Display lists owned the accumulated visual context tree through a
ref-counted pointer. That tied visual-context state to display-list
lifetime and made compositor updates treat the two as one unit, even
though AVC trees need to become independently updateable compositor
state.
Make accumulated visual context trees plain versioned values, have each
display list store the compatible tree version, and pass the matching
tree alongside display-list updates and replay calls. Replay verifies
that the provided tree matches the display list before executing it.
This prepares the compositor for receiving AVC tree updates separately
from display-list updates: it now accepts the tree as a separate update
parameter, stores it next to the display list, and uses that stored tree
for replay and async-scroll hit testing. Nested display-list resources
carry their own tree snapshots for the same version check.
DisplayListPlayer::execute() used to flush the active painting surface
as part of replay. That made replay and submission inseparable, so
callers could not add extra painting after replay without either
accepting a stale submission boundary or flushing again.
Expose flush(PaintingSurface&) on the player and make execute() only
replay commands. Existing callers now issue an explicit flush at the
same point where the implicit flush used to happen, keeping behavior
unchanged while making the submission boundary visible to compositor
code.
Use StyleValue's computational-independence predicate when deciding
which specified values to retain for inherited-style recomputation.
Compound values such as shadows, filters, lists, and calc trees now
reuse their existing dependency tracking instead of the old bare-length
check.
Treat missing tuple entries and keyword-only edge offsets as
computationally independent. These nullable slots are part of the
parsed value representation, and the inherited dependency tracking now
uses the shared computational-independence query for specified values.
Add viewport resize coverage for media query changes that update root
font metrics while a descendant uses font-relative lengths inside
box-shadow and filter values.
Parse subgrid track listings for grid-template rows and columns,
including fixed and auto-fill name-repeat line-name lists. Preserve
the subgrid keyword through computed values and shorthand serialization,
and make interpolation discrete instead of routing subgrid lists through
the explicit track interpolation path.
Import WPT coverage for subgrid grid-template parsing and computed
values, and add text coverage for discrete subgrid track-list animation.
Record when computed properties depend on viewport metrics while
resolving lengths. Carry that information through font metrics so
font-relative lengths can be associated with viewport-sized fonts.
This keeps the dependency tracking local to style computation and gives
later viewport resize invalidation a way to find affected elements.
ImageStyleValue stored GC-managed request, stylesheet, and animation
timer references as strong GC::Ptr fields even though image style values
are refcounted objects. When such a style value outlives the GC object
that normally visits it, those fields can keep stale pointers after GC
collects the referents. On Steam this allowed a stale image resource
request to be read as unrelated image data, making carousel SVG arrows
render at the wrong size.
Store these back references as GC::Weak instead. Reachable style values
still use live requests, stylesheets, and timers normally, but detached
values observe null after the GC collects the referent and can reload or
skip the now-dead association instead of dereferencing reused GC memory.
Keep a local timer handle while installing the timeout callback so setup
does not rely on the weak member.
With the image style values no longer hiding strong GC edges, remove the
obsolete IGNORE_GC annotation from CSSStyleSheet's pending image list.
Represent WebIDL C++ types with a single CppType model that tracks
nullability, optional presence, and contained storage.
GC-like values now use GC::Ref/GC::Ptr directly, while containers choose
"plain", "Root", or "Conservative" container types depending on what
they contain. For example, sequence<Element> becomes a RootVector of
GC::Ref values, while sequence<SomeDictionary> becomes a
ConservativeVector only when the dictionary contains GC-like values.
This moves the generated bindings away from wrapping GC values in
GC::Root by default.
This has broad fallout as the types passed to interfaces for GC
objects changes almost fully across the board.
Keep animated ImageStyleValue frame advancement owned by the
style value. The current frame and loop state live there, so a
separate document scheduler would duplicate ownership of that state.
Start the ImageStyleValue timer only while it has layout clients.
Stop it when the last client unregisters, or when a finite animation
completes. Expose a document-scoped active timer count through
internals for focused regression tests.
Clear image observers when layout nodes detach. Use current-node
cleanup for per-DOM-node clearing, and explicit subtree cleanup for
tree replacement, full tree teardown, and synthetic pseudo-elements.
This keeps large document clearing linear.
Unregister generated-content image providers during layout detach
instead of waiting for GC to finalize the provider.
Cover hidden animated background images, generated content images,
layout node replacement, full layout tree teardown, and document
scoping for the internals counter.
Parameter values are absolutized at style computation time (or are
confirmed to be resolvable in the case of reification or DOMMatrix) so
there is no reason this function should fail.
All relative lengths will have been absolutized at style computation
time (or disallowed at parse time in the case of DOMMatrix) so we don't
need to resolve them again here.
LibWeb still exposed the concrete CompositorThread to Page,
Navigable, and EventHandler, so compositor IPC would have leaked the
thread implementation into callers. The old thread APIs also bundled
page presentation callbacks and main-thread wakeups into the same
object, which made it awkward for WebContent to put an actor boundary
in between.
Introduce CompositorHost and context handles as the caller-facing API,
and move shared compositor protocol values out of CompositorThread. Add
WebContentCompositor IPC endpoints and route PageHost through a paired
in-process transport. The actor owns CompositorThread with explicit
main-thread and UI presentation clients, while screenshot completion is
serialized on the WebContent event loop using request IDs.
The intention for introducing IPC here is to prepare for moving the
compositor thread into a separate process.
Similar to GC::Root<T>, make GC::RootVector<T> constructible without
explicitly passing a Heap.
This is implemented by having RootVectorBase use GC::Heap::the() for
heap-free construction.
The resolution is always absolutized at style computation time so we can
just use `Resolution::from_style_value` and select the image once in
`load_any_resources` since the layout node won't change the result.
