Add memory reporter integration for the fragment and box trees that are
persisted in the layout thread.
Testing: Looked at the numbers for https://servo.org and
https://html.spec.whatwg.org/. The former was very small, but the latter
was 700mb.
---------
Signed-off-by: Josh Matthews <josh@joshmatthews.net>
Start storing a link to laid-out `Fragment`s in `LayoutBoxBase`, so that
these are accessible for queries and eventually for incremental layout.
Some box tree data structures lacked a `LayoutBoxBase`, such as table
tracks and table track groups[^1].
In addition, start using these `Fragment`s for queries instead of
walking the entire `Fragment` tree. Currently, this isn't possible for
most queries as `Fragment`s do not cache their absolute offsets (which
are often necessary). This change uses the new box tree `Fragment`s for
most resolved style queries.
[^1]: Note that only rows and row groups store `Fragment`s as columsn
and
colgroups do not produce any.
Testing: This is covered by existing tests.
Fixes: This is part of #36525.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
This is a follow-up to #36518, which only addressed inline formatting
contexts. However, flex formatting contexts had the same problem, so it
seems safer to address it in general.
Testing: this makes a WPT test pass
Fixes: #36570
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
It's typically a given that the min-content size can't exceed the
max-content size. However, it was possible to break that assumption when
an inline formatting context had contents with a negative outer size
(due to margins). This could lead to assert failures.
This patch avoids the problem by flooring the max-content size to not be
smaller than the min-content size. Note there is no interoperability:
https://github.com/w3c/csswg-drafts/issues/12076
Testing: adding new reftest and crashtest
Fixes: #36481
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
`LayoutBoxBase` will soon contain laid out `Fragment`s of a box tree
node in order to facilitate incremental layout and also layout queries.
This is currently missing for inline boxes, so this change adds a
`LayoutBoxBase` to them.
Testing: This should not change any observable behavior, so existing
WPT suites should suffice for testing.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
This PR introduces an initial, straightforward implementation for
displaying text selection and the caret.
This is achieved by passing the selection range and insertion point
index down to `TextFragment`, along with the starting offset of each
`TextFragment` to determine the proper range for displaying the caret
and text selection. Additionally, the `selected_style` was passed into
`TextFragment` to specify the background color.
During the final build phase, although whitespace is typically ignored
when constructing glyphs, we still need to retrieve it to render both
the caret and text selection at the correct location. This ensures that
whitespace is not overlooked when the `TextFragment` contains an
insertion point or selection range.
There are several improvements yet to be made, including:
- The caret is static and does not flash.
- The caret is not rendered when the input field is empty. (I suppose
there should be an easy fix somewhere but I haven't found it yet)
**Working Examples**
macOS
https://github.com/user-attachments/assets/f3622cbe-9fa6-40c0-b2d8-b3a8f9842c28
Windows
https://github.com/user-attachments/assets/9b008a0d-0011-4c76-a2e2-0e35869a216c
Linux
[Screencast from 03-07-2025 11_05_41
AM.webm](https://github.com/user-attachments/assets/09a311ad-f975-4450-a66c-b20be525a5ed)
---
- [x] `./mach build -d` does not report any errors
- [x] `./mach test-tidy` does not report any errors
- [x] These changes fix part of #33237 (But the cursor isn't blinking
yet)
- [x] These changes do not require tests because there's no behavior
change
Signed-off-by: DK Liao <dklassic@gmail.com>
Use `is_zero()` instead of comparing with `Au::Zero()` for zero checks.
Testing: This change does not cause behaviour change, a test is not
necessary.
Fixes: #36300
---------
Signed-off-by: Barigbue <barigbuenbira@gmail.com>
This starts to enable the fragment cache for all layout modes, except
grid. The main tricky bit here is that update points are absolutes and
these need to be laid out again in their containing blocks. We punt a
little bit on this, by forcing ancestors of update points to rebuild
their Fragments. This is just the first step.
Testing: We do not currently have layout performance tests, but will try
to run some tests manually later. Behavior is covered by the WPT.
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
This replaces `IndependentLayout` with `CacheableLayoutResult` and
stores it in `LayoutBoxBase` so it can be reused when we need to lay out
a box multiple times.
This is a generalization of the caching that we had for flexbox, which
is now removed in favor of the new one.
