This part is mainly to mark the channel as urgent-start if src related
attributes in HTMLImageElement and HTMLInputElement is set and the channel is
open due to user interaction. Unfortunately, we cannot just check the event
state just after creating channel since some loading image tasks will be queue
and execute in stable state. Thus, I store the event state in elements and
pass it to the place where create the channel.
MozReview-Commit-ID: GBdAkPfVzsn
--HG--
extra : rebase_source : 715352317b4b600f8a7f78b7bc22b894bb272d27
That function has a special path for when there's no parent reflow input, and if
there's a parent reflow input there should always be a parent frame.
MozReview-Commit-ID: EYh0aE4ozBg
The patch also renames Layer::SetScrollbarData() to Layer::SetScrollThumbData()
for clarity.
MozReview-Commit-ID: DVwJ3DMl3Zs
--HG--
extra : rebase_source : 7b2bfccf1351c82bb16296635e69d5488c87a50f
This patch implements chrome-only Selection#setColors and
Selection#resetColors methods, and use it to set the background color of
the preferences search highlight.
MozReview-Commit-ID: 2U92aBCAyeh
--HG--
extra : rebase_source : b07af1f37309d8184584b298a720cd5c1382929a
In bug 1359868 we started to do this, but we bounded the pre-render region for
the entire scrollbar by the widget bounds, which is not helpful for tall
scrollframes with short thumbs.
This time, we are bounding the pre-render region of the thumb only, so a small
thumb will always be completely painted.
MozReview-Commit-ID: 5LuP5Lfahdm
--HG--
extra : rebase_source : 3ab45f979160d7991aec71020cf57c9a1e57d1ce
Again, always initialized before use, and debug-only in any case, but worth
doing I think.
MozReview-Commit-ID: BHLReSWUU29
Signed-off-by: Emilio Cobos Álvarez <emilio@crisal.io>
This avoids conflicts with mozilla::dom::FrameType.
MozReview-Commit-ID: 7aEMbHRaTFk
--HG--
extra : rebase_source : 2d01321f5ce0ec8c0e3f70984674f82678034b3c
If the percentage basis isn't provided it will be calculated from the ancestors
as before. Note that that algorithm is only valid if the frame's containing
block is the ancestor's content box, which isn't the case for grid items.
So grid layout MUST provide a percentage basis, although the given value may
be NS_UNCONSTRAINEDSIZE for either axis if the value is indefinite.
MozReview-Commit-ID: 7nvAuR43xfM
This patch aims to speed up the lookup and storage of DisplayItemData objects, by removing a level of indirection and preventing the previously required hashtable lookup in order to access these. Instead it stores an array of pointers on each frame that allows direct access to the DisplayItemData object by dereferencing the frame. Since most frames get either 1 or 2 DisplayItemData objects attached to them a specialized class is used that is of minimal size (2 * sizeof(void)) and that performs well for sizes 1 or 2.
MozReview-Commit-ID: HONKAmpk5H8
Instead of the parent style context itself. This also fixes bug 1360530.
It's not clear what should we use in this case, it depends on the resolution in
[1] and [2].
While those get resolved, this is probably ok, and gets rid of the only
styleContext->GetParent() outside nsRuleNode.
[1]: https://github.com/w3c/csswg-drafts/issues/1249
[2]: https://github.com/w3c/csswg-drafts/issues/1281
MozReview-Commit-ID: LSOgFCwQi1W
Signed-off-by: Emilio Cobos Álvarez <emilio@crisal.io>
As we are often converting from LayoutDevicePixel to LayerPixel types
in WebRenderDisplayItem code, I added a convenience overload of
RelativeToParent that takes a LayoutDeviceRect and returns a LayerRect,
even though this is a potential footgun if abused.
MozReview-Commit-ID: DABAWdOBsbV
This also happens to fix other bugs, like making display: contents pseudos
animatable, which weren't before.
MozReview-Commit-ID: LhwTPNbFvSZ
--HG--
extra : rebase_source : 785105b08d6bfa15ad257e61b769a263c6810ad0
This renames ShouldClampMinSize to ShouldApplyAutoMinSize, which is more
accurate, and caches the value on the grid item for later use.
No functional changes other than that.
