When "handling" an uncaught exception, make an attempt to stop
profiling. In case profiling was active, this will avoid losing
the profiling buffer.
Bug: 26291225
Change-Id: Ie5f5517b46ed89d0456a4aac9d4ff3b713d1d8aa
The issue was traced to go/aog/854901. This revert was only temporary.
Bug: 26863668
This reverts commit d9c31db2e9e4da08ed43c88135fa7e220bc26ebf.
Change-Id: I6724059db18263936cc392651c621d27836f1c33
When the roaming status of the cell network changes, it is handled as
follows:
-DcActiveState gets an event (either EVENT_DATA_CONNECTION_ROAM_ON or
EVENT_DATA_CONNECTION_ROAM_OFF) and calls
NetworkAgent#sendNetworkInfo with updated NetworkInfo.
-ConnectivityService then gets an EVENT_NETWORK_INFO_CHANGED and calls
updateNetworkInfo with the new info.
With the previous code, the NetworkAgent's NetworkInfo object would be
updated internally, but because the state of the connection had not
changed from CONNECTED, the change would be considered a duplicate and
otherwise ignored.
With the new code, we call notifyIfacesChanged() when the state is the
same but the roaming status has changed. That should be all that is
necessary, as roaming is not exposed as a link property or network
capability today.
Bug: 26545456
Change-Id: I33d32d1fa84ce41dbe01542e5b498eda4b744b74
This feature is very old and hasn't been used in many releases. In
addition, it was never taught about multi-user layouts, and it would
be incredibly hard to maintain in the new FBE world where CE keys
may not appear until several reboots after an OTA.
Bug: 26854442
Change-Id: I3e88d38fde671e41ea9bdd790173d8b93c8a5c12
If we try scheduling a pass before the system is ready, record a
pending event and dispatch once we're actually ready.
Bug: 26863668
Change-Id: I028285383c8bbe8b653aeaa7544eefe3d41277bc
Log to two new histograms - one for operations originating from system
providers, the other for operations originating from external providers.
Events in these histograms are bucketed by
- type (copy, move, delete, etc)
- destination (system vs external provider, with a special bucket for
intra-provider operations)
- result (success vs failure)
BUG=26737132
Change-Id: Ic24c7a22fa3c1227e23cfbd42ae80ca1f70a19ba
Orientation change handling was done using a config change listener
but it is not a sufficient trigger for when the toolbar needs to be
resized. It is called before the layout has changed. A better trigger
for resizing the toolbar will be on layout change. At this time, we
get an accurate measurement of the drawing area.
Bug: 21816857
Change-Id: Ia4d752f12b64b64256c1c41d5ede5d2e29a4f408
And you might ask yourself, what is this state_single? You might ask
yourself, how did it get here?
Change-Id: Ib7f95c2f8bd2028e2bf96281830755abdbcf0527
Changes activity manager and window manager to use resizeMode
as defined by ActivityInfo#resizeMode instead of a boolean.
Bug: 26774816
Change-Id: I8cef46d9fba6bfdd21df7da63ed5d5330ad03d4b
The code to size and position the floating toolbar and it's content
when there was no overflow menu was omitted. This caused bad
rendering of the toolbar and it's content when an existing toolbar
object was updated to go in the "no overflow" state.
Fixed the issue by properly updating the floating toolbar size
and positioning of it's contents in the "no overflow" state.
Bug: 25973599
Change-Id: I9305140f493bc36b87328571cd7645f184d9cf30
We allow the profile owner of a managed profile to set the name of the
managed organization.
This name is used as the default header message shown in the confirm
credentials screen a.k.a. work challenge.
Bug: 26638631
Change-Id: I03c5acc9fffe06cdb9d0d60dd1580b20e21783b1
