Main frame scrolling after the WebRender upgrade was broken because the
main frame no longer has content size content bounds and quirks mode
leads to a body with zero height.
Fixes#30368.
* Take forced line breaks into account for intrinsic size
Fixes#30350.
* Don't linebreak on collapsible whitespace
This whitespace can hang off the end of the line, because it will be
trimmed LineItem layout.
* Update float placement after line breakage
Also don't ever line break for collapsible whitespace.
* Fix a few more test cases and clean up
* Renaming according to review comments
Refactor the scrolling and scrollable area calculation on the window
object, to make it better match the specification. This has some mild
changes to behavior, but in general things work the same as they did
before. This is mainly preparation for properly handling viewport
propagation of the `overflow` property but seems to fix a few issues as
well.
There is one new failure in Layout 2020 regarding `position: sticky`,
but this isn't a big deal because there is no support for `position:
sticky` in Layout 2020 yet.
Co-authored-by: Rakhi Sharma <atbrakhi@igalia.com>
This makes the names of flow relative geometry consistent with what is
used in the style crate and removes them from a module. With this change
it's more obvious what makes these types different from the ones in
`euclid`.
* servoshell: fix lockups while animating
* move comment to external_present declaration
* disable needs_recomposite optimisation for now due to breakage
* fix compile error that only happens on ci
* fix more compile errors
* Update mozjs
* moztools4 in bootstrap
* no autoconf
* tidy
* switch to servo-build-deps
* update mozjs for real
* glue mozjs
* fmt
* move to servo/mozjs
* Upgrade vendored version of WebRender
* Patch WebRender: upgrade version of gleam
* Restore hit testing implementation
* Fix WebRender warnings
* Adapt Servo to new WebRender
* Update results
* Add a workaround for #30313
This slightly expands text boundaries in order to take into account the
fact that layout isn't measuring glyph boundaries.
Before WebRender would ignore these, but newer version of WebRender have
issues with them. This change simply prevents legacy layout from
creating display items for these types of gradients. This is already the
behavior of non-legacy layout.
The border image outset support in WebRender is going to be removed and
even in versions of WebRender where it still exists, it fails to render
properly.
A border image is a type of border composed of slices of images. The
"outset" of this kind of border is a property in CSS that makes the
border boundaries expand. Previously, the value was passed to WebRender
which would expand the border by this amount and render the images into
the expanded rectangle.
Since this is going to be removed, we handle this property outside of
WebRender. The change is simply to expand the border area by the outset
before calculating the rest of the border values.
This is necessary for the WebRender upgrade.
* MallocSizeOf for Index{Set, Map}
* like as iterable in WebIDL
* Codegen magic for like interfaces
* TestBinding for like
* Test for Setlike and Maplike test bindings
* Some fixes
* Switch to any.js
* nit
* Keep order
* cleanup and move user input logix into servoshell
* fix fmt
* moves test from servoshell file
* move command-line args into servoshell
* remove feature media-gstreamer
* fix fmt
* move user input logic code into lib to make it more testable
* remove opts_matches in fn instead get it from main2
* remove pub and fix import
* add licence in new file
* revert passing Matches, instead pass Option String
* review update, also move sanitize fn to parser file
* fmt fix
* review fix: remove extra line
* constellation: crash to a new “sad tab” page
* check in resources/crash.html
* use a separate enum variant instead of keying on reason
* fmt + tidy
* rename Resource::Crash to Resource::CrashHTML
* clean up crash page and add details (reason + backtrace)
* avoid repeating crash errors in script::script_thread warn log
* make new LoadData init more idiomatic
* clarify comments and new fields
* fix doc comment style
This makes the nesting architecture of inline layout a lot simpler. Now
PartialInlineBoxFragment and InlineNestingLevelState are replaced by
two structs:
- InlineContainerState
- InlineBoxContainerState
InlineContainerState holds state for the root of the inline formatting
context, while InlineBoxContainerState holds state for inline boxes.
InlineBoxContainerState has an InlineContainerState as the first
element, thus "extends" it.
Now the inline iterators are stack variables directly in the `layout()`
method. They are no longer stored in the state. This avoids the weird
nesting of state and instead relies on a normal vector to hold the
stack of state.
