...by constructing ImmutableBitmap directly from SkImage.
This is a huge optimization for the case when content of canvas is
painted onto another canvas, as it allows pixels to remain in GPU memory
throughout the process.
Fixes performance regression on https://playbiolab.com/ introduced by
switching to GPU-backend for canvas.
By caching the SkImage that is reused across repaints, we allow Skia t
optimize GPU texture caching.
ImmutableBitmap is chosen to own the SkImage because it guarantees that
the underlying pixels cannot be modified. This is not the case for
Gfx::Bitmap, where invalidating the SkImage would be challenging since
it exposes pointers to underlying data through methods like scanline().
Gap values are now represented by Variant<LengthPercentage, NormalGap>.
NormalGap is just an empty struct to represent the `normal` keyword.
This fixes a long-standing issue where we were incorrectly storing gaps
as CSS::Size, which led to us allowing a bunch of invalid gap values.
This method was being used to check for invalid `PropertyKey`s.
Since invalid `PropertyKey`s are no longer created, and since the
associated method has also been removed in the latest edition of
ECMA-262, the method can now be removed here as well.
While we are removing all its calls, lets also update any surrounding
spec comments to the current latest edition, where possible.
This object property kind had completely different behaviour.
By adding an instruction for it, we can remove a bunch of special
casing, and avoid creating dummy `PropertyKey` values.
This constructor was creating an "invalid" `PropertyKey`, but every
code path constructing such a `PropertyKey` was either never used or
immediately `VERIFY`ing that the `PropertyKey` was valid.
The `VERIFY` call has been moved to the `PropertyKey::from_value(...)`
call, and the array/object literal spreading code could be refactored
to not take a `PropertyKey` but creates dummy values for now.
The default constructor for `Reference` has similairly be deleted,
because it was never used.
This also adds support for `xyz` as it defaults to `xyz-d65`. We now
pass the following WPT tests:
- css/css-color/xyz-001.html
- css/css-color/xyz-002.html
- css/css-color/xyz-004.html
- css/css-color/xyz-d65-001.html
- css/css-color/xyz-d65-002.html
- css/css-color/xyz-d65-004.html
This will be usefully on its own later on when supporting this color
space directly, but it also allows us to do some factorization in the
current codebase.
If a & simple selector is on a style rule with no parent style rule,
then it behaves like :scope - but notably, :scope provides 1
specificity in the class category, but & is supposed to provide 0.
To solve this, we stop replacing it directly, and just handle the & like
any other simple selector. We know that if the selector engine ever
sees one, it's equivalent to :scope, because the nested ones will have
been replaced with :is() before that point.
This gets us one more subtest pass. :^)
When we first parse a nested CSSStyleRule's selectors, we treat them as
relative selectors and then patch them up with an `&` as needed.
However, we weren't doing this when assigning the `cssText` attribute.
So, let's do that!
This gives us a couple of subtest passes. :^)
We were delaying sending an IPC message until the HTML PortMessage
task was run, but we were not queuing the task in on the receiver
side. The result of this was that message port posting was
needlessly creating an HTML task just to send an IPC message.
On the flip side, we were directly calling dispatch_event from the
socket notifier of the target port's message queue. This is a huge
problem, because it means that we were effectively running
javascript-aware code from an 'in parallel' context.
By switching around which side of the IPC interface is responsible
for queuing a task, we can avoid problems where a document is
destroyed from a port message-attached callback and crashes.
The Gfx::Bitmap encoder and decoder have been replaced
by BitmapSequence in #1435, making them redundant and safe to remove.
Additionally, the base IPC::encode and IPC::decode implementations
will trigger a compiler error if these methods are called.
.ipc files already exclude references to Gfx::Bitmap.
This change fixes accessible-name computation for:
- an element that has empty text content but that also has a title
attribute (“tooltip”) with a non-empty value
- an img element whose alt attribute is the empty string but that also
has a “title” attribute with a non-empty value
Otherwise, without this change, the accessible name unexpectedly isn’t
computed correctly for those cases.
As shown in the test added by this patch, it was possible to re-assign
the `this` value of a member function call while it was executing.
Let's copy the original this value like we already do with the callee.
Fixes#2226.
