This isn't actually part of CSS-FLEXBOX-1, but all major engines honor
these properties in flex layout, and it's widely used on the web.
There's a bug open against the flexbox spec where fantasai says the
algorithm will be updated in CSS-FLEXBOX-2:
https://github.com/w3c/csswg-drafts/issues/2336
I've added comments to all the places where we adjust calculations for
gaps with "CSS-FLEXBOX-2" so we can find them easily. When that spec
becomes available, we can add proper spec links.
Extending the borders on reference frames so that motion vectors that
point outside the reference frame allows `predict_inter_block()` to
avoid some branches to clamp the sample coordinates in its loops.
This results in about a 25% improvement in decode time of a motion-
heavy YouTube video (~20.8s -> ~15.6s).
Moving the clamping of the coordinates of the reference frame samples
as well as some bounds checks outside of the loop reduces the branches
needed in the `predict_inter_block()` significantly.
This results in a whopping ~41% improvement in decode performance
of an inter-prediction-heavy YouTube video (~35.4s -> ~20.8s).
Changing the calculation of reference frame scale factors to be done on
a per-frame basis reduces the amount of work done in
`predict_inter_block()`, which is a big hotspot in most videos.
This reduces decode times in a test video from YouTube by about 5%
(~37.2s -> ~35.4s).
This changes the order of the loop copying data to a reference frame
store so that it copies each row in a contiguous line rather than
copying a column at a time, which caused unnecessary branches.
This reduces the decode time on a fairly long 720p YouTube video by
about 14.5% (~43.5s to ~37.2s).
This prevents the seek slider in VideoPlayer from skipping forward and
then back when fast seeking is enabled. Previously, it was possible for
a single frame to render before the actual seek position in the seek
bar was set.
Previously, the time would read "00:00:01" when the timestamp was
merely 1 millisecond past the start of the video. If a video does not
start with a sample at timestamp 0, then, seeking to the start would
display that text rather than "00:00:00".
The state could change and cause a timestamp change without the video
frame event firing, which could desync the seek bar from the video's
actual time.
Previously, there was some leftover logic in `SeekingStateHandler`
which would avoid presenting a frame if it didn't find a frame both
before and after the target frame. However, this logic was unnecessary
as the `on_enter()` function would check if it needed to present a
frame and exit seeking if not.
This allows seeking to succeed if the Seeking handler cannot find a
frame before the one to be seeked to, which could occur if the first
frame of a video is not at timestamp 0.
This adds a timestamp to the debug output when presenting a frame, so
it can be clear the frame spacing between a presented frame and a state
change.
The seeking state will also now print when it early-exits if the seek
point is within the current sample's duration.
The consume(size_t) overload consumes "at most" as many bytes as
requested, but consume() consumes exactly one byte.
This commit makes sure to avoid consuming past EOF.
Fixes#18324.
Fixes#18325.
This makes YouTube's thumbnails start appearing on the homepage.
Yes,seriously.
Simply put, this is because this check failed when Comment had the
incorrect prototype:
90cb97f847/packages/shadycss/src/style-util.js (L397)
This causes it to try and reconvert style sheets that are already in
Shady format, which would cause it to spuriously add things such as
class selectors on the end of tag selectors. This caused nothing to
match the selectors.
When YouTube is generating the thumbnails, it checks if the thumbnail
grid container has a non-zero clientWidth. If it's zero, it simply
bails generating thumbnails. Since the selectors for this container did
not apply, we would not properly create a paint box for it, causing
clientWidth to return zero.
Previously, if we copied the last byte for a length of 100, we'd
recalculate the read span 100 times and memmove one byte 100 times,
which resulted in a lot of overhead.
Now, if we know that we have two consecutive copies of the data, we just
extend the distance to cover both copies, which halves the number of
times that we recalculate the span and actually call memmove.
This takes the running time of the attached benchmark case from 150ms
down to 15ms.
Function `CellSyntaxHighlighter::rehighlight()` direct inserted spans
to TextDocument `m_span` vector missing out important reordering and
merging operation carried out by `TextDocument::set_spans()`.
This caused overlapping spans for a cell with only a `=` symbol
(one for the actual token and one for the highlighting) to
miscalculate `start` and `end` value and a `length` value (of
`size_t` type) with a `0-1` substraction (result: 18446744073709551615)
causing `Utf32View::substring_view()` to fail the
`!Checked<size_t>::addition_would_overflow(offset, length)` assertion
This remove the possibility to directly alter `TextDocument`'s spans
thus forcing the utilization of `HighlighterClient::do_set_spans()`
interface function.
Proper refactor have been applied to
`CellSyntaxHighlighter::rehighlight()` function
The {sin,cos,tan} functions in AK are used as the implementation of the
same function in libm. We cannot use the __builtin_foo functions as
these would just call the libc functions. This was causing an infinite
loop. Fix this by adding a very naive implementation of
AK::{sin, cos,tan}, that is only valid for small inputs. For the other
functions in this file, I added a TODO() such that we'll crash, instead
of infinite looping.
We were crashing on the VERIFY_INTERRUPTS_DISABLED() in
RecursiveSpinlock::unlock, which was caused by the compiler reordering
instructions in `sys$get_root_session_id`. In this function, a SpinLock
is locked and quickly unlocked again, and since the lock and unlock
functions were inlined into `sys$get_root_session_id` and the DAIF::read
was missing the `volatile` keyword, the compiler was free to reorder the
reads from the DAIF register to the top of this function. This caused
the CPU to read the interrupts state at the beginning of the function,
and storing the result on the stack, which in turn caused the
VERIFY_INTERRUPTS_DISABLED() assertion to fail. By adding the `volatile`
modifier to the inline assembly, the compiler will not reorder the
instructions.
In aa40cef2b7, I mistakenly assumed that the crash was related to the
initial interrupts state of the kernel threads, but it turns out that
the missing `volatile` keyword was the actual problem. This commit also
removes that code again.
While the window geometry overlay is centered inside the tile overlay
neither the text nor the location change, so there is no need to
re-render and update it every time the window moves.
