Add AnimationComponents.ATOMIC_OVERVIEW_PEEK_COMPONENT, and rename
previous ATOMIC_COMPONENT to ATOMIC_OVERVIEW_SCALE_COMPONENT.
When SWIPE_HOME is enabled:
- Overview lives to the left of Workspace, which is encoded in
LauncherState.NORMAL.getOverviewScaleAndTranslation().
- Create atomic animation based on ATOMIC_OVERVIEW_PEEK_COMPONENT
and OVERVIEW_PEEK state when swiping and holding from home screen.
Bug: 111926330
Change-Id: Iab6dbef7238dae15b3036d4b2a026b781eee6b4b
- After the atomic animation ends, overview jumps slightly because the
normal -> all apps transition puts it at a different position than
normal -> overview
Bug: 111926330
Change-Id: I6ca359b3ef2fc4d0b6b96229d8bf118bd0db9649
Added new SpringObjectAnimator class that wraps an ObjectAnimator so the
Object can be controlled via the Animator or via a SpringAnimation. It extends
ValueAnimator so that it remains compatible with AnimatorPlaybackController.
Code is behind feature flag toggle QUICKSTEP_SPRINGS.
Bug: 111698021
Change-Id: I1b20179ede37e89a6a6bb2a45d407cc74c99ac4e
Previously we were comparing the controller's progress at the time of
the end of the animation... which is always 1. Instead, we should be
comparing the effective final progress based on the interpolator.
Change-Id: I18110b5a3b914839860931187f39cfa11182b3e2
The public API of the library has finalized; flakiness has been removed;
code polished.
Bug: 110103162
Test: TaplTests suite
Change-Id: Ic156bbfeedb1cb9c4a48ef907f97e396e8e81936
Before, we were just controlling the components as far as we had left,
which was fine since they are just a subtle effect anyway. But now that
we don't fade out until the very end, this means that long swiping from
home usually kept recents in the background during the entire swipe and
then abruptly disappear after letting go. Now we make sure the entire
atomic animation plays by the time we reach all apps, so recents will
fade out in all cases.
Bug: 79867407
Change-Id: I7cb6790d9055bc76b4b73ed761604042a308c987
- All apps content fades in quickly so that icons are opaque by the time
they are on screen
- Recents fades out late so that we don't see it as translucent while
the transition is continuing (the translucent icon top of tranclucent
task view looks bad, for instance)
- Fix colored scrim that appears over recents - was using 0 to 1 instead
of 255
Bug: 79867407
Change-Id: I4f50423157f7870c8d0708f586a72e3e5a7b6559
> Finishing the atomic animation to that proper states are applied
> Offsetting the duration of atomic animation so that it completes along with
the swipe animation
Bug: 80432882
Bug: 86308723
Bug: 80549582
Change-Id: I8a3b44c2c7017e241b2fbdbb96c5d67edbb359d1
- Don't have double haptic on quick scrub
- Correctly check interpolated progress to determine final state,
so that it always aligns with the haptic (i.e. passing the haptic
means letting go will go to the new state)
Bug: 109709720
Change-Id: I702bb76a4c15f932f923e81a14cc49f6a9126cb8
This includes swiping/fling up from a fullscreen app and swiping/fling
down a task to a fullscreen app.
Test: swipe up when app is fullscreen
Fixes: 80543530
Change-Id: Iee259284cc6d4fe1e784532f0822ffd38cdee280
When you swipe up to overview from home, we should allow touches on
recents even if the overshoot animation isn't finished. If the
touch is below the all apps progress, we continue to intercept to
allow for double swipe up.
Bug: 80093992
Change-Id: Ifdca61e7cfc1f7a713c067596cc0c1bd3154b44e
This was leading to a pending animation running while the state had changes,
leaving user in an inconsistent state.