With this, the number of runs per second in the Chromium perf test
`flexbox-deeply-nested-column-flow.html` are multiplied by 3.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
* Migrate to 2024 edition
Signed-off-by: Simon Wülker <simon.wuelker@arcor.de>
* Allow unsafe_op_in_unsafe_fn lint
This lint warns by default in the 2024
edition, but is *way* too noisy for servo.
We might enable it in the future, but not now.
Signed-off-by: Simon Wülker <simon.wuelker@arcor.de>
* Compile using the 2024 edition
Signed-off-by: Simon Wülker <simon.wuelker@arcor.de>
---------
Signed-off-by: Simon Wülker <simon.wuelker@arcor.de>
* Use 2024 style edition
Signed-off-by: Simon Wülker <simon.wuelker@arcor.de>
* Reformat all code
Signed-off-by: Simon Wülker <simon.wuelker@arcor.de>
---------
Signed-off-by: Simon Wülker <simon.wuelker@arcor.de>
The specification doesn't say how to deal with percentages when
determining the minimum and maximum size of a table grid, so follow the
approach that Chromium uses.
Essentially, figure out the "missing" percentage from the non-percentage
columns and then use that to work backwards to fine the size of the
percentage ones.
This change is larger than one might expect, because this percentage
approach shouldn't happen for tables that are descendants of a flex,
grid or table container (except when there is an interceding absolute).
We have to pass this information down when building the box tree. This
will also make it easier to improve propagated text decorations in the
future.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
Some layouts like table need some style overrides. We were handling this
in `ComputedValuesExt`, but it was messy, unreliable and too limited.
For example, we were assuming that a style with `display: table` would
belong to a table wrapper box or table grid box. However, certain HTML
elements can ignore their `display` value and generate a different kind
of box. I think we aren't doing that yet, but we will need this.
Also, resolving the used border of a table needs layout information,
which we don't have in `ComputedValuesExt`. This patch will allow to
improve border collapsing in a follow-up.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
There were two kinds of layout tracing controlled by the same debugging
option:
- modern layout: Functionality that dumped a JSON serialization of the
layout tree before and after layout.
- legacy layout: A scope based tracing that reported the process of
layout in a structured way.
I don't think anyone working on layout is using either of these two
features. For modern layout requiring data structure to implement
`serde` serialization is incredibly inconvenient and also generates a
lot of extra code.
We also have a more modern tracing functionality based on perfetto that
we have started to use for layout and IMO it's actually being used and
more robust.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
The new version of rust allows us to elide some lifetimes and clippy is
now complaining about this. This change elides them where possible and
removes the clippy exceptions.
Fixes#34804.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Several structs and enums had a `inline_content_sizes()` method, but it
wasn't clear which ones would try to cache the result, and which ones
would always compute it.
Therefore, this performs some clarifying renaming:
- Cached ones stay as `inline_content_sizes()`
- Uncached ones become `compute_inline_content_sizes()`
Also, to simplify calls to `LayoutBoxBase::inline_content_sizes()`,
`compute_inline_content_sizes()` is moved into a new trait.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
- Remove the `LayoutBox::InlineBox` variant that was only used for
inline level boxes. Now they are stored in `LayoutBox::InlineLevel`
along with other kinds of out-of-flow and atomic inline items.
- Reduce the size of `InlineItem` by 260 bytes per item by using atomic
indirection / pointers. This adds a bit of overhead to access items in
exchange for a lot of memory saved.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
This might make caching these values a bit easier in the future.
Correcting the visibility of `ContainingBlock` also exposed some new
rustc and clippy warnings that are fixed here.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
Add a new struct `LayoutBoxBase`, that will be used throughout the box
tree. The idea of this struct is that we have a place to consistently
store common layout information (style and node information) and also to
cache layout results such as content sizes (inline and maybe later box
sizes) and eventually layout results.
In addition to the addition of this struct,
`IndependentFormattingContext` is flattened slightly so that it directly
holds the contents of both replaced and non-replaced elements.
This is only added to independent formatting contexts, but will later be
added to all block containers as well.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
* Refactor computation of preferred aspect ratios
Computing min/max-content sizes required a ContainingBlock in order to
resolve the padding and border when determining the preferred aspect
ratio. However, all callers already knew the padding and border, so they
can compute the ratio themselves, and pass it directly instead of
the ContainingBlock.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
* Put preferred aspect ratio into ConstraintSpace
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
---------
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
In order to support size keywords in block layout, we may need to call
`inline_content_sizes()` in order to compute the min/max-content sizes.
