It was very easy to forget about using `.with_specific_layout_info()` to
set the specific layout info, so it's better to make it a parameter.
In fact this already happened in the past: #36993 fixed the missing
specific layout info for flex items.
This patch fixes it for floats and atomic inlines. It also propagates it
in other cases where not doing so was not a big deal because the
specific layout info was None, but that was a fragile assumption.
Testing: Various WPT improvements
Fixes: #37898
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
Layout: Add incremental box tree construction for inline floats and
abspos
Due to false positives in the memory benchmark on CI, the previous PR
[37868](https://github.com/servo/servo/pull/37868) reverted. Now it is
resubmitted.
Signed-off-by: sharpshooter_pt <ibluegalaxy_taoj@163.com>
This reverts commit 19ceccc8eb due to a
significant increase in resident memory usage (See
https://github.com/servo/servo/issues/37887).
Testing: This is a revert due to a regression
Signed-off-by: Jonathan Schwender <schwenderjonathan@gmail.com>
The logic for laying out block-level replaced elements wasn't taking
floats into account when resolving a `stretch` inline size. By handling
them with the same logic as non-replaced elements, we fix that problem,
and reduce the amount of code.
Testing: Adding new tests
Fixes: #37861
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
This changes extend the incremental box tree construction for inline
out-of-flow-box, including the
`InlineItem::OutOfFlowAbsolutelyPositionedBox` and
`InlineItem::OutOfFlowFloatBox`.
Testing: This should not change observable behavior and is thus covered
by existing WPT tests.
Signed-off-by: sharpshooter_pt <ibluegalaxy_taoj@163.com>
This changes extend the incremental box tree construction for inline
atomic
Testing: This should not change observable behavior and is thus covered
by existing WPT tests.
Signed-off-by: sharpshooter_pt <ibluegalaxy_taoj@163.com>
Layout: Continue support incremental box tree reconstruction for
flex&taffy level box
This change reuse the flex/taffy level box from old box slot if the box
slot is valid and there is no LayoutDamage to the element.
Testing: This should not change observable behavior and is thus covered
by existing WPT tests.
Signed-off-by: sharpshooter_pt <ibluegalaxy_taoj@163.com>
This change extends incremental box tree updates to table columns.
Testing: This should not change observable behavior and is thus covered
by existing WPT tests.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
Previously, our Servo-specific spatial tree scroll offsets were opposite
to
that of WebRender and also the web platform. This is due to the fact,
likely, that `winit` wheel directionality is also flipped. This change
has both the Servo spatial tree and the API take offsets that are
consistent with the web.
Any possible changes to the meaning of wheel directionality will be
handled in a followup change.
This is a breaking change to the Servo API.
Testing: This change updates unit tests.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
This change extends incremental box tree updates to table cells. In
addition, for simplicity this refactors `BoxSlot::take_layout_box()`
into `BoxSlot::take_layout_box_if_undamaged()`.
Testing: This should not change observable behavior and is thus covered
by existing WPT tests.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
This change extends incremental box tree updates to table captions. In
addition, calls to `LayoutBox::invalidate_cached_fragment()` are moved
to the damage calculation traversal.
Testing: This should not change observable behavior and is thus covered
by existing WPT tests.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
In quirks mode, the ::marker of a bare `<li>` should be
`list-style-position: inside`,
Testing: Some WPT improvement
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
This change:
- Adds a new type of LayoutDamage that signifies that a box needs its
children recollected, because one or more of them need to be rebuilt.
- During restyle damage propagation, propagate this new damage upward in
the tree. Then box tree construction should be able to preserve any
still-valid box tree nodes from box slots.
- During BlockLevelBox job finalization, if a box slot is valid and
there is not LayoutDamage to the element, use the old box slot,
ensuring that its fragment cache is invalidated.
Testing: This should not change observable behavior and thus is covered
by existing WPT tests.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
Co-authored-by: coding-joedow <ibluegalaxy_taoj@163.com>
This improves naming of layout categories and adds tracing for each
layout phase.
Testing: This just adds / adjusts profiling categories, so doesn't need
tests.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
In layout, some parts of the code were still using parallel iterators
from Rayon even when single-thread layout was activated. This change
modifies those parts to use non-parallel iterators when
`LayoutContext::use_rayon` is not active.
