Before this patch, we store each computed values in a hashtable,
nsRefPtrHashtable<nsUint32HashKey, RawServoAnimationValue>, for all
KeyframeEffectReadOnly on an element, and convert the ServoAnimationValues of
the hashtable into an nsTArray<ServoAnimationValue*> and then convert
the ServoAnimationValues of the nsTArray into PropertyDeclarationBlock
in rust. This way was really inefficient.
In this patch, we store the computed values into AnimationValueMap and
convert all AnimationValue in the map into PropertyDeclarationBlock
after EffectCompositor::GetAnimationRule.
MozReview-Commit-ID: EJ2Kl65fVeF
Later in this patch series we will replace ServoAnimationRule with a hashmap.
At that point, we would like to pass the hashmap to ComposeStyle. In order
to achieve that, this patch templatizes the 'animation rule' parameter of
ComposeStyle in both Animation and KeyframeEffectReadOnly so that it can
represent a hashmap instead.
MozReview-Commit-ID: H5bMxjN3W8n
We will need to remove animation from timeline before calling Animation::UpdateTiming() in order to fire the cancel event.
In bug 1264125, we request one more tick after calling Animation::Cancel(), however we won't need to call this request if we apply this changeset.
MozReview-Commit-ID: h0dxUdtgkl
1. Introduce ServoAnimationRule, which is an equivalent of AnimValuesStyleRule.
2. Wrap ServoAnimationRule and AnimValuesStyleRule into a struct, and
use it as the parameter of Animation::ComposeStyle and
KeyframeEffectReadOnly::ComposeStyle.
3. Uncompute the RawServoAnimationValues in ServoAnimationRule when we
need it.
MozReview-Commit-ID: HahXDYBCAhH
Before this patch we skipped KeyframeEffectReadOnly::ComposeStyle() for
animations that are not in effect.
After this patch we call KeyframeEffectReadOnly::ComposeStyle() even if the
animation is not in-effect state in order to prepare the base style for
properties that can be run on the compositor because the in-effect animation
will be sent to the compositor and might be composed onto the base style on the
compositor after the animation gets out of its delay phase.
MozReview-Commit-ID: FuAZv4jqVMJ
Note that in this patch, the mSyncWithGeometricAnimations member is never set
to true since no one calls NotifyGeometricAnimationsStartingThisFrame yet.
MozReview-Commit-ID: GSTQmfkSdoy
This should be easier to read and provide us a convenient place to check for
other cases where we need to synchronize with the main thread (such as the
change introduced in this bug where we synchronize with other animations
started at the same time).
MozReview-Commit-ID: 8iuA7P4ycwM
The check of negative elapsedDuration is basically no longer valid since
animation delay is not factored into start time any more. But still we have
somtimes met negative elapsedDuration sice we use a previous vsync time stamp
for async animations to make the animations more sync. This is not a problem
in most cases but makes two reftests intermitent failure because both of them
used steps(1, start), the steps(1, start) composed different results in the
before phase and in the active phase. To avoid this difference this patch
replace the steps(1, start) with steps(1, end).
Once we incorpolate playbackRate into GetCurrentOrPendingStartTime, we don't
need to call AnimationTimeToTimeStamp for deviding delay by playbackRate since
the time passed to AnimationTimeToTimeStamp does not contain delay any more.
MozReview-Commit-ID: IVE2IFfNgm0
This patch also makes composite order lowest to highest, as a result we also
need to replace mWinsInCascade checks with the the properties.
The mWinsInCascade membed itself will be removed in a subsequent patch.
Now we call RequestRestyle(Layer) respectively for transition and animation,
so a test case in test_restyles.html works as expected.
And now lower-priority animations are also sent to the compositor so this patch
fixed some tests in test_running_on_compositor.html and
test_animation_performance_warning.html
MozReview-Commit-ID: BchUsJbmatg
Use SetEffectNoUpdate() for the procedure of setting effect, and let SetEffect()
also calls PostUpdate(), so we can use SetEffectNoUpdate() if we don't want to
call RequestRestyle(RestyleType::layer).