But this required a mutable reference in order the update the cache,
and in various places we already had mutable references.
So this switches the cache into a RwLock to avoid needing mutable refs.
Note OnceCell wouldn't work because it's not thread-safe.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
To compute the min-content and max-content inline sizes of a replaced
element, we were only using the aspect ratio to transfer definite block
sizes resulting from clamping the preferred block size between the min
and max block sizes.
However, if the preferred block size is indefinite, then we weren't
transfering the min and max through the aspect ratio.
This patch adds a `SizeConstraint` enum that can represent these cases,
and a `ConstraintSpace` struct analogous to `IndefiniteContainingBlock`
but with no inline size, and a `SizeConstraint` block size.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
This is the second flexbox caching change. It seeks to detect when a
relayout can be avoided in the case of a stretching flex item. This
heuristic can be combined, because currently we still do relayout
sometimes when we do not need to.
For instance currently we always relayout when a flex child is itself a
column flex. This only needs to happen when the grandchildren themselves
grow or shrink. That optimization is perhaps a lower priority as
`flex-grow: 0 / flex-shrink: 1` is the default behavior for flex.
Since this change means we more consistenly zero out the percentage part
of `calc` expressions when they have circular dependencies, this causes one
test to start failing (`/css/css-values/calc-min-height-block-1.html`).
This is related to w3c/csswg-drafts#10969, which is pending on further
discussion in the working group.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
When computing the min-content size of an inline formatting context,
we could allow a soft wrap opportunity at the start of a text run.
This shouldn't happen with `text-wrap-mode: nowrap`.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
The result of `inline_content_sizes()` may depend on the block size of
the containing block, so we were always recomputing in case we got
a different block size.
However, if no content has a vertical percentage or stretches vertically,
then we don't need to recompute: the result will be the same anyways.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
Adds support for min-content, max-content, fit-content and stretch,
for atomic inlines.
There are some new test failures because we don't support vertical
writing modes nor `transition-behavior:allow-discrete`.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
Instead of a blocking a layout thread on the generation of WebRender
`FontKey`s and `FontInstanceKey`s, generate the keys ahead of time and
send the font data to WebRender asynchronously. This has the benefit of
allowing use of the font much more quickly in layout, though blocking
display list sending itself on the font data upload.
In order to make this work for web fonts, `FontContext` now asks the
`SystemFontService` for a `FontKey`s and `FontInstanceKey`s for new web
fonts. This should happen much more quickly as the `SystemFontService`
is only blocking in order to load system fonts into memory now. In
practice this still drops layout thread blocking to fractions of a
millisecond instead of multiple milliseconds as before.
In addition, ensure that we don't send font data or generate keys for
fonts that are used in layout but never added to display lists. This
should help to reduce memory usage and increase performance.
Performance of this change was verified by putting a microbenchmark
around `FontContext::create_font` which is what triggered font key
generation.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
This will allow callers to start obeying `min-content`, `max-content`,
`fit-content` and `stretch` in follow-up patches.
The old functionality is kept as deprecated methods that we should
eventually remove.
This patch has very little impact on the existing behavior, just some
very minimal implementation of the keywords for css tables.
This also overhauls fixed-layout-2.html since:
- It had code that wasn't doing anything
- It had wrong expecations in prose
- The logic seemed broken in general
- All browsers were failing one testcase
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
This also makes a couple small improvements:
- Rename `IntrinsicSizes` to `NaturalSizes` which reflects more
modern spec language.
- Move the conversion of Stylo's `ImageRendering` to WebRender's
version to a `ToWebRender` trait implementation.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
If we encountered a preserved line break after some whitespace, we were
including the space in the max-content size of the following line.
So just like `line_break_opportunity()` was already clearing
`self.pending_whitespace.min_content`, `forced_line_break()` needs to
clear `self.pending_whitespace.max_content` too.
Also some cosmetic refactoring.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
When computing the ascent and descent in an inline formatting context,
we weren't taking into account that app units have precision limitations.
Therefore, in some cases we were getting a line height that was slightly
taller than the value specified in `line-height`.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
There were various cases like `text-wrap-mode: nowrap` and
`white-space-collapse: break-spaces` that weren't handled well.