MozReview-Commit-ID: E1mvBZsMIxh
The COMPUTE_SIZE_SHRINK_WRAP flag inhibits stretching in the inline axis
which means the item won't shrink below its min-content size. In this
case min-width:0 should allow it to do that and since the column size is
smaller it should shrink, which in turn affects the measured block size.
MozReview-Commit-ID: AxQcMkYrxS5
Practical changes:
1) Some additional method arguments are nullable or optional, which
matches Chrome/WebKit. They make more sense non-nullable and
non-optional, but Chrome is afraid of the compat impact of changing.
2) Added [CEReactions] to deleteFromDocument().
MozReview-Commit-ID: Kg9EDubnEui
--HG--
extra : rebase_source : 1d47ee0b12b0b719159c326f789dd6e6b6000c8e
Calling CreateReferenceRenderingContext currently returns a non-recording DrawTarget even
when the nsIDeviceContextSpec within the nsDeviceContext has a DrawEventRecorder, due to
bug 1354624. Since we've already called BeginPage we can call CreateRenderingContext instead
though and that will give us a recording DrawTarget as we require to record the canvas drawing.
ScrollFrameActivityTracker::NotifyExpired() will be invoked by
nsExpirationTracker::TimerCallback() from an unlabeled runnable.
We provide a SystemGroup EventTarget for the invocation of this
callback since there's nothing within a page that would rely on
the timer firing at a particular time (i.e., it doesn't matter
when this timer's callback is scheduled, relative to other
runnables dispatched for the page).
MozReview-Commit-ID: 9QEjxCtFhve
--HG--
extra : rebase_source : 06b979835363b9c4288dd218d2a4ca2dc111169b
This patch is just flipping some logic in a way that should produce the same
outcome, so it shouldn't affect behavior.
MozReview-Commit-ID: LM4HbJD3D9w
This patch shouldn't change our layout order or paint order for flex items
(though it will change our behavior for the better when an abspos child is
present, as discussed in bug 1345873).
This patch *will* change the tab-index behavior of flex items. Previously, the
default tab order would match the visual order (i.e. it would respect "order"),
because it depends on the frame tree, and we sorted the frame tree by
"order". Now, the tab-index will come from the DOM order (the unmodified frame
tree), as the spec requires.
MozReview-Commit-ID: 9OsqQX1sEn3
This patch just adds an optional codepath that isn't taken yet, so it shouldn't
affect our behavior. (The next patch in the series will make use of this new
codepath.)
Note: the large code-comment that this patch adds is taken mostly-verbatim from
some nsFlexContainerFrame.cpp code. (The original copy will be removed by the
next patch in this series, when we switch to take advantage of this new
mechanism.)
MozReview-Commit-ID: 9pkJ346rrXg
NotifyApproximateFrameVisibilityUpdate gets the displayport so we want the base rect set before calling it.
We also don't want to record the displayport if we ignored it in the actual visibility pass.
This protects all accesses to the frame property table with a bit stored
on the frame. This means we avoid hashtable operations when asking
about frame properties on frames that have no properties.
The changes to RestyleManager, and the new HasSkippingBitCheck API, are
needed because RestyleManager depended on being able to ask for
properties on a deleted frame (knowing that the property in question
could not have been set on any new frames since the deleted frame was
destroyed), in order to use the destruction of the properties that
happens at frame destruction as a mechanism for learning that the frame
was destroyed. The changes there preserve the use of that mechanism,
although it becomes a bit uglier. The ugliness is well-deserved.
MozReview-Commit-ID: BScmDUlWq65
--HG--
extra : transplant_source : %C8%C0%CD%DC%12g%5B%8ER%3A%FF%A7a%F8%91%D4%2C%9BF%2B
This makes it so that, given a |const nsIFrame*|, a caller can retrieve
properties but not set or remove them, but with an |nsIFrame*| all
operations are allowed. I believe this is sensible since properties act
as extended member variables for things that are needed rarely, and
these are the const-ness semantics of member variables.
This also avoids the need for const_cast<nsIFrame*> to cast away const
in the following patch, which guards property access with a frame state
bit.
MozReview-Commit-ID: IJ9JnGzdH51
--HG--
extra : transplant_source : %D4%DF%04%91_q%E6%CF%B3N%82%2C%A5%CB0%3A%B6%810%ED
This replaces the usage of |PR_LogPrint| with either |printf_stderr| or
|MOZ_LOG| where appropriate. |printf_stderr| is used where a logger is not
actually available or if log levels are not being used as expected.
