There are some nuances to creating these wrappers, such as manually
propagating certain text styles that are not inherited by default. We
already have the logic for this in
`NodeWithStyle::create_anonymous_wrapper()`, so reuse that method in our
implementation of the button layout.
Fixes applying certain text styles (such as `text-decoration`) to the
text of a `<button>`.
...by another GFC. When searching for grid item that contains an
abspos box positioned relative to GFC, it's incorrect to assume that the
closest ancestor box whose parent establishes GFC is always the one we
are looking for, because there may be non-positioned GFC in between.
Fixes regression introduced in 80c8e78.
This enabled WebGL on Linux. It uses ANGLE's OpenGL backend running atop
EGL_PLATFORM_SURFACELESS_MESA. Eventually we should probably switch to
the Vulkan backend but that doesn't seem to be enabled in the vcpkg
angle package. Anyway, switching later should be trivial.
The painting surface is allocated through Vulkan and then imported into
EGL as a dma-buf. The DRM format modifier mechanism, along with Vulkan
initializing the image with VK_IMAGE_LAYOUT_GENERAL, should ensure
surface compatibility across the two APIs.
For now, we will synchronize rendering and presentation using glFinish,
although this is admittedly suboptimal. Really we should grab an
EGLSync, export that to an fd, import it into Skia and have it wait for
it before reading from the image. That can be implemented in a future
change, though.
If a WebGL canvas is resized through the set_size function, we will
re-create the painting surface. However, this currently leaks all of the
associated EGL/OpenGL objects. This change introduces the
free_surface_resources function which will free all resources associated
with the painting surface. It will be called before allocating a new
surface and during context destruction. It keeps track of the OpenGL
texture for the color buffer in m_impl instead of just storing it on the
stack.
This adds a new PaintingSurface creation function, create_from_vkimage,
which returns a PaintingSurface backed by a vulkan image. It's analogous
to the existing create_from_iosurface function. In both cases the
backing object will be imported into Skia as a render target and then an
SkSurface will be wrapped around that.
In order to ensure that the image will not be freed while still in use
by Skia, we will manually bump the refcount of the VulkanImage object
before passing it to Skia and then use the releaseCallback parameter of
WrapBackendRenderTarget to register a callback that drops this
reference.
This introduces the ability to allocate Vulkan images which can be
shared across graphics APIs or across processes using the Linux dma-buf
interface.
The create_shareable_vulkan_image function takes a VulkanContext, image
width, height, and format, and an array of DRM format modifiers
representing image memory layouts accepted by the caller. The function
will intersect this list with the list of modifiers supported by the
Vulkan implementation, ensuring the resulting image uses one of those
layouts (exactly which one is up to the implementation). The function
will return a VulkanImage object, which is reference-counted and
encapsulates the VkImage itself as well as the backing memory
allocation. It also stores various image parameters such as usage,
tiling, etc.
The VulkanImage::get_dma_buf_fd function will create a file descriptor
representing the image's backing dma-buf which can then be imported by a
different API or sent over a unix domain socket.
When serializing the "style" attribute, we were incorrectly assuming
that some of the grid-related CSS properties would never contain var()
substitution functions.
With this fixed, we can now book appointments on https://cal.com/ :^)
This changes the maximum number of decimal places from 5 to 6, but 5 was
previously a guess, and not specified behaviour:
> For all of the decimal changes (except color) I couldn't really find a
> spec that mandates any required precision, so I just copied what
> Firefox seems to do, which is limit the output to 5 decimal places.
> https://github.com/SerenityOS/serenity/pull/23449
If `will-change` is set to a property value where that property could
create a stacking context, then we create a stacking context regardless
of the current value of that property.
This property provides a hint to the rendering engine about properties
that are likely to change in the near future, allowing for early
optimizations to be applied.
Previously, it was possible for an up/down arrow press to place the
cursor in the middle of a multi-code point grapheme cluster. We want to
prevent this in a way that matches the behavior of other browsers.
