This matches patch 2, and also fixes an incorrect use of eRestyle_Self
on the parents of pseudo-elements in order to restyle those
pseudo-elements, where it would not previously have been effective.
This should all be temporary, since this code can go away with bug
960465, when animation phases are removed.
This (like patch 1) posts restyles directly to the pseudo-element
content nodes, which is a new thing as of this bug. Previously we'd
have posted eRestyle_Subtree restyles to the pseudo element's real
element (i.e., the parent of the pseudo-element content node).
This changes the way we post animation restyles for ::before and ::after
pseudo-elements with animations on them.
This (like patch 2) posts restyles directly to the pseudo-element
content nodes, which is a new thing.
This isn't needed right now since AddStyleUpdatesTo is currently only
used when updating main-thread-suppressed animations running on the
compositor. However, it will be needed once we depend on
AddStyleUpdatesTo for bug 960465. And it will have an effect now since
AddStyleUpdatesTo actually adds all animations rather than only the ones
that are suppressed from running on the main thread.
This patch stores the animation name on the Animation object rather than its
AnimationPlayer. This is because Animation objects don't have a reference to
their AnimationPlayer but their AnimationEffect needs access to the animation
name.
This patch also adds an accessor for AnimationPlayer to get the name from its
Animation (since players *do* have a reference to their source animation
content).
This patch fixes a regression from
https://hg.mozilla.org/mozilla-central/rev/31695984cfe2 (bug 1025709). That
patch replaced the EnsureStyleRuleFor method on ElementTransitions and
ElementAnimations with a common method in CommonElementAnimationData.
ElementTransitions::EnsureStyleRuleFor would create a new style rule if
there was no style rule (mStyleRule == nullptr) or if the refresh time was
old (mStyleRuleRefreshTime != aRefreshTime).
ElementAnimations::EnsureStyleRuleFor, however, would create a new style rule
only if mStyleRuleRefreshTime was null or old since a null style rule may
still be valid for animations (unlike transitions). If we bail as soon as we
a null style rule we would never update mNeedsRefreshes when the animation
finishes.
The unified version of EnsureStyleRuleFor in CommonElementAnimationData
adopted the behavior from ElementAnimations checking for a null or old
mStyleRuleRefreshTime.
However, nsTransitionManager::StyleContextChanged sets mStyleRule to nullptr
to indicate that we need to generate a new style rule. This means that we
will fail to create a style rule for the transition in some cases.
This patch addresses this by making nsTransitionManager::StyleContextChanged
set mStyleRuleRefreshTime to a null timestamp. Setting mStyleRule to nullptr
is no longer necessary since EnsureStyleRuleFor will do this when necessary
and nsTransitionManager::mStyleRule is only used after calling
EnsureStyleRuleFor.
bug512295-1.html and bug512295-2.html are sending key events which hides
touch caret. Turn off the touch caret preference to fix test cases that
ought to be failed by definition of the touch caret behavior.
There's no need for this method to turn off blinking anymore. Its only
caller already calls SetCaretReadOnly to achieve the same effect. That
means we don't actually need the mIsBlinking flag after all.
--HG--
extra : rebase_source : 9d4c31282ed280c0f822f1d9f7fa8ae1c2ba6cab
A few things got mashed together here:
-- Inline KillTimer/PrimeTimer into their callers.
-- Instead of having to call StopBlinking and StartBlinking together,
change StartBlinking to ResetBlinking and have it set up
the correct blink state and reset the blink cycle.
-- nsCaret::NotifySelectionChange needs a SchedulePaint
-- nsCaret::DrawAtPosition needs a ResetBlinking
--HG--
extra : rebase_source : abc7fd78c4f20b787b212e1e3f13226a1ccff16b
This is the core of the whole patch set.
Now GetPaintGeometry/PaintCaret figure out on their own almost all the state
they need every time we paint. So when caret flags change, all we need
to do is SchedulePaint. We don't need to fiddle with mDrawn and most of the
logic in DrawCaret is obsolete. (In fact, it was duplicated by GetGeometry
and friends, and we're removing that duplication.) EraseCaret, CheckCaretState
and UpdateCaretPosition are also obsolete.