When calculating the thickness of the spellchecker underline, we can
directly use aPresContext to get default font size, instead of creating
a temporary nsStyleFont object. nsStyleFont does more work than default
font size calculation, and destructing its member mFont object is
expensive (see bug 1355600), so it's better to avoid it on this hot path.
MozReview-Commit-ID: GSvnZHULtL5
--HG--
extra : rebase_source : bf301c74933cf19dac45b52ff0b8b3e3675a9b1b
This protects all accesses to the frame property table with a bit stored
on the frame. This means we avoid hashtable operations when asking
about frame properties on frames that have no properties.
The changes to RestyleManager, and the new HasSkippingBitCheck API, are
needed because RestyleManager depended on being able to ask for
properties on a deleted frame (knowing that the property in question
could not have been set on any new frames since the deleted frame was
destroyed), in order to use the destruction of the properties that
happens at frame destruction as a mechanism for learning that the frame
was destroyed. The changes there preserve the use of that mechanism,
although it becomes a bit uglier. The ugliness is well-deserved.
MozReview-Commit-ID: BScmDUlWq65
--HG--
extra : transplant_source : %95%A2%9B%A1M%1F%86%A8%E0%FF%7B%E4%83%24%83%16%BE%FA%08T
This makes it so that, given a |const nsIFrame*|, a caller can retrieve
properties but not set or remove them, but with an |nsIFrame*| all
operations are allowed. I believe this is sensible since properties act
as extended member variables for things that are needed rarely, and
these are the const-ness semantics of member variables.
This also avoids the need for const_cast<nsIFrame*> to cast away const
in the following patch, which guards property access with a frame state
bit.
MozReview-Commit-ID: IJ9JnGzdH51
--HG--
extra : transplant_source : %91%D6%C7%01hC%B3z%90%B6%93%93qcAK%CB%09%D6z
This switches over layout's usage of PLArena to ArenaAllocator. This allows
us to build more files in unified sources and gets rid of various CONST masks.
MozReview-Commit-ID: Aaf3Dl2kaoz
This fixes the failure of
layout/reftests/css-grid/grid-min-max-content-sizing-002.html with the
primary patch in bug 1308876 (which causes a child whose parent is dirty
to pick up the dirty bit from the parent only the first reflow of the
child if the parent reflows the child multiple times). A simplified
testcase for that failure is
https://bugzilla.mozilla.org/attachment.cgi?id=8849771 .
The failure was caused by an error in height calculation of the first
<x> in the test. The div that is the parent of that x has a definite
height (presumably due to rules in grid), and the x has a specified
height. The div gets three reflows: two measuring reflows (from
MinContentContribution and then from MaxContentContribution) and then a
final reflow from nsGridContainerFrame::ReflowInFlowChild. Prior to the
primary patch in this bug, the div was marked dirty on all three
reflows, but with it it is marked dirty only on the first. This means
that, without the block-resize flag, the div optimizes away the reflow
of its children, since ShouldReflowAllKids returns false because
IsBResize() is false, even though NS_FRAME_CONTAINS_RELATIVE_BSIZE is
correctly set.
In order to fix this, we need to make sure the BResize flag on the
reflow state in at least some cases (see the comments in the patch for
when, and for how the cases could be optimized in the future).
Note that:
* when the dirty bit is set on the grid container, the new behavior
(with the combination of the patches) is strictly more efficient than
the old, since we will sometimes do non-dirty reflows on the grid
items (with the b-resize flag)
* when the dirty bit is *not* set on the grid container, the new
behavior is less efficient than the old, since we will set the
b-resize flag when we did not do so before. However, this slowdown
fixes existing bugs such as the one in the reftest.
Given that I was able to construct a reftest that triggers the failure
without the changes from bug 1308876, I've moved this to a separate bug.
Without the patch, grid-measuring-reflow-resize-dynamic-001.html fails,
but grid-measuring-reflow-resize-static-001.html passes. With the patch
both tests pass. (And without the patch, doing a text zoom on the
dynamic test fixes the layout error.)