Display lists used to own the resource storage needed to replay their
command bytes. That kept the compositor tied to in-process object
ownership: sending a display list update also meant sharing the same
resource container with the recording side.
Move resource storage out of DisplayList and make display list updates
carry a transaction of resources to add and remove. Navigable now tracks
the resources referenced by the current display list, sends only the
delta to the compositor, and trims its recording-side storage to the
active set. The compositor applies those transactions to its own storage
before replacing the cached display list.
This still carries in-process resource objects, but it puts the
ownership boundary in the right place. Command bytes and resource
lifetime are now synchronized explicitly, which is the shape needed
before the compositor can receive serializable resource updates across a
process boundary.
CSS Animations Level 2 makes animation-duration default to auto. We
still initialized it to 0s and converted auto to zero while collecting
CSS animation properties, before the animation effect had its scroll
timeline. Scroll-driven animations with an omitted duration therefore
behaved like zero-duration animations, so scroll(root) progress bars did
not track the scroll position.
Make animation-duration initialize to auto and carry that value into the
effect timing normalization. Time-driven animations still resolve auto
to 0s there, while progress-based timelines resolve it against the
timeline duration. Keep treating that resolved 0s as the default when
serializing the animation shorthand so computed style remains compatible
with existing CSS Animations expectations.
Add a text regression covering an omitted duration on a scroll(root)
progress animation. The test fails before this fix because the computed
duration is 0s and the transform never advances at 25% scroll.
This will be used in generated code to represent missing optionals i.e.
`<foo> <bar>?` with only a `<foo>` value will be represented by a
`StyleValueList` containing the relevant `<foo>` style value and an
`EmptyOptionalStyleValue`
Store overflow-clip-margin as a structured style value instead of
collapsing it into generic keyword/length lists during parsing.
Keep omitted visual-box as a distinct state so painting can apply
the spec default per element type (content-box for replaced,
padding-box otherwise), while preserving canonical serialization.
Delayed preload and image animation callbacks can outlive the objects
they notify. Incremental sweeping makes this easier to hit because stale
callback state may be reclaimed before the delayed work runs.
Use a weak link element when firing preload load and error events, and
use a weak image style value from animated image timers instead of a raw
pointer.
DecodedImageFrame only wraps a ref-counted Bitmap and color-space
metadata. The frame object itself does not provide shared mutable
state or lifetime ownership beyond those members, so ref-counting it
adds an unnecessary layer of indirection.
Previously we stored these within the `URLStyleValue` which didn't
itself have an `absolutized` method so wouldn't absolutize the fallback
color. We now store the two values alongside each other in a
`StyleValueList` which correctly handles absolutization
DecodedImageFrame now owns decoded bitmap pixels directly, so the
separate ImmutableBitmap wrapper no longer carries useful semantics.
Remove the class and pass decoded image frames or bitmaps at the
boundaries where pixels are actually required.
The Skia image cache now keys off DecodedImageFrame, matching the
display-list commands that paint decoded images. Video frames stay
owned by LibMedia, with the explicit YUV-to-bitmap conversion living
at HTMLVideoElement's decoded-frame entry point for canvas and WebGL
callers.
Decoded image data should not continue to traffic in ImmutableBitmap now
that the bitmap wrapper is being retired. Introduce DecodedImageFrame as
the paintable decoded-image unit and store a Bitmap plus ColorSpace in
it directly.
Thread the new frame type through decoded image data, display-list
image commands, filters, canvas drawImage, patterns, WebGL texture
upload, and CSS/SVG image consumers. ImmutableBitmap remains only at
the legacy boundaries that still need it, such as HTML video snapshots
and callers that explicitly ask for a bitmap snapshot.
This keeps color-space ownership with the decoded frame while making
the expensive or legacy ImmutableBitmap path explicit at the few call
sites that still need it.
Previously, `Document::notify_css_background_image_loaded()` walked the
entire `PaintableBox` subtree and cleared each box's paintable cache
whenever any CSS image finished loading.
Replace this with per-image observers owned by the layout node. During
`apply_style`, each node registers as an `ImageStyleValue::Client` for
the images its style references. On load, only the affected layout
node's paintables are invalidated.
ImageSetStyleValue::set_style_sheet() previously stored the style sheet
on itself but didn't propagate to its candidate images. As a result,
candidates were never registered as pending image resources, so their
fetches didn't start until layout time and didn't delay the document's
load event.
This caused css-image-set-background-type.html to be flaky: under
load, the screenshot could be captured before the selected SVG
candidate had finished decoding, producing an empty box instead of
the expected color.
Propagate set_style_sheet() to each candidate image whose type()
filter does not exclude it, mirroring StyleValueList and
ShorthandStyleValue. The candidates now register themselves as
pending so SharedResourceRequest's load event delayer correctly
delays the load event until decoding completes.
Add an abstract image style value for image-set() and parse both the
standard and -webkit-prefixed spellings through the existing <image>
value path. The parser accepts URL and string image candidates,
optional resolution descriptors, and type() filters.
Track attr-taint through substituted component values so image-set()
candidates using attr()-derived URL-producing tokens are rejected when
resolved for URL-using properties.
Update the relevant WPT baselines now that image-set() parsing is
supported in additional value contexts.
Currently there are multiple style values which are essentially the same
thing, a function holding a value, just with different names. This
commit adds a generic style value to replace them with and the following
commits will do so.
In a future commit we will use the allowed literal values as the bounds
for calculated values rather than the existing `ValueParsingContext`
based system. Since the computed bounds are different from the allowed
literal values here we need to handle clamping them manually.
This allows us to avoid the ugly hack in
`property_accepted_type_ranges()`.
This also updates the `ValueType` to be `opacity-value` rather than
`opacity` to match the spec.
