When one handed mode activated, user swipe-up to exit usually
cross over the NavBar region, and then invoke TouchController
intercept touch event to trigger All Apps drawer on Home.
To enhanced the UX of gesture conflict of exit OHM & All Apps,
notify TouchController throught LauncherActivityInterface,
and Launcher dispatch onOneHandedModeStateChanged() event to
all mTouchControllers in DragLayer that touchController can
adjust the touch slop by it's SingleAxisSwipeDetector.
Test: manual trigger One handed mode and swipe-up to exit
Test: monitor minDisplacement of SingleAxisSwipeDetector
OHM activated : touchSlop x multiplier
OHM deactivated : touchSlop x 1
Test: check All Apps doesn't mis-trigger when exit one handed mode
Bug: 186235522
Change-Id: I7b9e6e7fa898231697d1866186a5f9b1717a9aa3
There is a case when one handed mode triggered(Activated), all apps
drawer is very easy to trigger while user swipe up around NavBar
region to exit one handed mode. Since System Gesture monitor regsion
is small on screen bottom, swipe-up gesture usually cross over NavBar
monitor region and invoke launcher touch controller intercept touch
event and introduce unexpectedly trigger all apps drawer.
Adding onOneHandedModeStateChanged(boolean activated) for controller
be able to adjust the touch slop by multiplier, we can set a larger
multiplier when the visible window size translate become smaller
and make swipe gesture not too sensitive.
Test: manual swipe up to swich "home <-> all apps" and monitor
minDisplacement of SingleAxisSwipeDetector
Test: Trigger one handed mode and swipe up to exit one handed mode
check the minDisplacement of SingleAxisSwipeDetector
Bug: 186235522
Change-Id: I9729cd408d85b2b22582bf800e28d1471fc06980
- When calculating motion delta, it should consider the scale of pagedView
- Adjust significant threshold of tablet quick switch to 15% of page width (roughly 100dp)
Fixes: 188786242
Test: manual on quick switch
Change-Id: Idaa6c5b721decb573e97158c5fbcd67fa224f9d1
Also, scroll after collapsing as well to preserve the header position.
LinearLayoutManager seems to not count/double count the top padding of
the RecyclerView when scrolling to the position. Scrolling with the
offset and deducting the top padding works around the issue.
There's still some occasional weirdness that needs to be investigated,
but this works well...most of the time.
Bug: 183378651
Test: locally
Change-Id: I0ba85fb65411991ef781f08a69faaa993a7d7fd0
This keeps the row in view. Currently, this will scroll the row to the
centre-bottom of the screen due to the top padding being double counted,
but that will remedied in a future CL.
This also resolves the issue where the last row's widgets aren't visible
by handling that case specially
Fix: 188665456
Bug: 183378651
Test: verified locally
Change-Id: I9acb9087a8cdaf130ac5955c810c96462b368f36
During animations, the view group child may not correspond to the
visible views that appear at the top.
The previous logic is kept in the case that the layout manager returns
null for any reason, which can happen before layout has occurred for the
position.
Fix: 189588014
Test: verified locally
Change-Id: Ie8b5dcef50287e9e90a21f86e30a1ebcbbcba30f
Holding a reference to AssistContentRequester prevents it from being deleted by the garbage collector while it is still in use.
Bug: 189985872
Test: Manual (Pixel 3A)
Test: m -j RunLauncherGoGoogleRoboTests
Change-Id: I204aab907175a171c496c3bd146a8b23d2bbcf16
Don't allocate animators when there is no animation to do. The work was
not required.
Bug: 189492167
Test: Local build, run and trace analysis
Change-Id: I111768b055ed636aa92d5d9d6b799f316a568380
The call into system server is synchronous, so make the call off the main
thread to avoid jank.
Bug: 189251291
Test: Local with flag enabled
Change-Id: I1787a0ad68488755bf19e813ecfe9fc079cfaed8
Split display support was previously based on the sizes of supported device profiles. If a device supporting split display increased display size too much, it could trigger a grid migration (due to number of hotseat icons displayed vs in the DB changing as it was no longer seen as split display). The migration would not run however, as the grid had not actually changed, but would still have cleard the DB in preparation of the migration.
By counting the number of supported profiles instead, we can estimate the number of screens invariant of display size changes and avoid undesired grid migrations.
Test: manual
Fix: 187689871
Bug: 187689871
Change-Id: If740c501cab0e80ef6144356ec5618ee30134ed7
Styles are forked from SettingsLib to avoid using any private APIs.
Test: Settings root, all apps, and about have new style on S
Bug: 187732263
Fix: 188254205, 188012913
Change-Id: I0ca55eecbc82cfa5d0b8707beb4bb7c82b2aa2b1
Bill Lin
Zak Cohen
Alex Chau
Jon Miranda
Stevie Kideckel
Shawn Lin
Alina Zaidi
Pat Manning
Jon Spivack
Weilun Du
TreeHugger Robot
Wei Sheng Shih
Brian Isganitis
vadimt
Lucas Dupin
Tony Wickham