MozReview-Commit-ID: JQOdVTQIkU0
--HG--
extra : transplant_source : %8B%2ARO%3B%D0%7B%EC%C9_%B3%60Sp%F9T%14X%85%DC
Note that this removes the useless (because this anon box is never absolutely
positioned) "clip: inherit" style.
MozReview-Commit-ID: iJJdv5VL06
--HG--
extra : rebase_source : d7a8697a89391eb6a80683dcbe8e3fa6f76c6a41
This makes the display-reflow information for
layout/reftests/w3c-css/submitted/css21/pagination/moz-css21-float-page-break-inside-avoid-4.html
more clear, since it means we include previously-omitted
inline-break-before statuses (not really inline in this case!), such as:
status=[Complete=Y,NIF=N,Truncated=N,Break=B,FirstLetter=N]
It was useful when debugging various tests for bug 1308876.
MozReview-Commit-ID: AL38FH6wuOa
--HG--
extra : transplant_source : 15%21%86%3C%1B%5C%C5s%7C9%26%FEo%B5%20%F7%E6%E4x
::first-letter is not support by stylo yet. (bug 1324618)
MozReview-Commit-ID: KwQk2jhn53c
--HG--
extra : rebase_source : 9bafad53a38fd11aae7be09780ee7680d9a40e1c
In stylo, nsStyleContext::GetParent() is not valid.
MozReview-Commit-ID: 5FbgUwobG3U
--HG--
extra : rebase_source : 47910f056eaf59af4e68276dd8e48b2a86634874
Per bug 1322570 comment 46, it's not easy to replace ComputedJustifyItems()
and UsedJustifySelf()'s internal nsStyleContext::GetParent() without
correctness penalty, so we use GetParentAllowServo() for now.
Also, fix the reftest.list added in bug 1334403 which incorrectly wrote test
page as reference page.
MozReview-Commit-ID: 6kAAWSFojd5
--HG--
extra : rebase_source : 2d03302115dd40281cec27bc18750b6933682855
The cleanup work includes:
part 1: make all methods 'const' for the abstract class of PropertyProvider.
part 2: make nsFontMetrics's StubPropertyProvider final.
part 3: make nsTextFrame's PropertyProvider final.
Make some methods 'const' and some variables 'mutable', so we could let all the
overridden methods stay const.
We also need to make the pass-in parameter of gfxFontGroup's GetHyphenWidth const.
Note that the comment of GetHyphenWidth seem outdated, so I fixed it as well.
part 4: make the member variables in nsTextFrame's PropertyProvider to be 'const'.
Make all the member variables 'const' except mStart, mLength,
mJustificationArrayStart, and mJustificationSpacings.
The static function AdvanceToNextTab is fixed since we only use 2 of the 4 parameters.
part 5: coding style fix for nsTextFrame's PropertyProvider.
MozReview-Commit-ID: 1kbWPwx27aQ
The relevant frame classes are:
SubDocumentFrame
ListControlFrame (only when used for (non-e10s?) comboboxes)
PluginFrame
ViewportFrame
MenuPopupFrame
The view is now created in the frame's Init() method, except for
ViewportFrame which has its view assigned by the frame constructor
via a SetView() call.
MozReview-Commit-ID: 4O7Hm1yqwIp
Views are used for PluginFrame which inherits nsFrame, not nsContainerFrame,
so it's more appropriate that these methods should live in nsIFrame/nsFrame.
MozReview-Commit-ID: 87EgCnkF5YT
I found this problem because I was debugging the failure of
layout/reftests/w3c-css/received/css-writing-modes-3/clearance-calculations-vrl-008.xht
with my patch for bug 1308876. It was failing because the red reference
box that was intended to be covered up was being mispositioned leftwards
by the width of the scrollbar, since we were not reflowing it when we
decided that the viewport did not need scrollbars. This patch fixes
that failure.
This led me to this bug, where
nsAbsoluteContainingBlock::FrameDependsOnContainer was incorrectly
testing conditions for when the values of 'top', 'right', 'bottom', and
'left' require reflow due to changes in the size of the containing
block.
The old code is incorrect in a number of cases, such as:
1. in RTL, with 'right: 100px', it will say that the frame does not
depend on its container's width since 'right' (offset-inline-start)
is a fixed offset and 'left' is 'auto'. However, since the
positioning is relative to the right edge, a change in container size
does require that the absolutely positioned element be repositioned
relative to the container's left edge.