Testing: It's very hard to make an automated test for this, but I've
manually
verified this by building with tracing and observing that layout runs
only on
a single thread now when loading https://servo.org.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
This stylesheet was being included both in quirks mode and in limited
quirks mode. It should only be the former.
Testing: adding new test.
Part of #37813
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
This implementation is quite confusing as it makes it harder to tell
that we are just looking for the case that `Contents` contains
`NonReplacedContents`.
Testing: This shouldn't have any functional change, so is covered by
existing tests.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
#37433 didn't handle intrinsic contributions. This patch computes the
correct SizeConstraint to be used as the ConstraintSpace's block size
when computing intrinsic inline sizes.
Testing: Adding new test
Fixes: #37478
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
Add methods to retrieve the natural, default and fallback sizes per each
axis (inline or block), rather than computing both simultaneously.
Sometimes we only want one size, and I think it's clearer this way.
Testing: Unnecessary (no behavior change)
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
`FlexContainer::layout()` was forcing `depends_on_block_constraints` to
true for `flex-direction: column`. However, it doesn't make sense to do
that only for `column` but not for `column-reverse`. Also no test is
complaining, so I'm just removing that logic.
Testing: no behavior change is intended, so just existing test coverage
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
The creation of `LayoutContext` does more work than necessary if layout
just needs to do something like make a display list and not restyle and
relayout. This change makes it so that these kind of non-restyle layouts
do not need to create a display list. In addition, the creation of
`LayoutContext` is better encapsulate
Testing: This should not change observable behavior and is thus covered
by existing WPT tests.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
- Move some of the image handling code to a separate function.
- Move reflow event debugging into layout itself and use the `Debug`
implementation to print the event.
- A few other small cleanups
Testing: This should not change behavior and is thus covered by existing
WPT
tests.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
This reduces the amount of work necessary when running layout, by making
restyle information optional in the `ReflowRequest`. When restyling
isn't
necessary, the option is `None`.
Testing: This shouldn't change any observable behavior and thus is
covered
by existing WPT tests.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
This will allow Servo to add custom types of damage in the near future
which correspond to minor phases layout. The damage exposed by Stylo
only corresponds to the major layout phses. In the future, both phases
will likely be managed by Servo itself and implementors will need to
provide their own damage system entirely.
Testing: This shouldn't change behavior and thus is covered by existing
tests.
Stylo PR: https://github.com/servo/stylo/pull/207
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
This is used to capture information about what layout phases are
necessary for a given `ReflowGoal`. It's moved closer to where these
decisions are made and it should be easier to understand what the values
mean. They had gotten a bit out of sync with how queries and layout were
implemented.
Testing: This shouldn't change observable behavior and thus should be
covered
by existing WPT tests.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Instead of doing a restyle whenever layout is requested, only do one if
script believes that the `Document` has changed in a way that needs a
restyle. In addition, track the different reasons this might be the
case. This will be used later to provide better debugging output.
In layout, when a restyle isn't requested, provide:
- an early return if layout is up-to-date enough for the reflow goal.
- skipping restyle and reflow if it isn't necessary.
Testing: This should not change observable behavior, and thus is covered
by existing tests.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
This change fixes an issue and makes a few more minor improvements to
the `ImageAnimationState`:
1. Image rooting and unrooted now happens in one step from
`Window::update_animations_post_reflow`.
2. The `node_to_animating_image_map` is now stored as a shared `RwLock`
so that it doesn't need to be taken and then replaced in the
`ImageAnimationState` during reflow. This should prevent a hypothetical
issue
where image animations are restarted during empty reflows.
3. General naming and idiomatic Rust usage improvements.
Testing: This doesn't really have any obvious behavioral changes,
because all
reflows currently trigger a restyle. It becomes a serious problem with
#37677
and this change fixes the failing test there.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Now that we are standardizing on the `_traits` crates becoming `_api`
and exposing the API of the crate that they get their name from [^1],
`script_layout_interface` becomes `layout_api` as it exposes the API for
`layout` that is used by `script` This brings the crate in line with the
naming of the other ones in `shared`.
[^1]:
https://servo.zulipchat.com/#narrow/channel/263398-general/topic/Organizing.20*_traits.20crates/with/396893711
Testing: This should not change any behavior and thus is covered by
existing tests.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
No longer store scroll offsets for elements in the DOM. Instead
consistently get and set these in layout's `ScrollTree`. This more
consistently requires layout to run when querying scroll offsets, which
ensures that they are up-to-date and properly bounded by scrollable
overflow area.