Both Chrome and Firefox will map the starting position to a visually
equivalent position in the target line with harfbuzz and ICU segmenters.
The need for this is explained in a code comment. The result is a much
more natural feeling of text navigation.
When starting transitions we compute the after-change style, for any
inherited properties this should include the non-animated value.
Previously we were only inheriting the animated value and treating it as
non-animated so were instead including the animated value.
This commit fixes that by inheriting both the animated and non-animated
values (with the former being stored in `m_animated_property_values`,
and the latter in `m_property_values`).
This gains us 12 new WPT passes.
This brings with it 252 new WPT fails from the various 'events' tests in
css/css-transitions/properties-value-inherit-001.html, however these
also fail in other browsers (Chrome, Edge and Firefox) and the behaviour
that causes these failures is specifically mentioned in the spec.
This commit updates the CSSTransition constructor to:
- Leave the KeyframeEffect start time unresolved
- Set the KeyframeEffect start delay
Gains us 14 WPT passes but exposes one false positive in
properties-value-inherit-001.html
For button layouts, we were overriding the computed `width` value with
`fit-content` in `TreeBuilder::wrap_in_button_layout_if_needed()`. But
the spec asks us to set the _used value_ instead, so we now actually
calculate the fit-content width and set the box' content width to it.
Fixes#2516.
This suits the spec a bit better, and exposes the fact that we were
allowing `::ImageButton` to use the button layout although it is never
specified that it should do so. Tests were rebaselined for this.
We should calculate whether we need to truncate text with an ellipsis
exclusive of any trailing collapsible whitespace.
This was causing issues where an inline element was automatically sized
(e.g. min-content) as we would calculate available width exclusive of
trailing collapsible whitespace and then our text chunk would be wider,
always inserting an ellipsis.
Fixes the visual element of #4048 but we still are incorrect in how we
collapse whitespace more generally
As of 7dc8062 paintables compute and cache their visibility (which
depends on opacity) at construction - this cached value can fall out of
sync with reality if if the opacity changes to/from zero within the
lifetime of that paintable.
This commit invalidates layout when an opacity changes to/from zero so
that we reconstruct paintables with the correct visibility.
The limitations right now are:
- We don't know if a property is a list or not.
- We always reify as a CSSStyleValue directly.
So, we pass some tests but only ones that expect a CSSStyleValue.
When no better reification form is available, we produce an opaque
CSSStyleValue with a serialized value. For starters, this will be the
only way to reify, and then we'll add others later.
After looking into this more, the `[[declarations]]` slot does not seem
to need to be a literal map of property names and values. Instead, it
can just point at the source (an element or style declaration), and
then direct all read or write operations to that.
This means the `has()` and `delete()` methods actually work now.
A few remaining failures in these tests are because of:
- StylePropertyMap[ReadOnly]s not being iterable yet
- We're not populating an element's custom properties map, which get
fixed whenever someone gets around to producing proper computed
values of them.
This isn't part of the CSSOM API directly, but will be used by
StylePropertyMapReadOnly.has() to avoid doing the work to serialize a
string version of the property's value, just to throw it away again.
We occasionally (frequently) time out in CI. If ctest triggers this
timeout, it sends a SIGSTOP followed by a SIGKILL. Since we want to
gracefully exit the test runner, ask ctest to send a SIGTERM instead.
This should cause active test status reports to show up in CI.
If you interrupt the test runner (Ctrl+C, SIGINT) or if the test runner
is gracefully being terminated (SIGTERM), it now reports the current
status of all the spawned views with their URL and, if an active test is
still being run, the time since the start of the test.
Hopefully this will help gain insight into which tests are hanging.
The local handling of some messages might cause the transport to get
closed. If that's the case, we shouldn't try to send back a response.
This fixes many of the "Trying to post_message during IPC shutdown"
errors I was seeing when terminating Ladybird or when abnormally exiting
from LibWeb tests.