2. In vertical-rl, again with 'right: 100px', it will make the same
mistake, since 'right' (now offset-block-start) is a fixed offset.
This is the case from the test I was debugging.
3. In vertical-rl with rtl direction and 'bottom: 100px', we will make
the same mistake because 'bottom' (inline-start) is fixed and 'top'
is 'auto', and we use 'bottom' rather than 'top'.
However, in cases (1) and (3) we actually avoid hitting the bug in these
simple-ish cases because ReflowInput::ShouldReflowAllKids() returns true
whenever IsIResize() is true, which means that
nsAbsoluteContainingBlock::Reflow doesn't even call
FrameDependsOnContainer. However, FrameDependsOnContainer should still
do the right thing because it's needed for
nsAbsoluteContainingBlock::MarkSizeDependentFramesDirty, which is only
used (from nsBlockFrame) when we reflow again for clearance or for
interruptible reflow. I haven't attempted to write a testcase for that
because it seems likely to require spending hours in the debugger trying
to trigger the right code.
This means that the only test that fails prior to the patch is
dynamic-offset-vrl-001.html, which exercises case (2), and also happens
to be the most similar to problem in clearance-calculations-vrl-008.xht.
This patch also makes the tests stricter so that we do optimize away
resizes in some cases where we're able to do so, such as
'left: 100px; right: auto' in RTL. (Or, rather, we would if it weren't
for the IsIResize() in ShouldReflowAllKids().)
MozReview-Commit-ID: 8xm1AHC21oh
--HG--
extra : transplant_source : %06%B4%40%EB%A9%C8M%F3%99%80%A9%DE%1F%1E%90%D3%F1%04W.
The calculated 'fr' value might change in the second round (after
applying min/max-size) even if it's zero in the first round.
MozReview-Commit-ID: 60moiyoWwuo
The "origSizes.isSome()" condition on the outer if-block was
a logical mistake. We should check it before re-assigning
mSizes though (this was the optimization originally intended).
MozReview-Commit-ID: AooUHYKG3jB
Currently, Selection::NotifySelectionListeners() moves focus before setting mCalledByJS to false. Therefore, if moving focus causes some calls of internal Selection methods, it may cause moving focus due to mCalledByJS being still true. So, mCalledByJS should be set to false before moving focus in NotifySelectionListeners().
MozReview-Commit-ID: F879bOmhZlv
--HG--
extra : rebase_source : 80790d24f1c78d9aeb20e5735e0c7c45111e69b3
Since GetParentObject has a chance returning nullptr, we keep the original
code path as a fallback.
MozReview-Commit-ID: LCJefr1ZH6t
--HG--
extra : rebase_source : 16b791d9555e9d3fb6e00233868249cbd08d944f
This change is to use gecko_enum_prefix in helpers.mako.rs, so that we do not
need to manually write code for nsStyleDisplay::mTransformBox.
MozReview-Commit-ID: 7UAL0iUcSIO
--HG--
extra : rebase_source : e99b7c163991df7ef3e7c0404fcef1832718a150
If we disable APZ on an individual scrollable element by setting the "disable APZ"
flag on the ScrollMetadata, we should also disable paint-skipping for that element.
If we don't do this, we end up in a situation where the APZ code is not applying
the async transform but is sending repaint requests expecting the main thread to
do repaints. Meanwhile the main-thread thinks that it can send empty transactions
and have APZ update the async transform (a.k.a. paint-skipping). So visually
neither APZ nor main-thread have an effect and the element doesn't appear to
scroll except with the tile-aligned displayport shifts. Disabling paint-skipping
on the element fixes the issue.
MozReview-Commit-ID: H4wpVLw8r8X
nsStyleContext::CalcDifference had an optimization where, when we knew
that the old and new style context have the same rule node, we knew that
the only change hints that would need to be handled are those in the
"not handled for descendants" category, generated due to explicit
'inherit' values on reset properties. This was because any changes due
to differences in inherited properties should only have generated
"handled for descendants" change hints (and thus would already have been
handled on an ancestor).
Before bug 931668, this let us avoid calling CalcDifference on structs
that only would have generated hints that we knew we already would have
handled. However, after bug 931668, we compare all structs anyway so
that we can set the aEqualStructs outparam, so we don't gain anything
from this optimization. We can still return these change hints we know
will not need to be handled, and rely on ElementRestyler::CaptureChange
to filter them out.
