new timeout was set that was smaller than the previously set timeout
> Using uptimeMillis in Alarm to avoid errors due to system time
changes
> Adding an extra check in Wait in case Thread.sleep eats up
all the timeout
Change-Id: Id1fac5e8fdb81a0c3c7a6a5e50586b2a2f180d06
(cherry picked from commit a2125e1d10)
> Allow touch events on hotseat while in accessible drag as drag now
happens in spring loaded state.
> Allow drop target buttons to ignore thershold check when in
accessibility drag
Bug: 30900444
Change-Id: I88274367983fc027b2ddde3a719ca943f4f48587
(cherry picked from commit 4583d09c05)
This ensures that the intent has the package corresponding to the
shortcut publisher, rather than a market intent. It also ensures that
the intent has the EXTRA_SHORTCUT_ID attached.
Bug: 31123204
Change-Id: I05d56396b629880322e915f52bfc0605b921b0b1
(cherry picked from commit fc02c1b446)
The import logic following the behavior: Improt everything and force
run GridMigrationTask to automatically remove broken icons.
This logic would fail for hotseat as the replacement happens before
the GridMigrationTask, which will not replace the broken targets
appropriately
The cl changes some logic only for hotseat import
> After import remove any broken icons/empty folders
> When adding default icons, only add as much icons as required. Since
GridMigrationTask uses weights, it sometimes removes imported icon, if
the hotseat size becomes too large.
Bug: 30909630
Change-Id: I6ca1f25dac81649794d81aaf82c3c38d1c918d91
Since we are doing a equality check on the intent, items which already have
the bounds attached to the intent will not get matched, causing duplicate
icons
Bug: 27176189
Change-Id: Ia9ed9120b04e9c234304995aed13777a5773119e
(cherry picked from commit cbfe71dba3)
There was a race condition that caused the new color to not be
applied. For example, here's how the hotseat would become transparent:
1 Launcher is loaded for the first time; as such, there is not yet a
color defined for the hotseat, so we start animating to the provided
default (Color.TRANSPARENT). Meanwhile we start the color extraction.
2 When the color extraction finishes, we set the hotseat to the new
color. However, if launcher is paused at the time (perhaps some
retail mode content is showing), then we don't animate the change.
3 If 2 happens before the animation in 1 is complete, the color from
2 will be overriden by subsequent animation frames and thus the
hotseat remains transparent until the wallpaper changes.
Bug: 30956221
Change-Id: Iddf72379b0162f1b32883ad26ce267473e172849
Before, everything is set to APP_ICON
With this changed, pinned shortcuts are set to DEEPSHORTCUT
Change-Id: I3e17de63f58693525236290ef5cb1f909f1d6098
(cherry picked from commit 8ce6063c4a)
When touching down within the drag handle's bounds, we remove the
onClickListener temporarily (restored when touching down outside
of the drag handle's bounds). Long clicks still start the drag.
Also increased drag threshold from 12dp to 16dp.
Bug: 30816665
Change-Id: I0b33dc34bf95c0532376f2f7cf50865fa50093de
- We only want to log when the container is opened and potentially
used, not when a long press is followed by a drag-and-drop.
- Also cleaned up code that was determining the container of the
app icon, since LaunchSourceProvider.fillInLaunchSourceData()
can do that instead (it's more robust and consistent).
Bug: 30791570
Change-Id: I05b6750f26182fda8a9940ac66f1371c2d228ca9
> Check for permission on every onResume
> If the permission is different than last known permission,
reload and rebind workspace.
Bug: 30789422
Change-Id: Idfa445815e29e2336505779545507d106b33a253
mBgDeepShortcutMap is only accessed on the background thread. But
the same instance of list of values was getting passed to the UI
thread, instead of being cloned.
Change-Id: Ie7d0442d895304489ce9323ea872b9091d668ae5
This change makes the padding consistent regardless of where the app
is (e.g. folder vs workspace vs all apps) by ignoring the app's
padding and adding our own to the shortcuts container.
Also note that this padding is relative to the icon, excluding the
text beneath it. So we also hide the text when the container opens
downwards, and re-show it when the container closes.
Bug: 30604007
Change-Id: I6e51c4983a8b5d495833f86e483ebaa229ed2099
Otherwise the previous active controller will continue to handle touch
events even if it doesn't re-intercept future touches. For instance,
All Apps was handling the swipe gesture after DragLayer intercepted to
close a shortcuts container, which led to the weird behavior described
in the bug.
Bug: 30590854
Change-Id: I247b39b03d336a04659f6ce644380bf3cef8de3f
- Any change to the TextView text will cause mQuery to be set,
which will cause a new search next time refreshSearchResult
is called. We should also be checking there if it is a valid
search query before starting a new search.
Bug: 30606307
Change-Id: I08640c56199211f2aeea2386fcf699810853ab58
- Just skip the animation if we are animating to the same wallpaper
offset (which is the case when we are adding from all apps)
Bug: 28587903
Change-Id: Ib7b1828c1b099a665d68c22cb33ee62693f33f35
Sunny Goyal
Hyunyoung Song
Tony Wickham
Winson
Peter Schiller