KeyframeEffectReadOnly::CanAnimatePropertyOnCompositor has a comment that says
it, "Returns true |aProperty| can be run on compositor for |aFrame|" but it
does nothing of the sort.
What it *does* do is check answer the question, "If there happened to be an
animation of |aProperty| on |aFrame|, should we still run animations on the
compositor for this element?".
This patch renames the method accordingly and moves the step where we iterate
over a given effect's animated properties from
AnimationCollection::CanPerformOnCompositor to inside this method, making this
method a class method rather than a static method at the same time.
As noted in the expanded comment, the approach of blocking opacity animations
in these situations seems unnecessary but for now this patch just preserves the
existing behavior.
This patch also moves AnimationUtils out of the dom namespace since it seems
unnecessary. We typically only put actual DOM interfaces in the dom namespace.
This is so that when we have code like:
elem.animate({ opacity: 0 }, 1000)
the resulting Animation object is kept alive by |elem| based on the following
ownership chain:
elem --(strong)--> KeyframeEffectReadOnly --(strong)--> Animation
Now, there is an ownership cycle introduced here because KeyframeEffectReadOnly
objects also store owning references to their target elements. This is broken
when the Animation finishes (if it does not fill forwards) or is cancelled
since either event will trigger a call to
KeyframeEffectReadOnly::UpdateTargetRegistration.
If the Animation fills forwards, the resource will not be released until
it is cancelled. For Animations corresponding to CSS Animations / CSS
Transitions this happens when the Element is unbound or when the corresponding
style property is updated causing the animation to be replaced or removed.
For the general case of script-generated animations, however, this cycle won't
be broken until the Element is unbound and all external references to the
Animation or KeyframeEffectReadOnly are dropped.
It's unfortunate that we can't more aggressively prune these objects but it's
what the spec currently says. I've posted to the mailing list[1] about this but
have yet to find a good solution.
[1] https://lists.w3.org/Archives/Public/public-fx/2015OctDec/0029.html
The behavior of unthrottling in case of not current animations there is the
same as on current trunk.
There are two cases to reach there I can think of:
a) 0s duration time and fill-forwards animation
b) Calling pause() after fill-forwards animation finished.
I can provide these automation tests once bug 1222326 is fixed.
The preference check has been removed from CanThrottleTransformChanges
because we already perform that check that when deciding if we should run
an animation on the compositor (in CanPerformOnCompositorThread, as called
by GetAnimationsForCompositor). Hence if the "is running on compositor" flag
is true, we can assume the preference is set (or was set when we decided to
put the animation on the compositor-- we don't worry about pulling the
animation off the compositor immediately if the preference changes while
it is running)
Based on AnimationCollection::CanAnimatePropertyOnCompositor.
The first argument has been changed to nsIFrame* so that we don't need to
get style frame for CanAnimateTransformOnCompositor again.
If this patch (and part 9) is an overkill to throttle animations having both
of properties, one can be run on compositor and another can not be, a test
case in test_running_on_compositor[1] will fail.
The test case is for an animation which has transform and background-color
properties.
Animation::CanThrottle() returns true
(then, AnimationCollection::CanPerformOnCompositorThread() returns false)
on current trunk in the test case.
Animation::CanThrottle() returns false with this patch in the test case.
If the test passes, it proves the transform animation is running on compositor
in both cases.
[1] http://hg.mozilla.org/mozilla-central/file/6c7c983bce46/dom/animation/test/chrome/test_running_on_compositor.html#l77
Do some minor revisions in struct ComputedTiming.
1. Use Nullable<double> mProgress, so remove the static const kNullProgress.
The generated ComputedTimingProperties dictionary uses "Nullable" variable,
so we replace the origin type in ComputedTiming to make it more consistent
with that in ComputedTimingProperties dictionary.
2. Use scoped enums for AnimationPhase.
In a subsequent patch, we will have another struct like
KeyframeValueEntry, but storing an StyleAnimationValue and an
ComputingTimingFunction object (not a pointer). So we split
KeyframeValueEntry into two, retaining the KeyframeValueEntry name for
the base class and naming the current one KeyframeStringValueEntry.
The bulk of this commit was generated with a script, executed at the top
level of a typical source code checkout. The only non-machine-generated
part was modifying MFBT's moz.build to reflect the new naming.
