The constructor is now only concerned with creating the required
streams, which means that it no longer fails for XZ streams with
invalid headers. Instead, everything is parsed and validated during the
first read, preparing us for files with multiple streams.
This was causing a huge slowdown when loading some pages with weirdly
huge number of style sheets. For example, amazon.com has over 200 style
elements, which meant we had to resort the StyleSheetList 200 times.
(And sorting itself was slow because it has to compare DOM positions.)
Instead of sorting, we now look for the correct insertion point when
adding new style sheets, and we start the search from the end, which is
where style sheets are typically added in the vast majority of cases.
This removes a 600ms time sink when loading Amazon on my machine! :^)
Setting the `data` of a text node already triggers `children changed`
per spec, so there's no need for an explicit call.
This avoids parsing every HTMLStyleElement sheet twice. :^)
We now select between nearest neighbor and bilinear filtering when
scaling images in CRC2D.drawImage().
This patch also adds CRC2D.imageSmoothingQuality but it's ignored for
now as we don't have a bunch of different quality levels to map it to.
Work towards #17993 (Ruffle Flash Player)
We currently decode back-references one byte at a time, while writing
that byte back out to the output buffer. This is only necessary when the
back-reference refers to itself, i.e. when the back-reference distance
is less than its length. In other cases, we can read the entire back-
reference block in one shot.
Using the "enwik8" file as a test (100MB uncompressed, commonly used in
benchmarks: https://www.mattmahoney.net/dc/enwik8.zip), decompression
time decreases from:
5.8s to 4.89s on Serenity (cold)
2.3s to 1.72s on Serenity (warm)
1.6s to 1.06s on Linux
Huffman codes have a useful property in that they are prefix codes. That
is, a set of bits representing a Huffman-coded symbol is never a prefix
of another symbol. This allows us to create a table, where each index in
the table are integers whose prefix is the entry's corresponding Huffman
code.
With Deflate, we can have codes up to 16 bits in length, thus creating a
prefix table with 2^16 entries. So instead of creating a table fit all
possible codes, we use a cutoff of 8-bit codes. Codes larger than 8 bits
fall back to the binary search method.
Using the "enwik8" file as a test (100MB uncompressed, commonly used in
benchmarks: https://www.mattmahoney.net/dc/enwik8.zip), decompression
time decreases from 3.527s to 2.585s on Linux.
We currently mix normal and bit streams during GZIP decompression, where
the latter is a wrapper around the former. This isn't causing issues now
as the underlying bit stream buffer is a byte, so the normal stream can
pick up where the bit stream left off.
In order to increase the size of that buffer though, the normal stream
will not be able to assume it can resume reading after the bit stream.
The buffer can easily contain more bits than it was meant to read, so
when the normal stream resumes, there may be N bits leftover in the bit
stream that the normal stream was meant to read.
To avoid weird behavior when mixing streams, this changes the GZIP
decompressor to always read from a bit stream.
Previously we used a parsing context with no access to the document, so
any URLs in url() functions would become invalid.
Fixes the images on Steam's store carousel, which sets
Element.style.backgroundImage to url() functions.
Some fonts (like the Bootstrap Icons webfont) have bogus glyph offsets
in the `loca` table that point past the end of the `glyf` table.
AFAICT other rasterizers simply ignore these glyphs and treat them as if
they were missing. So let's do the same.
This makes https://changelog.serenityos.org/ actually work! :^)
In situations where we need a width to calculate the intrinsic height of
a flex item, we use the fit-content width as a stand-in. However, we
also need to clamp it to any min-width and max-width properties present.
The spec tells us that for glyph offsets in the "loca" table, if an
offset appears twice in a row, the index of the first one refers to a
glyph without an outline (such as the space character). We didn't check
for this, which would cause us to render a glyph outline where there
should have been nothing.
Some versions of clang, such as Apple clang-1400.0.29.202 error out on
the previous out of line operators. Explicitly defaulting comparison
operators out of line is allowed per P2085R0, but was checked in clang
before version 15 in C++20 mode.
When debugging in Xcode, the waitpid() for the initial forked process
would always return EINTR or ECHILD. Work around this by blocking all
signals until we're ready to wait for the initial child.
In `flex-direction: column` layouts, a flex item's intrinsic height may
depend on its width, but the width is calculated *after* the intrinsic
height is required.
Unfortunately, the specification doesn't tell us exactly what to do here
(missing inputs to intrinsic sizing is a common problem) so we take the
solution that flexbox applies in 9.2.3.C and apply it to all intrinsic
height calculations within FlexFormattingContext: if the used width of
an item is not yet known when its intrinsic height is requested, we
substitute the fit-content width instead.
Note that while this is technically ad-hoc, it's basically extrapolating
the spec's suggestion in one specific case and using it in all cases.
When calculating the intrinsic width of a box, we now make its content
width & height indefinite before entering the intrinsic sizing layout.
This ensures that any geometry assigned to the box by its parent
formatting context is ignored.
For intrinsic heights, we only make the content height indefinite.
This is because used content width is a valid (but optional) input
to intrinsic height calculation.
When scrolling the page, we may need to flush any pending layout
changes. This is required because otherwise, we don't know if the target
scroll position is valid.
However, we don't need to *force* a layout. If the layout tree is
already up-to-date, we can use it as-is.
Also, if we're scrolling to (0, 0), we don't need to update the layout
at all, since (0, 0) is always a guaranteed valid scroll position.
Currently significant portion of requests coming to WebDriver server
fails with error while parsing json body because requests are parsed
when they are not complete yet.
This change solves this by waiting for more data to arrive if HTTP
request parser found that there is not enough data yet to parse
the whole request.
In the future we would probably want to move this logic to LibHTTP
because this problem is relevant for any HTTP server.
Fixes the issue that XML parser fails when loader passes input that is
prefixed with byte order mark.
Also it generally makes sense to pass text source through encoding
decoder before parsing. Probably we would even want to introduce method
similar to `create_with_uncertain_encoding` in `HTMLParser` but for
`XMLParser` to be make harder unconsciously pass non-UTF8 input to XML
parser.
The "flex item automatic minimum size in the main axis is the
content-based minimum size" behavior should only apply to flex item
sizes in the main axis. There was one case where we incorrectly applied
this behavior in the cross axis
The "flex item automatic minimum size in the main axis is the
content-based minimum size" behavior should only apply to flex items
that aren't scroll containers. We were doing it for all flex items.
When calculating one of the intrinsic sizes for a flex container, we
already go through the flex layout algorithm.
There's no need to perform some of the algorithm steps a second time.
This is a relic from an earlier time when we tried to bail early from
the layout algorithm in the intrinsic sizing case. Now that we go
through the whole thing anyway, this is much simpler. :^)
This fixes a few issues I noticed when playing around with radio
buttons. Previously radio buttons would uncheck checkboxes with
the same "name" attribute, uncheck inputs across different forms,
and treated no name attribute as a group.
This now implements the radio button group check from the HTML spec.
