static Outer_Inner_Delegate#get(Outer instance, Outer$Inner instance, ...)
Change-Id: Ie70f2b8e5e5f311ed9c7f26b7f64637ae6157a51
The "resizeable" attribute of supports-screens was never well documented,
so many apps don't set it. Assuming that if they are explicitly saying
they support large or xlarge screens then they are also implying that they
are resizeable.
Change-Id: Iaa1ad431c9868254af7581499477bff98ed109e5
Bug 3295544
Only the last line of text includes the bottomPadding (extra line
spacing below the characters' descent. When The text is clipped using
maxLines, the desired height correctly added this value, but getLineTop
and getLineDescent are also used when the layout is drawn.
The fix is to make the layout aware of its clipping so that these
values are correctly updated.
Change-Id: I703656cf45022d34a90f55f0ed8fc5e4b30f80b1
...an already-closed object: android.database.sqlite.SQLiteQuery
It turns out there is a state we are missing -- the loader is
still needed, but in the inactive list. In this case the loader
needs to continue holding on to its current data, and not deliver
any new data (which would result in it releasing its old data).
This introduces the new state to Loader, and uses it in
AsyncTaskLoader so all subclasses of that should get the new
correct behavior.
A further improvement would be to unregister CursorLoader's
content listener when going in to this state, but that can
wait for later.
Change-Id: I6d30173b94f8e30b5be31d018accd328cc3388ec
Bug: 3393571
Limit the event queue to a single pending SET_SCROLL_OFFSET. The
problem on slashdot.org was that the event thread was being flooded
with SET_SCROLL_OFFSET messages faster than it could process them,
causing the handler to get jammed up.
Change-Id: Ia4a90a030959d0b3769067791089cdb261ae403a
We need some additional debugging to help diagnose the cause
of this bug. There is no logic changes just additional debugging.
Change-Id: Ifc5d1c82904042fa87fd3b5b1066f58b62e7359c
The bug caused libagl to return 0 from eglGetError if an EGL error value
(including EGL_SUCCESS) was set on a different thread but not yet on the
current thread.
Bug: 3403756
Change-Id: Ifd965091d116745c2e22c121151ade9e78eb14c6
If the app doesn't have that permission, just mask out the bits.
Also makes the systemUiVisibility field visible in hierarchyviewer.
Change-Id: Ic3d3a3aed96b5c6448832755b4b2e6175117e5be