Various atomic animation fixes
> Ensuring that there is only one success listener on atomic animation, so that atomic
controller is created only once and to the final mToState
> If atomic controller is already running, skip animating the atomic conmonenets as
part of main animaiton
> Cancel atomic controller if it is going to a different state
Bug: 80549582
Bug: 109583168
Change-Id: Ie7a032e0fa73b1f1c2ef53055c08d16444f0385e
Previously we were logging new states as we swipe past them; now we just
log the start and end states when lifting your finger.
Bug: 80102083
Change-Id: Icec6c3dab1441023a3cdcadb7b56ecef8313cb8c
Add BackButtonAlphaHandler to set back button alpha, instead of setting
it from multiple places.
Also force back button alpha to be 1 if swipe up is disabled (b/80091187)
Change-Id: I49b63a0e6b033a3a947a847669a398f1b9ff0564
Back button changes opacity when moving the shelf during swipe up
between home screen and overview. The alpha changes depending on the
progress of the swipe up animation. When going from app to home and vice
versa, the fade animation does not tie with the swipe up progress. The
fade animation also masks the back button drawable when ime visibility
changes.
Change-Id: I51e42930640ba711e81880b385bb722d7ee8ad33
Fixes: 74581837
Fixes: 76900236
Test: swipe up from home screen to overview
- Don't update the animation to go from 0 to 1; instead, update the
interpolator to clamp to the remaining progress (b/79773309)
- Fix NPE that can happen in a race between the atomic animation
ending and the non-atomic animation canceling/ending
Change-Id: I313251dc5cbd7b931b043fc3e840bb4ab368a790
When we enter overview (overview appears, workspace disappears):
- Workspace scales down from 1f to .8f with OvershootInterpolator(1.2f) at 200 ms
- Workspace fades from 1f to 0 with OvershootInterpolator(1.2f) at 200 ms
- Overview scales down from 1.33f to 1f with OvershootInterpolator(1.2f) at 200 ms
- Overview fades from 0 to 1f with OvershootInterpolator(1.2f) at 200 ms
When we exit overview (overview disappears, workspace appears):
- Workspace scales up from .92f to .1f with DecelerateInterpolator() at 200 ms
- Workspace fades from 0 to 1f with AccelerateInterpolator() at 200 ms
- Overview scales up from 1f to 1.1f with AccelerateInterpolator() at 180ms
- Overview fades from 1f to 0 with DecelerateInterpolator(1.7f) at 200 ms
Parallax while the finger moves: Workspace translates half the distance as the shelf
Bug: 79776746
Change-Id: I319d982cf202bcd6dbbcd68ffc5c0c7853629c7e
There are 3 places we can block a fling:
- Swiping from home to all apps (through overview)
- Swiping from an app to all apps (through overview)
- Dismissing a task (in the same gesture that started by swiping down)
In all of these cases, we block the fling when crossing the threshold
to a new state (e.g. OVERVIEW), but unblock if the user pauses their
drag. With this change, the logic is consistent:
- Unblock the fling after pausing a short amount of time
- If a fling was blocked, increase the settling duration based on
velocity
Bug: 78089840
Bug: 78658678
Change-Id: I5ef52b74229418b867b26c3c6d3db2cf6e48914b
Previously we were ending the atomic animation with the assumption
that it should be complete/almost complete by the time you drag to
the next state. However, it is very easy to drag quickly enough where
that assumption doesn't hold, and thus you just see the atomic
animation pop to the end (i.e. recents showing without animation).
Now instead of ending the atomic animation, we let it continue. But
because the new state animation will have an atomic component that
interferes with the still playing atomic animation, we have to
control the atomic component separately; we control the non-atomic
components until the atomic animation ends, at which point we create
a separate controller to control the atomic components.
