Bug 3400770
TextUtils#commaEllipsize creates a MeasuredText "mt" object with the
text to be ellipsized. It calls setPara which initializes mt's mPos
member to be 0. It then calls addStyleRun which moves mPos to the end
of the string. The loop back in commaEllipsize then calls mt addStyleRun
again and this causes IndexOutOfBoundsException, because the paint
object is trying to measure text past the text's length.
It seems this was a typo and that the tempMt variable should be used,
because the code is trying to measure the format string...not the
the string to be ellipsized. This makes the saner parts of CTS test
for this method pass now.
Change-Id: Ib6aa6e4bbd6afff4c95ad4c4d51a384cc1389875
Or at least make it better. Now if we get a failure locking the surface,
we mark to do a full relayout pass later to try to get a new good surface.
Also fix some bugs in how activity manager was classifying processes for
their OOM adjustment to make better choices in what to kill.
Change-Id: I8e4aa86744211ba7693f9828291d8bbf2698274f
Also removed android:detachWallpaper="false" from lock_screen_exit
since it's not guaranteed to do something sane when windows aren't
owned by applications.
Change-Id: I28b5fc6b68d1aef93f092538d1318ce2b2a835ef
IOException on close() can be useful to indicate that in-progress transactions
were canceled.
I also audited all of our tech classes to make sure every function that needs
to throw IOException does so.
Change-Id: Iaa9c43d79d59ff85772d5c3e4b4d57a6fa8df4cf
Bug 3370191
The documented behavior is to hide the error when the text changes.
However, this should not be the case if the error was reset by a text watcher.
Comparing errorBefore and errorAfter as was done before is not sufficient in the
case where the error is reset to the same value. String pool optimization will re-use
the same Object and it will look like the error has not been modified (hence the
blinking behavior reported in the bug).
For this reason, TextView has a mErrorWasChanged flag. The fix is to export methods
that can use this flag as in done inside TextView when a physical keyboard is used.
These methods are hidden.
Change-Id: Ie3ec59a368f3b1588b81242890b971ac48e8ff7e
This is an initial API that will allow the plugin to request to
keep the screen on.
companion change is in external/webkit
bug: 3331493
Change-Id: Ic18787e7ecd705a5b2e31a34ea884fd39ad9d11a
Bug 3381317
Also generalized and uniformized the use of peekInstance. Added null
tests, and isActive tests before hiding.
Change-Id: Ifd1a053fd920841333e0ebab3e2a8d26b469a0f6
o Some javadoc updates
o writeBlock -> writePages (Block means something else in NFC Forum).
o validate page offset
Change-Id: Icae54db3397d57aaa451caaa86d56e8ba82507f2
There was logic in ViewGroup that assumed that an accelerated
view must always be able to get a display list for any child
that it was drawing. One situation occurred, however, that
caused a problem with this - a contacts activity was started
and not yet attached, but was being asked to render into an
accelerated canvas. We assumed that the child would have a display
list and simply called getDisplayList(). But since that call
returned null, we later deref'd the null object.
The fix is to check whether a child can have a display list
instead of assuming that it can just because the container view
is accelerated.
Change-Id: I7de62fd597ad50720c9585d621bec02e77c171df
Bug 3266751
The boot time is 43 compared to 44 seconds, which is within the precision of the measure.
Starting 15 activities in a sequence takes 20-24 seconds with the old version and 17-18
seconds with the new pre-load (12/13 when ran a second time, all apps being loaded).
The standard deviation is high, but it seems consistently better.
Change-Id: Ieac3cb93be03e8186979b894adda29f0549b9734