CLOSED TREE makes big refactorings like this a piece of cake.
# The main substitution.
find . -name '*.cpp' -o -name '*.cc' -o -name '*.h' -o -name '*.mm' -o -name '*.idl'| \
xargs perl -p -i -e '
s/nsRefPtr\.h/RefPtr\.h/g; # handle includes
s/nsRefPtr ?</RefPtr</g; # handle declarations and variables
'
# Handle a special friend declaration in gfx/layers/AtomicRefCountedWithFinalize.h.
perl -p -i -e 's/::nsRefPtr;/::RefPtr;/' gfx/layers/AtomicRefCountedWithFinalize.h
# Handle nsRefPtr.h itself, a couple places that define constructors
# from nsRefPtr, and code generators specially. We do this here, rather
# than indiscriminantly s/nsRefPtr/RefPtr/, because that would rename
# things like nsRefPtrHashtable.
perl -p -i -e 's/nsRefPtr/RefPtr/g' \
mfbt/nsRefPtr.h \
xpcom/glue/nsCOMPtr.h \
xpcom/base/OwningNonNull.h \
ipc/ipdl/ipdl/lower.py \
ipc/ipdl/ipdl/builtin.py \
dom/bindings/Codegen.py \
python/lldbutils/lldbutils/utils.py
# In our indiscriminate substitution above, we renamed
# nsRefPtrGetterAddRefs, the class behind getter_AddRefs. Fix that up.
find . -name '*.cpp' -o -name '*.h' -o -name '*.idl' | \
xargs perl -p -i -e 's/nsRefPtrGetterAddRefs/RefPtrGetterAddRefs/g'
if [ -d .git ]; then
git mv mfbt/nsRefPtr.h mfbt/RefPtr.h
else
hg mv mfbt/nsRefPtr.h mfbt/RefPtr.h
fi
We need to do this so effects can query their owning animation for the current
time and avoid falling out of sync. Furthermore, this pointer is needed
for a number of other bugs (e.g. bug 1166500 comment 12, or bug 1190235)
anyway.
Since getFrames() must gather all properties set at a given keyframe
offset time for a given easing function, we need to provide a total
ordering for ComputedTimingFunction objects. Until the spec defines how
to do this, we sort first by NS_STYLE_TRANSITION_TIMING_FUNCTION_*
value, then second by the four values in a cubic-bezier() function (in
order) or the integer and optional keyword in a steps() function.
Because we don't support automatic spacing of keyframes yet,
ComputedKeyFrame.computedOffset is always the same as Keyframe.offset.
Another assumption made is that the value of easing for a Keyframe
object at 100% should be the same as the value from the previous
Keyframe for the same property. An alternative would be to leave off
easing from that Keyframe, which would need the default value for that
IDL dictionary member removed (otherwise it would always be set to
"linear").
Since Keyframe.easing should reflect the {transition,animation}-timing-
function value relevant to each keyframe, we'll need to store on
nsTimingFunction the specific timing function value that was used, and
copy it down into ComputedTimingFunction for
KeyframeEffectReadOnly.getFrames() to access. This includes storing
whether the optional start/end keyword in a steps() function was
specified.
KeyframeEffectReadOnly uses IsFinishedTransition to exclude finished transitions
from certain tests. This check, however, is redundant in each case.
This is because any effect marked as IsFinishedTransition will have the
following properties:
- owning animation's PlayState() == Finished or Idle
- animation phase = after or null
- progress = null (this is because transitions don't fill forwards)
The long-term plan is to drop the mozilla::css namespace altogether. Before we
go to much further with refactoring code in AnimationCommon, we should drop
usage of the mozilla::css namespace. Specifically, this patch moves the
CommonAnimationManager and AnimValuesStyleRule classes to the mozilla namespace.
The long-term plan is to drop the mozilla::css namespace altogether. Before we
go to much further with refactoring code in AnimationCommon, we should drop
usage of the mozilla::css namespace. Specifically, this patch moves the
CommonAnimationManager and AnimValuesStyleRule classes to the mozilla namespace.
This patch is a fairly minimal rename of the AnimationPlayer interface. It
leaves a bunch of local variables and helper classes still using the word
"player". These will be addressed in subsequent patches that don't require DOM
peer review.
