When deleting by key, make a defensive copy of selection...probably fixing a latent bug.
Bug: 28173827
Change-Id: Ib1ad2214c1f95fa703be227da33b25a5d4021b28
bug:28144676
Calling ClipArea::clipRegion from within ClipArea::clipPathWithTransform
has us handling op-based special casing twice, which caused all clip
paths to appear to be replace ops.
Change-Id: Ib842db53ffed4eee29470f773d59a3a1d07a1a0e
USER_PRESENT is sent via the background queue. A delay
there can cause us not to recognize that the user has
unlocked and prompt for the credential again, when trust
or fingerprint would be sufficient.
Also removes an obsolete reference to USER_PRESENT from
TrustManagerService.
Change-Id: Ie8d1a180170df5f0c8f9e71660504fd71eeacd99
Fixes: 27830458
This updates Android Keystore developer documentation to clarify how
to replace the self-signed certificate create by Android Keystore when
it generates a new key pair. Some developers are attempting to use
KeyStore.setCertificateEntry which is the wrong method for this. The
correct method is KeyStore.setKeyEntry.
Bug: 28152878
Change-Id: I306447b7792ecad5fbb49bd691a57bedb5207003
The new getIccAuthentication should be used. All callers have been updated
to the new API. Remove the old API in this change.
Bug: b/27360179
Change-Id: I160974d53bb6477666b3e1d457accac45cc06bfc
We need to wait until third-party apps can start before we try
binding to the wallpaper.
Bug: 28166684
Change-Id: I6ef559a667104e830e97da68b437ff592816f6f3
Repeating a 0-duration animation makes no sense. In the case of
battery saver mode, all animators are set to 0 duration, and
repeating the 0 duration animations not only waste battery power
but also potentially produce flickers on screen. In this CL,
0-duration animations are skipped to the end, regardless their
repeat count.
Bug: 25451472
Change-Id: I20f9dc2f0ff9c027782a8363ff4cf4a4d390736c
For pre-N, we have inconsistent behavior between ValueAnimator and
AnimatorSet in the case of calling end() when already ended:
ValueAnimator would start() and immediately end, whereas AnimatorSet
would be no-op. We made a decision to be consistent within Animation
Framework from N forward, which means that AnimatorSet will have the
new behavior of starting and immediately ending just like
ValueAnimator. This new behavior will be guarded by an API check.
Bug: 25601129
Change-Id: I2d952a93d8521c547ec8cde173c80d1d8ead0639
If the thread a toast is shown on is shut down,
the attempt to post the hide message to it's handler
will fail and it will never request removal. If this was
only some application background thread we will also not
receive a death notification. It seems best to use a timeout
to ensure we don't need the clients cooperation, espescially
as toasts can keep the screen on.
Bug: 21693547
Change-Id: I1d6e54ded5b9e2050daedc4d263e2e21fbe69862
When starting the "App may not work in split-screen" activity which
is translucent we used to transfer the starting window, and then the
frame was drawn quickly so the starting window from the behind was
removed, leading to a black hole.
Bug: 28094732
Change-Id: I828f0ea8b0465a5a113fd76a82d33177c8faa566
In findDropDownPosition it's entirely possible for width/height
to be -1. If this is so, and the popup is anchored to the right,
we could fail to see that it is offscreen (since we think the width
is -1), and so fail to do our position adjustment to move it to the
left. I think this was previously covered up by window manager bugs
with child windows that requested coordinates outside of their
parent frame. To fix this, we ontinue to pass the same value to
the window manager, but use the width/height we expect to receive
for local layout calculations.
Bug: 28085451
Change-Id: Ia04ca3fcd17ad8819615b5ff42f7923462ce4b42
