When the computed restyle damage is empty, do not do a layout. Instead,
just rebuild the display list. In the future, even that can be omitted,
but that requires changes to the compositor.
These kind of relayouts commonly happen when the cursor is moving around
the page and no style rules cause changes to :hover.
Testing: This is covered existing WPT tests and should only have
performance
impacts. Unfortunately there are currently no performance tests.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
When making last-minute changes to the repaint-only layout pass, damage
propagation was broken, meaning that full layout was always done. This
change fixes that, meaning that times in the `blaster.html` test case
now reflect those described in the original commit message from #36978.
In addition, some style repair is now fixed:
- `InlineFormattingContext`s now keep a `SharedInlineStyles` for the
root of the IFC
which is updated during style repair.
- `BlockFormattingContext`s now properly update their style.
These changes are verified by turning on repaint only layout for more
properties
in Stylo via servo/stylo#183.
Testing: Manual performance testing via `blaster.html`.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
This change adds the simplest kind of incremental layout. When Servo
detects that all style changes only require a repaint, only run stacking
context tree and WebRender display list generation. This means that
these kind of restyles do not need a re-layout. Instead, the existing
box and fragment trees will be used and the styles of damaged nodes will
be updated in their box and fragment tree nodes.
This requires a new style repair DOM traversal for nodes that have had
their style damaged. In addition, careful accounting of all the places
where we store style must happen in order ot update those styles.
Testing: This is covered by existing WPT tests as it should not change
observable behavior.
We have created a test case which shows a 50% speedup when run
in Servo, even though there still a long way to go to match the speed
of other browsers:
https://gist.github.com/mrobinson/44ec87d028c0198917a7715a06dd98a0
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 reworks the structure of reflow in `layout_thread_2020` in order to
make it more modular. The goal here is to allow possibly adding a new
fragment tree traversal and to, in general, make the code a bit more
organized.
Testing: This should not change any behavior so is covered by existing
WPT tests.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
Now that legacy layout has been removed, the name `layout_2020` doesn't
make much sense any longer, also it's 2025 now for better or worse. The
split between the "layout thread" and "layout" also doesn't make as much
sense since layout doesn't run on it's own thread. There's a possibility
that it will in the future, but that should be something that the user
of the crate controls rather than layout iself.
This is part of the larger layout interface cleanup and optimization
that
@Looriool and I are doing.
Testing: Covered by existing tests as this is just code movement.
Signed-off-by: Martin Robinson <mrobinson@igalia.com>
We were already not compiling it and not running tests on it by default.
So it's simpler to just completely remove it.
Signed-off-by: Oriol Brufau <obrufau@igalia.com>
* 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>
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>
This change splits the style and layout data in DOM nodes that is
populated by style and layout passes. This makes Servo's data design
more like Gecko's. This allows:
1. Removing the various `StyleAndLayout` data structures used by layout.
2. Removing the `GetStyleAndLayoutData` and
`GetStyleAndOpaqueLayoutData` traits. Accessing style and layout data
are now just functions on the `LayoutNode` and `ThreadSafeLayoutNode`
traits.
3. Styling now doesn't populate layout data. This is is postponed until
layout itself.
4. Allows the DOM wrappers to no longer have to be generic over the
layout data. This data was already stored using `std::any::Any` and
the new code just makes layout responsible for downcasting. Cleaning
up the generic type parameter in the DOM wrappers can happen in a
followup change.
The main benefit to all of this is that we should be able to remove
unsafe creation of `ServoLayoutNode` in layout and
`TrustedLayoutNodeAddress` entirely, because `ServoLayoutNode` will be
able to be passed directly from script to layout. In addition, this
removes one more abstraction layer from the layout DOM wrappers, making
the code a lot more understandable.
Note: This increases the measured size of DOM types, but the same data
is stored. It's simply that before that data was stored behind a heap
pointer.
Layout asserts that it never creates stacking contexts that have a zero
scale, yet it doesn't prevent the creation of those stacking contexts.
This change stops their creation at an earlier stage.
Fixes#30118.
Move display_list_builder.rs and webrender_helpers.rs
along with the new file to components/layout/display_list/
Remove apparently unused IdType enum.
Only variant used was OverflowClip.
See #19676
This will allow Servo to create ClipScrollNodes later during display
list construction, which will be necessary once rounded rectangles
are removed from the LocalClip structure. Instead of keeping track
of the ClipId of each ClipScrollNode, we keep track of its index in an
array of ClipScrollNodes. This will allow us to access them without a
hash lookup.
In order to properly handle CSS clipping, we need to keep track of what
the different kinds of clips that we have. On one hand, clipping due to
overflow rules should respect the containing block hierarchy, while CSS
clipping should respect the flow tree hierarchy. In order to represent
the complexity of items that are scrolled via one clip/scroll frame and
clipped by another we keep track of that status with a
ClipAndScrollInfo.
This allows us to have ensure_data() and clear_data() functions on the TElement
trait, instead of hacking around it adding methods in random traits.
This also allows us to do some further cleanup, which I'd rather do in a
followup.
Just use WebRender's ClipId directly. This will allow us to create and
use ReferenceFrames in the future, if we need to do that. It will also
make it easier to have Servo responsible for creating the root
scrolling area, which will allow removing some old hacks in the future.