We need to have GetPaintGeometry/PaintCaret choose the correct content node
and offset, either getting them from the Selection or using specific data set by
DrawAtPosition. This logic, plus a bit of other code shared between them, is
put into the helper GetFrameAndOffset.
--HG--
extra : rebase_source : e777605dd2507ae043e9f82d0a30e23aa06e0c12
Instead of checking the mysterious mDrawn state (which is evil and will be
removed), let nsCaret::GetPaintGeometry take sole responsibilty for deciding
whether to draw. It takes the nsStyleUserInterface checks. It also needs to
check blink state, which is made possible by separating blink state into
the mIsBlinkOn flag.
--HG--
extra : rebase_source : 4a4796c37bc9ec7c25ffb2a320f9484cee1dc52f
This also fixes what appears to be a bug. MustDrawCaret returned true
when mShowDuringSelection is set even if the caret would otherwise be
hidden due a popup showing. That doesn't make sense.
--HG--
extra : rebase_source : e05e0892a85448dbc6666e23a0dbc7fa21a9c61e
This code is somewhat tricky. nsCaret::ComputeCaretRects was checking to see
if we have a bidi keyboard, and if so, what direction it's set to.
If the direction changed from the last direction seen for *this caret*,
we fired a SelectionLanguageChange notification on the caret's current
Selection. This looked bogus because the caret can be switched between
selections so it would seem some selections won't get a notification when
they should, but that's how it was. Also, when the SelectionLanguageChange
notification fired we then didn't draw the caret in that iteration, which
seems even more bogus.
This patch fixes all that by moving the logic to fire SelectionLanguageChange
out to GetPaintGeometry and firing the notification every single time without
trying to detect whether the state has changed or not. I carefully examined
the implementation of SelectionLanguageChange and I'm pretty sure it's
idempotent so this should be correct. That doesn't look like an
expensive function, and runs at most once per window paint, so I'm not
worried about perf. Because we now fire SelectionLanguageChange before
reading selection or frame state, it should be fine to carry on after
calling SelectionLanguageChange and drawing the caret based on whatever
changes SelectionLanguageChange has performed.
This also lets us remove mKeyboardRTL, which as noted above seems inherently
bogus.
--HG--
extra : rebase_source : 3ddfd10f6f30033e090e72b4bb43f2695218752e
Also, moves the "If the offset falls outside of the frame" check from
PaintCaret to GetPaintGeometry so we do less work in that case.
UpdateCaretRects is no longer needed.
--HG--
extra : rebase_source : 4b2925952a34d0388ae44c642129ce9015c367ea
This duplicates some code, but later patches will modify the callers and then
eventually we'll re-share common code.
--HG--
extra : rebase_source : 53f4756e87aadf22046972ef9102c190fbb35132
This is the start of the changes to caret-drawing proper.
The idea is to combine GetCaretFrame and GetCaretRect into a method
GetPaintGeometry which looks like GetGeometry but returns values
needed for painting (i.e. including bidi decorations, and returning
a null frame if we're not supposed to paint due to specific caret
state, e.g. in the "off" phase of the blink cycle).
Mostly a straightforward refactoring but there are a few interesting changes:
-- nsDisplayCaret stores its bounds instead of getting them from nsCaret on
demand. Eventually those bounds will not be stored in nsCaret at all.
-- nsDisplayCaret::GetBounds returns true for aSnap. nsCaret draws snapped
rects, so why not.
-- I removed "if (caretRect.Intersects(aDirtyRect))" in EnterPresShell.
As far as I can tell, this check is incorrect because it doesn't take
transforms into account. Since there's at most one drawn caret per window,
hence we do this at most once per paint, I don't think there's any real
performance advantage to having this check.
--HG--
extra : rebase_source : c98d3a5994478b482d19cc2e2ac83ab51bd17e00
The forward declaration of Selection in nsCaret.h will be used in later patches.
--HG--
extra : rebase_source : d1b749adac983c04d3365bb6bfb76a50101beeb5
This patch started an attempt to remove nsFrameSelection.h from nsCaret.h
and metastasized into a rather large refactoring patch that removed it
from some other header files as well, and changed nsFrameSelection::HINT
into a global-scope enum with better names. I also converted bools
into CaretAssociationHint in a few places where that was appropriate,
but there are still some more places (GetChildFrameContainingOffset)
where bools need to be converted. I figured this patch was big enough already.