Bug: 76449024
Change-Id: Ia4bf19e26d0838f952d9e500fbdd8aba19856a41
State handlers can now specify atomic and non-atomic components of
their animations to states, which can be specified when creating a
new animation. There is now one atomic animation, when going from
NORMAL to OVERVIEW (and in reverse):
- RecentsViewStateController's animation (scale/alpha) is all atomic
- WorkspaceStateTransitionAnimation has atomic and non-atomic:
- Hotseat and workspace alpha is atomic, as is workspace scale
- Everything else (scrim, translation, qsb and drag handle alpha) is
non-atomic
- All apps progress is non-atomic
Also simplified dragging through overview; no longer pulls against you,
so we use an OvershootInterpolator when flinging instead of our custom
interpolator for the spring effect.
Bug: 76449024
Bug: 78089840
Change-Id: Iafac84d0c2b99ee9cf9dd5b30e2218286713b449
We manually dispatch cancel when returning to the previous state in
onDragEnd(), but could end in a bad state if getting a second,
external cancel (e.g. by pressing home). Thus, we restore the
onCancelListener after manually dispatching cancel.
Bug: 79258868
Change-Id: Idc4c33cede1d8af1829a4a744b9348d379bcf8f7
This cleans up the code and ensures that the current state animation
is always initialized when we get drag events.
Also log when we pass through states.
Bug: 78017680
Change-Id: I54ab42923ed539940ea708973ad65f5793669c11
Previously, user-controlled animations weren't properly being canceled when a
non-user-controlled animation started, e.g. when hitting home. Thus, we could
end in the wrong or inconsistent state because the user-controlled animation's
end runnable was still used. Now we add a cleanup callback for when we reset
the user-controlled animation for one that isn't user-controlled.
Also fixed a couple typos.
Tests (easier with animation durations extended):
- Swipe up and hit home before reaching overview -> land on home
- Go to overview, swipe down slightly (before threshold to go to workspace)
and let go -> return to overview without flash (recents was resetting)
- Swipe up, press home while swiping -> goes home, stops responding to drag
- Start dismissing task and hit home before it finishes (or while dragging)
-> goes home, stops responding to drag
Bug: 78249220
Change-Id: If11d8999e3fadba38c987b25af67cd2304cd859b
I'm not sure how/when this case occurs (perhaps during some transition/state
change), but manually removing the floating view matches the symptoms in the
bug.
Bug: 72996404
Change-Id: I1e7c1a338fcd16c8e07b3c49fb9c9b2097eb2708
Some animation might be running from a previous orientation, which can cuase property changes
to get skipped.
Bug: 77848165
Bug: 77774619
Change-Id: I3e198196192746abdd72a1970ff2ef407bf4aff9
We now pass the log action (e.g. SWIPE or FLING) to the pending
animation, so that the end listener can log appropriately. This
is used when swiping down or up on a task, for example.
Bug: 73783784
Change-Id: I5c2eee24e8b23cf4af68d503d3435a6d8088dd8a
Example bug:
1. Swipe up to overview and let go
2. Swipe all the way to the top of the screen, past where all apps stops
3. Swipe down
Before this change, you get reset in NORMAL state instead of OVERVIEW.
By ensuring that getTargetState() checks the drag direction before
returning a new state, we guarantee we only re-init in the case that the
state is actually changing. Otherwise it's possible to change the state
to one that is impossible, such as NORMAL when swiping up from ALL APPS.
Change-Id: I19913dded9c94228d06289780b6400e99403f378
If you're ever in overview and swipe down on the hotseat, it will launch
the focused task *unless* you entered overview from the workspace and
have not scrolled past the first task. This is a hidden state to allow
for reversibility of the swipe up from workspace.
Also moved PendingAnimation from quickstep to launcher3.
Change-Id: Iea077bc0ef7c74f6bf7b98d0a638892b9c5fe36c
- Swiping down on hotseat from overview goes to workspace
- Can swipe up through overview to get to all apps
Bug: 76449024
Change-Id: I7f76d92da976e268cc2a97e55746cca4603e6620
> Increasing the distance to travel for the first swipe
> Adding support for custom interpolators when building an animation
> When quickly swiping twice from home, finished the first animation
Change-Id: Ibc3c8667e9b927376fd99f08f0ca027f2398914b