MozReview-Commit-ID: Ld1s2Js0i6r
There's an antipattern where nsLiteralString is used as an unnecessary intermediary in converting from CharT* to CharT*,
e.g. CallAFunctionThatTakesACharPointer(NS_LITERAL_CSTRING("foo").get());
or
NS_NAMED_LITERAL_STRING(foo, "abc");
CallAFunctionThatTakesACharPointer(foo.get());
This patch rewrites the callsites that can be trivially changed to use char*/char16_t*.
I'd somewhat like to remove nsTLiteralString::get() altogether, but in code that's less straightforward than these examples, get() is useful enough to keep.
MozReview-Commit-ID: Kh1rUziVllo
--HG--
extra : rebase_source : c21a65694d6e1c42fd88f73632f7ac8f38d005ae
Selection may be changed by methods of Selection or methods of Range retrieved by Selection.getRangeAt(). Selection::NotifySelectionListeners() is called after every selection change of each of them, so, this method must be a good point to move focus.
If new common ancestor of all ranges is editable and in an editing host, we should move focus to it. Otherwise, if an editing host has focus but new common ancestor is not editable, we should move focus from the editing host.
For consistency with the other browsers, this patch doesn't move focus to other focusable element.
MozReview-Commit-ID: 6sNsuzwqECX
--HG--
extra : rebase_source : 9ba52ab0bd1249abded81019847c85182ca1926e
Selection needs to be able to distinguish if every selection change is caused by JS (i.e., via Selection API) or the others.
This patch maps some methods of Range and Selection to *JS(). Each of them marks its instance as "used by JS" and calls corresponding method.
With this change, Selection::NotifySelectionListeners() can move focus only when it's caused by Selection API.
MozReview-Commit-ID: 1GoLHiIJ10Y
--HG--
extra : rebase_source : 02d497f9e55b3325a2e01c3041cadb90881dccb8
There's an antipattern where nsLiteralString is used as an unnecessary intermediary in converting from CharT* to CharT*,
e.g. CallAFunctionThatTakesACharPointer(NS_LITERAL_CSTRING("foo").get());
or
NS_NAMED_LITERAL_STRING(foo, "abc");
CallAFunctionThatTakesACharPointer(foo.get());
This patch rewrites the callsites that can be trivially changed to use char*/char16_t*.
I'd somewhat like to remove nsTLiteralString::get() altogether, but in code that's less straightforward than these examples, get() is useful enough to keep.
MozReview-Commit-ID: Kh1rUziVllo
--HG--
extra : rebase_source : c21a65694d6e1c42fd88f73632f7ac8f38d005ae
The <fill-rule> in the polygon() syntax is not handled because it doesn't
matter to shape-outside at all.
The reftests are numbered from 018 to avoid conflict with the w3c upstream
ones according to this list.
https://test.csswg.org/harness/results/css-shapes-1_dev/grouped/
Reftest 018 to 025 are under various writing-modes, 026 to 029 are testing
empty float area, 030 to 031 are testing polygon containing horizontal
lines.
MozReview-Commit-ID: FPUbMdZsvu6
Remove WritingMode parameter from LineRight() and LineLeft() in both
FloatInfo and ShapeInfo.
Bug 1316549 Part 3 added the parameter to compute the border radii under
writing-modes correctly. However, bug 1326407 Part 6 later made us cache
border radii, so the WritingMode parameter is no longer needed.
MozReview-Commit-ID: DFhm5d51OXJ
This is a largely uninteresting patch that just uses the DisplayListBuilder
directly. A wonderful cleanup patch will come after this. One of the more
interesting pieces is the use of PushBuiltDisplayList. This is needed for
handling empty transactions. See https://github.com/servo/webrender/pull/934
for more info.
It needs this #include in order to get the (inlined) impl for its call to
PropagateRootElementWritingMode().
MozReview-Commit-ID: 35OGSHlRqFA
--HG--
extra : rebase_source : 71cedd7ee4feb389cc8b41c626b41e642c26a0c2
This basically just splits the enum in two and does the necessary plumbing. The
worst part is that now DisplayItemLayers need to have two arrays of commands.
Fortunately, this will be going away in the future.
This makes it so we don't send the child side commands to the parent.