--HG--
extra : rebase_source : cc618ef60e707e1360644340a2648de389383da0
GetGeometry is used in two different ways. Sometimes it's used to get
information about a particular caret. Sometimes it's used to get
information about a particular selection that's not associated with
a caret. Splitting GetGeometry into a non-static version for the former
and a static version for the latter makes this more clear. Also it saves
code since for the latter version we don't have to get an nsCaret first.
--HG--
extra : rebase_source : b7730dac56b308a82b79b175749234c9a92b6f59
We don't need to store the blink rate. Instead we can just fetch it whenever
we need it. However we do need a flag to handle the case where
nsCaret::DrawAtPosition disables blinking.
--HG--
extra : rebase_source : 4f36825a8e22edd3a2e37edb1fae5675c85adadd
I was inspired by the previous patch to remove FrameLayerBuilder.h from
nsDisplayList.h too.
--HG--
extra : rebase_source : b34a01bea5865da4737817e2396b98d54cc6d1bf
mCaretAspectRatio doesn't need to be stored. We can recompute it
whenever we need it.
--HG--
extra : rebase_source : 065a6996e920800edf5cbcc7f072f0439ef84553
mCaretWidthCSSPx doesn't need to be stored. We can just recompute it
whenever we need it.
--HG--
extra : rebase_source : 850b556651afade0da543e281e5fa54235969ab4
These used to be needed for linking with non-libxul builds, but are no
longer needed. The obsolete comment is fixed in a later patch.
--HG--
extra : rebase_source : 034f69b647bd64d4353e51a15e166094b7b2e638
To make it more clear what's going on, make CaretBlinkCallback protected
and reorder the public methods to put the state-changing "API" methods first.
--HG--
extra : rebase_source : fddf0ea4b9269b2d431d0b75c40f99a9140138a1
GetCaretFrameForNodeOffset only uses the nsFrameSelection and mBidiUI
from its nsCaret. For mBidiUI we can just get the pref directly whenever
we need it. By modifying GetCaretFrameForNodeOffset to take nsFrameSelection
as a parameter, we can make it static to make it clear it isn't really
related to the state of the nsCaret.
This may also fix a bug in Selection::GetPrimaryFrameForFocusNode where
things would go unexpectedly wrong if mCaret is temporarily observing
a different selection to the Selection.
--HG--
extra : rebase_source : cdd59f6e20cd1060bc5d2325cb3adb5e5c4a1d2c
Fixes the assertion failure with text:
"###!!! ASSERTION: Wrong line container
hint: '!aForFrame || (aLineContainer == FindLineContainer(aForFrame) ||
aLineContainer->GetType() == nsGkAtoms::rubyTextContainerFrame ||
(aLineContainer->GetType() == nsGkAtoms::letterFrame &&
aLineContainer->IsFloating()))', file
/home/sgbowen/builds/mozilla-central/layout/generic/nsTextFrame.cpp, line 1259"
which occasionally appears when opening pages with ruby or when running ruby
reftests.
Updates the manifest for ruby reftests to the current expectations (adjust
assertion counts, etc.)
To account for spacing between bases or text boxes during reflow, the line
layout which manages the bases updates its inline direction coordinate based on
the preferred inline size for the corresponding text boxes. Next, the base is
reflowed at the correct inline coordinate. Each paired text box is then also
reflowed at the proper inline position determined by (1) the current position of
its corresponding base and (2) its own preferred width.
In computing intrinsic widths, accounting for spacing is less complicated. The
minimum intrinsic width is the width of the widest ruby column, and the
preferred intrinsic width is the sum of all the ruby column widths. Each ruby
column width is the maximum width of its base box and text boxes. These
individual widths are determined using GetPrefISize on the base and text boxes.
Ruby base container frames store a list of pointers to the ruby text container
frames in the segment they denote. This list of pointers is created in the ruby
frame reflow method before calling the reflow method for the ruby base
container. The list exists and is used only during reflow of the main ruby frame
and is cleared before returning from reflow.
With the full patch series, all 4 pairs of tests pass. However, with
patch 10 removed from the patch series, all of the added tests fail
(probably through some interesting interactions).
The second set of 14 tests all fail (both with and without the patch)
because we don't start transitions when the before state is
display:none (either for style contexts coming out of the undisplayed
map or because we have no current style context because an ancestor is
display:none).