MozReview-Commit-ID: CaWA34UebpH
Both Cancel() and SetEffect() need the procedure of "Reset an animation's
pending tasks", so factor it out and then we can reuse it.
MozReview-Commit-ID: C7Q5kF9aPrV
This is a pre-patch for part 5, which is trying to make our code closer to the
spec. Some methods in KeyframeEffectReadOnly belong to AnimationEffectReadOnly,
so first, use AsKeyframeEffect() to access those keyframe-related methods, and
then add virtual methods for timing-related methods to AnimationEffectReadOnly.
MozReview-Commit-ID: 1srA1f8JYeN
The previous patch in this series renamed nsCSSProperty to nsCSSPropertyID.
This patch renames nsCSSPropertySet to nsCSSPropertyIDSet accordingly.
This patch is generated by the following commands (note: if you're running
using OS X's sed, which accepts slightly different flags, you'll have to
specify an actual backup suffix in -i, or use gsed from Homebrew):
hg stat -c \
| cut -c 3- \
| tr '\n' '\0' \
| xargs -0 -P 8 gsed --follow-symlinks 's/\bnsCSSPropertySet\b/nsCSSPropertyIDSet/g' -i''
Then:
hg mv layout/style/nsCSSPropertySet.h layout/style/nsCSSPropertyIDSet.h
... and finally, manually renaming nsCSSPropertySet in the include guard in
nsCSSPropertyIDSet.h.
MozReview-Commit-ID: ASUNs7FWbKP
If we run following code, content process will crash.
------
anim1.timeline = timeline;
anim2.timeline = document.timeline;
anim2.timeline = timeline;
------
The AnimationTimeline has LinkedList variant. (called mAnimationOrder) And Animation is sub-class of LinkedListElement.
We will create the relationship of AnimationTimeline and Animation when calling the AnimationTimeline::NotifyAnimationUpdated. However we didn't remove these relation ship when setting new timeline.
So we should remove these relationship when setting new timeline object.
MozReview-Commit-ID: RDyranduTA
Currently endTime is calculated when getComputedTiming() is called. As a
result, the value returned there doesn't necessarily reflect what we are using
in the model. It would be more simple, consistent and useful if we simply
calculate this as part of GetComputedTimingAt and use it both internally and in
the result to getComputedTiming().
In some circumstances when composing style, we tweak the time of the animation
before telling the effect to compose style. This is to avoid visual flicker in
certain situations where the main thread progress is being synchronized with an
animation running on the compositor.
In the past, effects would store their latest sample time locally so when
tweaking the animation time, we would need to call UpdateEffect() after tweaking
the time, and then again after restoring it as otherwise the style composed by
the effect would not reflect the adjusted time.
Now, however, effect's always query their animation for the time so this is no
longer necessary. Furthermore, the actions triggered by UpdateEffect are not
desirable in this case because they can, amongst other things, cause the
associated EffectSet to be destroyed and recreated.
Specifically, Animation::UpdateEffect() calls
KeyframeEffectReadOnly::NotifyAnimationTimingUpdated() which:
* Calls UpdateTargetRegistration which can trigger EffectSet
destruction/creation which is undesirable in this case because we intend to
restore whatever changes we make to the Animation's state and deleting and
recreating the EffectSet will cause any pointers to it to dangle.
* Cause us to possibly reset the "is running on compositor" status.
This too is undesirable since we intend to restore the state of the
Animation immediately after tweaking the hold time so we don't want to
act as if any state has changed.
* Similarly for marking the cascade as possibly needing an update or
requesting a restyle.
In summary, all the actions performed by NotifyAnimationTimingUpdated are
unnecessary and undesirable in this situation where we are temporarily tweaking
an Animation's current time only to restore it immediately afterwards since the
actions are all involved with recognizing actual changes in state.
