Couple of overlapping issues that this CL fixes. The major change is to
offset the selection vertically this is similar to how the handles work
already (i.e. touch target is below what's being selected), and still
allows the user to see the selection start / end without covering it with
their finger.
This change fixes multiple issues:
1) Previously to ensure the finger wasn't covering the selection a
preceding or following offset was taken, this made it difficult to
select certain words and it made text on the edge of the screen hard
to select (b/21098345 and b/19965619)
2) The use of preceding and following on the word iterator was not correct
allowing grapheme clusters to be split, now the offset is calculated
with getWordStart/End which respect grapheme clusters (b/21045116)
Bug: 21098345
Bug: 21045116
Bug: 19965619
Change-Id: Id8392426cce20ad0ff47a4279c92f6ed1b0ad30e
First, when parceling a notification with no small icon:
Well, you shouldn't attempt to do this anyway, since NoMan
will reject a notification without a valid smallIcon. But
setServiceForeground parcels up the Notification on its own
before handing it off to NoMan, so it will crash on an
invalid small icon. (In general, parceling code should never
ever crash, even if the object is in an undesirable state.)
And when build()ing a notification: Same thing---don't build
a notification with no icon; you're going to have a bad
time. But maybe you're going to fix it before you hand it
off to NoMan. Or maybe it's just one page of a wearable
notification, so it doesn't really need its own icon. Either
way, Notification shouldn't crash.
Bug: 21286186
Bug: 21298403
Change-Id: Ie482cde0a3afe3aaabf07be0536551b8e4bceba0
In landscape do not allow starting selection modes if the
extracted mode has not yet started. Also remove the single
tap selection delayed runnable.
This allows for a smooth transition (as in no floating
toolbar artifacts) into extracted mode. Also fixes
(more as in hides) the double selections.
Bug: 20930071
Change-Id: Idb9c375dc25777761b11ca17ccfd9f5eedc642b4
With NLM_F_ACK set in RTM_NEWNEIGH requests we get some response from
the kernel, whether there was an error or not.
Additionally:
[1] add IpReachabilityMonitor#probeNeighbor() as a public
static method, since it actually depends very little on the
class internals and might be of larger use.
[2] add a unittest for parsing NetlinkErrorMessages.
Bug: 18581716
Change-Id: I5d62e7a9972c7440f0483c38c77677436d3a1a25
This forces the NetlinkSocketObserver thread to exit quickly, rather
than lingering until the next random netlink neighbor multicast message
arrives.
Additionally, add a small unittest to verify that multiple calls to
NetlinkSocket#close() are safe.
Change-Id: I101730fad7eee72f9c6e8a7e7bd10c634f2ceab4
Bug: 20170924
Don't pre-allocate buffers if there is a SurfaceView or other
View that has requested transparent regions. This doesn't fully
address the problem of allocating buffers when there's a full
screen SurfaceView but it at least gets back to KitKat levels
of only having 1 wasted buffer instead of 3 (the 1 being
a result of the fact that it will still draw one frame of emptiness
forcing a buffer dequeue).
Change-Id: Ied9553186bc7b111d180b63f87b92bd317cb4b97
Stash volume count from last scan, and use it to push initial storage
notifications state when listener is first attached.
Also omit disks with invalid size, which usually means they're an
empty slot with no media.
Bug: 20503551
Change-Id: I75097035aebaad70ba32437179a863f6a0910aa5
It was deprecated as part of the Apache deprecation, however it is used
by ConnectivityManager.getLinkProperties() and .getDefaultProxy().
The alternatives (e.g. java.net.Proxy) do not offer a complete
replacement as they lack certain fields (e.g. PAC URL).
Bug:21274137
Change-Id: I99163a2611a738f337fb9907349ef6255dbed6d6
...from text nodes in WebView
Add a new explicit API for setting the text style information associated
with a view structure.
Also, how about some documentation!
Change-Id: Ia948b2d66382b973d0d00a67172a281ad55ce592