The patch makes all of the first set of 14 tests pass. Without the
patch, most of these tests fail, except for the 4 testing e1 and the
root.
I presume the four tests that pass prior to the patch pass because the
restyle actually occurs and starts the transition (which is stored on
the element) before the style context is destroyed.
This isn't actually a problem for this patch series because the root
element can't have an ancestor that's reframed because it has no
ancestors, and reframes of the element itself trigger a restyling
operation that does actually start transitions.
That said, later patches in this bug hook in to ResolveStyleContext, and
other things might as well in the future, thus the FIXME.
- Verify that instant scroll-behavior is synchronous.
- Verify that smooth scroll-behavior is asynchronous.
- Verify that smooth scroll-behavior is triggered by CSSOM-View DOM methods.
- Verify that instant scroll-behavior interrupts smooth scroll-behavior
animation.
- Verify that smooth scroll-behavior is not framerate dependant.
- Verify that smooth scroll-behavior physics simulations used by animations
converge and allow the animation to reach completion.
- CSSOM-View scroll-behavior smooth scroll animations must produce the same
results indendently of frame-rate:
- Reference samples of scroll position for each frame are captured from a
smooth scroll at 120fps for variations in X-Distance, Y-Distance.
- Test samples are captured from an animation with the same parameters at
varying framerates.
- Variance in position at each sampled interval is compared to the 120fps
reference. To pass the test, the position of each test sample must match
the reference position with a tolerance of one test sample frame's range
of motion. This range of motion is calculated by the position delta of
the reference samples one test frame duration before and after.
- The duration of the reference sample animation and the test sample
animation must match within 1 frame to pass the test.
- The simulation driving the animation must converge and stop on the
destination position for the test to pass.
- Updated ScrollTo method in nsGlobalWindow to accept a
mozilla::dom::ScrollOptions parameter to select between the instant
and smooth MSD motion.
- Updated WebIDL binding boilerplate scrolling functions in nsGlobalWindow
to pass the correct value of mozilla::dom::ScrollBehavior to the
implementation and functions, activating smooth scrolling.
- These functions will need to be updated again to support the scroll-behavior
CSS property in Bug 1010538.
The assertion in nsTimeStamp::operator >= occasionally fails due to a null RHS
timestamp when called from AnimationPlayer::GetCurrentTimeDuration, which, in
turn, is called from AnimationPlayer::SetSource.
The issue appears to be that we're not correctly initializing
AnimationPlayer.mPlayState. This value can be running (0) or paused (1) but we
only ever compare against the paused state (i.e. every other value is
effectively treated as running). If we don't initialize this we can end
up in the paused state but with mPauseStart not set to a valid (non-null)
TimeStamp. This creates an inconsistent state since we assume that when we're
paused mPauseStart is non-null, hence the assertion fails.
This gets tickled because, since bug 1040543, we call
AnimationPlayer::SetSource from nsAnimationManager::BuildAnimations *before* we
set mPlayState and mPauseStart.
This patch initializes mPlayState to the running state so we don't accidentally
end up in an inconsistent state.
It also removes some (now unnecessary) initialization of mPlayState and
mPauseStart from nsTransitionManager::ConsiderStartingTransition.
In future we could also consider reworking nsAnimationManager::BuildAnimations
so that we call SetSource on the new animation player *after* setting up its
play state. That may be more correct and would further avoid this problem but
may not be necessary since all that code is likely to change in the near future.
This also includes a test that mozMatchesSelector is still exposed and works.
--HG--
rename : js/xpconnect/tests/chrome/test_mozMatchesSelector.xul => js/xpconnect/tests/chrome/test_matches.xul
rename : js/xpconnect/tests/mochitest/file_mozMatchesSelector.html => js/xpconnect/tests/mochitest/file_matches.html
This patch changes the inheritance of ElementPropertyTransition so that it is
a subclass of Animation not AnimationPlayer.
The only thing special about ElementPropertyTransition is it stores some extra
state for reversing transitions and an extra ValuePortionFor convenience method.
This reversing behavior is implemented by the transition manager by creating
a new AnimationPlayer (i.e. it is *not* a property of the AnimationPlayer). As
a result this extra state can be pushed down to Animation which simplifies the
code significantly.