Testing: This causes several WPT tests to start passing, and one to
start
failing. In the case of
`/shadow-dom/scroll-to-the-fragment-in-shadow-tree.html`, I believe the
issue
is that we don't properly handle scrolling and shadow DOM elements.
Before, the
faulty scrolling was hiding this issue.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Add debug option `dump-scroll-tree` to print the scroll tree that had
been stored after each reflow, or log that the scoll tree is not
initialized yet..
To reduce the coupling, the debug print operation will process the
scroll tree node list that have been constructed in the stacking context
tree. It will generate a adjacency list and do preorder traversal. The
order of the tree then will depends on the order of the node in the node
list that has been constructed. Which, in turn, correspond to the
declaration order of the nodes.
This would help with the analysis and development of post composite
queries and its caching.
cc: @xiaochengh
Signed-off-by: stevennovaryo <steven.novaryo@gmail.com>
Zero-sized gradient border images cause WebRender to panic, so simply
don't add them to the display list.
Testing: This change adds a WPT crash test.
Fixes: #37432
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Highlights from the devtools trigger reflows. Previously, incremental
layout would notice that nothing about the box tree changed and skip
generating a new display list. This caused the highlights to never be
painted.
Forcing a new DL in this case makes reflows that happen *while* there is
a highlighted DOM node (which did not change since the last reflow)
slightly less efficient. We could check if the highlighted node changed
and only force a new display list if it did, but I'm not sure if
`OpaqueNode`s can be compared like that. It also seems like a very niche
issue.
Testing: This is hard to test for, so there are no tests.
Fixes https://github.com/servo/servo/issues/37500.
Signed-off-by: Simon Wülker <simon.wuelker@arcor.de>
Previously, `rendered_text_collection_steps` ignores
`TextTransformCase::Capitalize` due to limitation of iterator. Now we
handle the case outside.
Testing: Added a new test as not covered by existing wpt-test, except
for the indirectly related WebDriver test.
`./mach test-wpt -r
tests\wpt\tests\webdriver\tests\classic\get_element_text\get.py
--product servodriver`
Fixes: #37469
---------
Signed-off-by: Euclid Ye <yezhizhenjiakang@gmail.com>
Previously, layout was handling scrollable overflow and srolling area
calculation separately, only excluding the "unreachable scrollable
overflow region" at the last step. In addition, `position: absolute` was
not included in scrollable overflow calculation.
This change combines the two concepts into a single scrollable overflow
calculation and starts taking into account `position: absolute`.
Finally, `BoxFragment::scrollable_overflow_for_parent` is converted to
use early returns which reduces the amount of indentation.
Fixes#35928.
Fixes#37204.
Testing: This causes some WPT test to pass, but also two to start
failing:
- `/css/css-masking/clip-path/clip-path-fixed-scroll.html`: This seems
to fail
because script is scrolling past the boundaries of the document. This is
a
failure that was uncovered by the fixed element now being added to the
page's scroll area.
- `/css/css-overflow/overflow-outside-padding.html`: One test has
started to fail
here because now the absolutely positioned element is included in the
scroll area,
and I think there is an issue with how we are placing RTL items with
negative margins.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
This amends #37457 so that the min-content and max-content inline sizes
of a replaced element stop depending on the value of the inline sizing
properties.
In practice I don't think this has any observable effect, but now the
code obeys https://drafts.csswg.org/css-sizing-3/#intrinsic-sizes, and
it's simpler.
Testing: Unneeded
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
This change aligns Servo with both Blink and WebKit in common cases.
When the `aspect-ratio` property is set to a different value than the
natural ratio, then Blink and WebKit disagree, we match Blink.
Gecko doesn't support intrinsic min/max block sizes at all.
Note this patch doesn't fix the intrinsic contributions, they will need
to be addressed in a follow-up patch.
Testing: Covered by WPT
Fixes: #37433
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
There are currently five places that scroll offsets are stored:
- DOM: A set of scroll offsets used for script.
- Layout: An array of scroll offsets that is used for tracking
layout-side scroll offsets.
- Layout: The scroll offsets stored in the `ScrollTree`. These are
currently unset and unused.
- Compositor: The scroll offsets stored in the `ScrollTree` mirrored
from layout.
