Previously, we were collapsing whitespace in Layout::TextNode and then
passed the resulting string for further processing through ChunkIterator
-> InlineLevelIterator -> InlineFormattingContext -> LineBuilder ->
LineBoxFragment -> PaintableFragment. Our painting tree is where we deal
with things like range offsets into the underlying text nodes, but since
we modified the original string, the offsets were wrong.
This changes the way we generate fragments:
* Layout::TextNode no longer collapses whitespace as part of its
stored "text for rendering", but moves this logic to ChunkIterator
which splits up this text into separate views whenever whitespace
needs to be collapsed.
* Layout::LineBox now only extends the last fragment if its end offset
is equal to the new fragment's start offset. Otherwise, there's a
gap caused by collapsing whitespace and we need to generate a
separate fragment for that in order to have a correct start offset.
Some tests need new baselines because of the fixed start offsets.
Fixes#566.
This updates `background-blend-mode`, `mix-blend-mode`, and `isolation`
to be animation type `discrete` rather than `none`. This reflects
169b0e1 but isn't reflected
on the published spec as the published spec is out of date - see
https://github.com/w3c/fxtf-drafts/issues/547
Gains us 26 WPT tests.
This also fixes a bug in the view transitions code that was
required to get the imported test to pass. The code for setting
the initial containing block size just did not set the right thing,
since doing so would trigger an error later on.
That later error resulted from walking up the tree, without
considering that the document element has a parent that is not
itself an element. (and then doing element things to it)
Now follows the same pattern as PaintableBox and StackingContext, where
it exits if hidden, then hit tests children, then hit tests itself if
it's `visible_for_hit_testing()`.
This ensures that we clamp values for properties like padding-* to valid
ranges (non-negative in this case) if they are specified with `calc()`.
The length-related changes in this commit combined with the ones from
the previous commit fix the primary layout issue on https://lwn.net
(yes, not the first place I would have expected problems either).
If a node with `contenteditable=true/plaintextonly` is the child of an
editable node or an editing host, we should make it editable instead of
an editing host. This effectively merges nested editing hosts together,
which is how other browsers deal with this as well.
Gains us 5 WPT subtest passes in `editing`.
`font-size` can end up with a negative value - either due to `calc`
being resolved using the old method which doesn't clamp the value, or
interpolation - in this case we should clamp negative values to zero.
Gains us 36 new WPT passes and fixes crashes in the three imported
tests.
Interpolation can leave `padding-*` values as negative - this should be
handled by interpolation clamping it to the allowed range of values
but we don't yet do that. As a stop gap we can clamp this before setting
it in ComputedValues.
This fixes 3 crashes and gains us 11 passes in the imported WPT tests
From the SVG spec
The value of the ‘viewBox’ attribute is a list of four numbers <min-x>,
<min-y>, <width> and <height>, separated by whitespace and/or a comma...
Currently try_parse_view_box will fail to parse the attribute if the
values are separated by commas.
This change replaces try_parse_view_box with a more correct
implementation. It will reside in the AttributeParser.cpp. This new
implementation correctly handles comma-separated viewBox values, and is
also more robust against invalid inputs.
Additionally, it adds a new test case to ensure viewBox values with
various syntax are parsed correctly and invalid values are rejected.
This fixes slow test execution on macOS where localhost resolution
has a 200ms delay due to IPv6 fallback behavior in libcurl.
The XMLHttpRequest-override-mimetype-blob.html test now runs in ~0.6s
instead of ~16s on macOS.
Fixes#4850
See also: https://github.com/curl/curl/issues/2281
If a node that establishes a StackingContext has `pointer-events: none`,
hit testing should first proceed with hit testing the SC's children
before deciding to bail. We were checking for `pointer-events` too
early, causing large parts of certain websites to be noninteractive.
Fixes#6017.
One MessagePort can be entangled with another MessagePort, either in the
same agent, or in another agent.
In the same-agent case, the MessagePort objects point to each other via
the MessagePort::m_remote_port field.
In the separate-agent case, they live in separate processes entirely and
thus can't point at each other.
In both cases, the MessagePorts have an underlying transport channel,
which means they are "entangled". However, we can't assume that being
entangled means having a non-null m_remote_port.
This patch simply adds a missing null check for m_remote_port and thus
makes https://vscode.dev/ stop crashing with a null dereference.
We originally had special handling for `:host()` as that had been the
only pseudo-class that could be both an identifier or a function.
However, this meant duplicating the serialization logic, and also we
had to manually remember to add the same hack for any other
identifier-and-function cases. Which I forgot to do with `:heading()`!
So instead, for these cases, detect if they actually have arguments
specified and use that to determine which form to serialize as. We do
still have to write a check for each one of these pseudo-classes, but
the VERIFY should make it easier to remember.
We now also store `outline-width` in ComputedValues as a `CSSPixels`
since we know it's an absolute length at `apply_style` time - this saves
us some work in converting to CSSPixels during layout.
Gains us 46 new passes since we now interpolate keywords (thick, thin,
etc) correctly.
Also loses us 4 WPT tests as we longer clamp negative values produced by
interpolation from the point of view of getComputedStyle (although the
'used' value is still clamped).
An SVGLength can be read-only, e.g. all animVal values cannot be
modified. Implement this for all instantiations of SVGLength.
While we're here, add `fake_animated_length_fixme()` so we can easily
find all sites where we need to improve our animated length game.
The transform of each paintable was being applied multiple times due to
the recursive nature of the hit testing methods. Previously it used
combined_css_transform to transform the position, and then it would pass
that position to children, which would then apply combined_css_transform
again, and so on.
PaintableBoxes are also not hit tested anymore when having a stacking
context. A similar check is done in PaintableWithLines, but it was
missing from PaintableBox. Without this check some elements can get
returned multiple times from a hit test.
StackingContexts with zero opacity will now also get hit tested, as it
should have been before.
Atlassian login gets the base URL for its module scripts by throwing an
error and pulling out the current script's URL from error.stack with
regex.
Since we only returned a basename for module scripts, it would fail to
match and try and use `/` as a base URL (because it does
[matched_string] + "/"), which is not a valid base URL.
Update a couple of focus-related spec steps and their implementations.
The most relevant change is that we no longer allow focusing on elements
that return false for `->is_focusable()`, which necessitates fixing a
broken test that tried to `.focus()` on `<div>`s that were not
focusable. That test's output now more accurately reflects the expected
outcome as seen in other browsers.