In future if we implement KeyframeEffect as a separate object we may be able to
push transition-specific state down to KeyframeEffect instead.
In this fourth step of dividing functionality between AnimationPlayer and
Animation, we move the mIsLastNotification and related methods/enums from
AnimationPlayer to Animation.
It is somewhat unclear where this belongs. This member is used to determine
which event to send for CSS Animations. The thinking behind moving this to
Animation is that if an animation that has already dispatched its animationstart
event was transferred to a new animation player with a similar current time then
I think it is expected that such an animation would *not* dispatch another
animationstart event. That suggests that event-state is a property of the
Animation not the AnimationPlayer.
Obviously, this needs to be defined somewhere (namely, the CSS Animations <->
Web Animations integration spec likely to become "CSS Animations Level 4"). Once
that behavior is agreed upon, if AnimationPlayer proves to be the more suitable
home for this member then it should be relatively straightforward to move the
member back at that time.
As the third step in dividing functionality between AnimationPlayer and
Animation this patch moves the mIsFinishedTransition member and related methods
from AnimationPlayer to Animation.
At the same time we rename SetFinishedTransition to SetIsFinishedTransition.
As the second step in dividing functionality between AnimationPlayer and
Animation, this patch moves the AnimationTiming member from AnimationPlayer to
Animation.
Most of this patch is simply moving code around. However, one significant
functional difference is that Animation::GetLocalTime() uses the mParentTime
member which is set when the Animation is updated by the player it is attached
to.
Other less significant differences are:
* AnimationPlayer::GetLocalTime is renamed to GetCurrentTimeDuration
In Web Animations, animation players have a (writeable) "current time" and
animations have a (read-only) "local time".
We would call the method simply "GetCurrentTime" (instead of
"GetCurrentTimeDuration") but GetCurrentTime is the name of the method used in
the content-facing API where it returns a double.
* "IsCurrent" is defined on both AnimationPlayer and Animation with the version
in AnimationPlayer serving mostly as a convenience shortcut to the version on
Animation.
* Animation::GetComputedTiming (previously on AnimationPlayer) now makes the
timing parameter optional since most of the time it is not needed.
As the first step in dividing the functionality currently contained in
AnimationPlayer between AnimationPlayer and Animation this patch moves the set
of keyframe properties to the Animation.
These properties are returned from the Animation by a couple of Properties()
methods that provide direct access to the member variable. In future it is
anticipated that the non-const version will be replaced with an appropriate
setter function. This will likely happen when we implement a separate
KeyframeEffect object as defined by the Web Animations API.
With regards to error checking, nsAnimationManager checks the result of
AnimationPlayer::GetSource() and handles the case where it is nullptr.
nsTransitionManager, however, simply asserts that GetSource() is never null much
like it also asserts that there is only one property with one segment in the
animation. Eventually this code should be made more generic which will probably
happen in bug 999927.
Now that we have both AnimationPlayer and Animation in use we need to clarify
which object we are referring to. This patch renames a number of member and
local variables to better reflect whether they point to an AnimationPlayer or an
Animation.
This patch is mostly renaming only with one exception. Since we are touching
a number of local variables used in loops (for looping over the array of
animation players) we take the opportunity to replace a number of instances of
uint32_t with size_t since that is the preferred type for array indices now.
This patch makes AnimationPlayers pass their current time down to the Animation
they are playing.
Since all Animations need from their players is their time, this avoids adding
a pointer back to their AnimationPlayer.
This patch renames mozilla::ElementAnimations to mozilla::dom::AnimationPlayer
and moves the code from layout/style/AnimationCommon.cpp to
dom/animation/AnimationPlayer.cpp.
It also moves various helper classes needed by AnimationPlayer to
AnimationPlayer.cpp and moves them from the mozilla::css namespace to the
mozilla namespace.
Beyond that, there are no functional changes contained in this patch.
The renaming of various members and variables that used to refer to
ElementAnimation objects but now refer to AnimationPlayer objects--to give them
a more appropriate name--is performed in a subsequent patch.
--HG--
rename : layout/style/AnimationCommon.cpp => dom/animation/AnimationPlayer.cpp
rename : layout/style/AnimationCommon.h => dom/animation/AnimationPlayer.h