- WebRender: The scrolled offsets stored in the WebRender spatial tree.
This change is the first step in combining the first three into the
layout `ScrollTree`. It eliminates the extra array of scroll offsets
stored in layout in favor of the storing them in the `ScrollTree`. A
followup change will eliminate the ones stored in the DOM.
- In addition the `ScrollState` data structure is eliminated as these
are
now stored in a `HashMap` everywhere when passing them via IPC.
- The offsests stored in layout can now never scroll past the boundaries
of the scrolled content.
Testing: This should not change behavior and is thus covered by existing
WPT tests.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: stevennovaryo <steven.novaryo@gmail.com>
Refactor the box tree update logic to improve readability and facilitate
the addition of incremental update logic in the following PRs.
Testing: No logic changed, just covered by existing WPT tests
Fixes: Nothing to fix
Signed-off-by: sharpshooter_pt <ibluegalaxy_taoj@163.com>
The callback can be used instead, since `Size::resolve()` already takes
care of creating a `LazyCell`.
Testing: Unnecessary, no behavior change
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
This change has two parts which depend on each other:
1. An early exit in the layout process, which allows for skipping
display list construction entirely when nothing would change.
2. A simplification and unification of the way that "fake" animation
frames are triggered. Now this happens on an entire ScriptThread at
once and is based on whether or not any Pipeline triggered a display
list update.
Animations are never canceled in the compositor when the Pipeline
isn't updating, instead the fake animation frame is triggered far
enough in the future that an unexpected compositor tick will cancel
it. This could happen, for instance, if some other Pipeline in some
other ScriptThread produced a new display list for a tick. This makes
everything simpler about these ticks.
The goal is that in a future change the ScriptThread-based animation
ticks will be made more generic so that they can throttle the number of
"update the rendering" calls triggered by script.
This should make Servo do a lot less work when moving the cursor over a
page. Before it would constantly produce new display lists.
Fixes: #17029.
Testing: This should not cause any web observable changes. The fact that
all WPT tests keep passing is the test for this change.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
This is a minor cleanup that simplifies the construct of
`StackingContextTree`. The two arguments removed are derived from the
`FragmentTree` which is already passed as an argument.
Testing: This should not change behavior and is thus covered by existing
tests.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
`PlacementAmongFloats` should guarantee that the inline size of the
placement rect is at least as big as the inline size of the box,
resulting in a non-negative free space.
However, that may fail when dealing with huge sizes that need to be
saturated to MAX_AU, so this floors the free space by zero.
Testing: New crashtest
Fixes: #37312
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
This applies some fixes for CSS Grid percentage sizing. These fixes are
mostly within Taffy, but there are some changes in Servo to allow it to
communicate whether an item is replaced to Taffy.
It also updates Taffy to v0.8.0. Taffy has switched to a tagged pointer
representation of length/size styles. Much of the diff here is updating
Servo's type conversion code to use the new representation.
---
<!-- Thank you for contributing to Servo! Please replace each `[ ]` by
`[X]` when the step is complete, and replace `___` with appropriate
data: -->
- [x] `./mach build -d` does not report any errors
- [x] `./mach test-tidy` does not report any errors
- [x] There are tests for these changes OR
---------
Signed-off-by: Nico Burns <nico@nicoburns.com>
Even though we were continuing the parent BFC, we weren't updating the
SequentialLayoutState to have the correct containing block info. That
caused problem in the presence of floats.
This patch establishes an independent BFC, which avoids the problem.
This seems reasonable since outside markers are out-of-flow-ish, and it
matches Firefox. Blink implements them as inline-blocks, so they should
also establish a BFC.
Testing: Adding new tests. Some still fail because of a different issue.
Also, adding an expectation for several existing tests that were missing
it.
Fixes: #37222
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
This reverts commit 5580704438.
Let's re-land that fix when a working solution is found. Keeping that
regression makes it hard to evaluate other potential improvements.
Signed-off-by: webbeef <me@webbeef.org>
Instead of computing scrollable overflow while constructing the fragment
tree, we will now do it later. In the future this will also allow to
only recalculate the overflow without rebuilding the tree when transform
properties change, but that's left for a follow-up.
Stylo PR: https://github.com/servo/stylo/pull/194
Testing: One test is now passing (more investigation is needed), but
otherwise this isn't expected to have any effect.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