Co-authored-by: Mukilan Thiyagarajan <mukilan@igalia.com>
PlacementAmongFloats would stop iterating when current_bands would be
empty, even if next_band wasn't at infinity.
Then the BFC root or replaced block was placed after all the floats,
even if it could fit next to some of them.
This patch moves the next_band into current_bands so that the loop
keeps considering bands.
This implements the rest of the bulk of float support. Now inline
element flow around floats and floats can be pushed down by inline
elements before them.
Co-authored-by: Oriol Brufau <obrufau@igalia.com>
PlacementAmongFloats was assuming that there would always be a FloatBand
at a position smaller than or equal to the given ceiling, but this was
not the case for negative ceilings.
The reason is that the FloatContext initialized the FloatBandTree with
a band at 0, and another at +∞.
This patch changes the initial bands to −∞ and +∞. This seems more
consistent and matches the expectation of PlacementAmongFloats.
`OnceCell` is now part of the standard library and we'll be able to use
it once we upgrade rust. For now we can use the version that's shipped
behind a feature flag in rust. This removes a dependency on one crate.
Now that our dependencies are using a newer version of rand we can
upgrade. This change also disables the WebRender debugger feature
because it has been removed in later versions of WebRender and allows
deduplicated the rand dependency as well as quite a few others.
When layout was split into two phases, floats were laid out as direct
children of the inline formatting context. This meant that they were
positioned properly, but not properly made children of their inline
ancestors' stacking contexts. This change maintains the proper
positioning of floats, but positions them relatively to their inline
ancestors.
The big change here is that `text-align` needs to be taken into account
before actually laying out LineItems. This has the added benefit of
setting inline layout for the implementation of `text-align: full`. Now
all line items are laid out at the real final position and we can adjust
the `start_corner` property of float `BoxFragments` when their ancestors
are laid out.
* Allow noidl files in script/dom/webidls
* Upgrade wgpu to 0.16 and refresh whole webgpu implementation
* Update WebGPU test expectations
* misc
* MutNullableDom -> DomRefCell<Option<Dom for GPUTexture
* Direct use of GPUTextureDescriptor
* Remove config from GPUCanvasContext
* misc
* finally blue color
* gpubuffer "handle" error
* GPU object have non-null label
* gpu limits and info
* use buffer_size
* fix warnings
* Cleanup
* device destroy
* fallback adapter
* mach update-webgpu write webgpu commit hash in file
* Mising deps in CI for webgpu tests
* Updated expectations
* Fixups
* early reject
* DomRefCell<Option<Dom -> MutNullableDom for GPUTexture
* Fix `mach test-wpt` to make crash tests work
There are two issues related to crash tests:
1. test-wpt is unable to find existing crash tests even when
called with --test-types=crashtests. The fix here is to
add crashtests to the default test suite types to python/wpt/run.py
2. When running in headless mode, crashes in style threads
don't cause servo to crash because the logic in constellation.rs
currently calls handle_panic only when the top-level browsing
context id is some value. Since style pool threads are shared,
they always generate Panic messages with None as top-level
browsing context id.
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
Signed-off-by: Mukilan Thiyagarajan <mukilan@igalia.com>
* Send bactrace to stderr and capture it in test runner
Servo's panic hook writes backtraces to stdout. This
patch changes it so they are written to stderr.
The crash test executor for servo in WPT grouping formatter
was also not capturing the output correctly for crashtests
as the log events were being aggregated based on thread name
which doesn't seem to match correctly in case of crashtests.
This patch also fixes the log grouping logic to be based on
test name.
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
Signed-off-by: Mukilan Thiyagarajan <mukilan@igalia.com>
* crashtests: update expectations for layout 2020
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
Signed-off-by: Mukilan Thiyagarajan <mukilan@igalia.com>
* crashtests: update expectations for layout 2013
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
Signed-off-by: Mukilan Thiyagarajan <mukilan@igalia.com>
* remove outdated & intemittent test expectations
Signed-off-by: Mukilan Thiyagarajan <mukilan@igalia.com>
---------
Signed-off-by: Mukilan Thiyagarajan <mukilan@igalia.com>
Co-authored-by: Martin Robinson <mrobinson@igalia.com>
The problem is a double-borrow of the `when_defined` member. The fix is
to let go of the borrow before resolving the promise which will
eventually try to borrow `when_defined` again.
Fixes#30120.