Fixes#33335
flexbox_flex-formatting-interop.html fails now because we don't support
`table-layout: fixed`.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
This change removes the `effective_writing_mode` concept and tries to
properly implement right-to-left layout support for all non-inline
writing modes. In general, what needs to happen is that rectangles
need to be converted to physical rectangles using the containing block.
A right-to-left rectangle's inline start is on the right physical side
of the containing block. Likewise a positive inline offset in
right-to-left text is a negative physical one.
The implementation here is pretty good for most layout modes, but floats
are still a bit in process. Currently, floats are processed in the
logical layout of the block container, but there still might be issues
with float interaction with mixed RTL and LTR.
While this does move us closer to supporting vertical writing modes,
this is still unsupported.
New failures:
- Vertical writing mode not supported:
- `/css/CSS2/floats/floats-placement-vertical-001b.xht`
- `/css/CSS2/floats/floats-placement-vertical-001c.xht`
- Absolutes inlines should avoid floats (#33323)
- `/css/css-position/position-absolute-dynamic-static-position-floats-004.html`
- No support for grid
- `/css/css-align/self-alignment/self-align-safe-unsafe-grid-003.html`
- `/css/css-position/static-position/inline-level-absolute-in-block-level-context-009.html`
- `/css/css-position/static-position/inline-level-absolute-in-block-level-context-010.html`
- Cannot reproduce these locally on any platform. Very mysterious:
- `/css/css-tables/row-group-margin-border-padding.html`
- `/css/css-tables/row-margin-border-padding.html`
- Exposes bugs we have related to hanging whitespace in preserved
whitespace inlines:
- `/css/css-text/white-space/trailing-space-and-text-alignment-rtl-003.html`
- `/css/css-text/white-space/white-space-pre-wrap-trailing-spaces-023.html`
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Rakhi Sharma <atbrakhi@igalia.com>
* Use app unit in `ComputedValuesExt`
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
Signed-off-by: atbrakhi <atbrakhi@igalia.com>
* Some miscellaneous fixes
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
* remove redundant defination of `containing_block_inline_size`
Signed-off-by: atbrakhi <atbrakhi@igalia.com>
---------
Signed-off-by: atbrakhi <atbrakhi@igalia.com>
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
This implements the requirements outlined in the [flexbox specification]
about how to position absolute children of flex containers. We must
establish a static position rectangle (to use if all insets are auto)
and also align the child into that rectangle.
[flebox specification]: https://drafts.csswg.org/css-flexbox/#abspos-items
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
This change fixes all rustdoc errors and also adds a tidy check for a
very common rustdoc URL issue. Eventually rustdoc warnings should likely
cause the build to fail, but this catches those issues sooner in order
to not waste so much developer time.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
When inline atomics establish containing blocks for absolute
descendants, layout should happen with those atomics as the containing
block. This ensures that the absolute descendents have the correct
containing block and Fragment parent. This wasn't happening before and
this change fixes that.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
In particular, this takes into account that flex items may be stretched,
and if they have an aspect ratio, we ma6y need to convert the stretched
size through the ratio.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
When computing the min-content or max-content size of an element we
need to ignore `inline-size`, `min-inline-size` and `max-inline-size`.
However, we should take the block-axis sizing properties into account.
That's because the contents could have percentages depending on them,
which can then affect their inline size via an aspect ratio.
Therefore, this patch adds `IndefiniteContainingBlock`, which is similar
to `ContainingBlock`, but it allows an indefinite inline-size. This
struct is then passed arround during intrinsic sizing.
More refinement will be needed in follow-up patches in order to fully
address the problem.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
Consider a block container that establishes an inline formatting context
and has a definite `block-size` which is clamped by `min-block-size` or
`max-block-size`.
We were already sizing such container correctly, however, its contents
were resolving their percentages against the unclamped `block-size`
value.
This patch fixes the `ContainingBlock` that we pass to the contents so
that they resolve percentages correctly.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
- Rename `InlineFormattingContextState` to
`InlineFormattingContextLayout`.
- Have `InlineFormattingContextLayout` hold a reference to the
`InlineFormattingContext`, so that it does not need to be passed
around as an argument
- Have `LineItemLayout` hold a reference to
`InlineFormattingContextLayout` to avoid duplicating so much data.
- Rename some members of `LineItemLayout` to make it clearer what
they do.
- Give beter names to many lifetimes and combine some that are
effectively the same.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Rakhi Sharma <atbrakhi@igalia.com>
