A couple of tweaks here to make it work better:
- Call `set_cursor_and_focus_line()` to make the Editor scroll to the
symbol's location.
- Remove focus from the Locator's text box so your cursor jumps to the
Editor instead of staying in the Locator.
This being a Popup window meant it behaved in a couple of janky ways:
- It would steal the focus each time it was shown, so after every key
press in the TextBox.
- It would disappear when you focused that TextBox again.
Using the Autocomplete window type fixes both of these. While the
Locator is not technically an autocomplete, it shares the general "type
and get suggestions based on the input, which you can select" behavior,
so this is close enough.
Also change a Vector<Handle> to a Vector<NonnullGCPtr> while we're
here, since there's no need to use handles for members of a cell.
Fixes an ASAN error on the HTML/Window-postMessage.html test.
We should be using WebIDL Promise in these methods, per the spec
references in other specifications. That is, the HTML spec always links
to WebIDL when it talks about `Promise<T>`. The Crypto seems to be
missing those cross-references.
While we're here, actually resolve the digest promise in parallel.
Previously, our code for the fixup of table rows assumed that missing
cells in a table row must be sequential. This may not be true if the
table contains cells have a rowspan greater than one.
We now split up symbols into zero-sized symbols that label single
instructions (no need to separate them by newlines; they are used for
jump labels and relocation targets within a larger block of code) and
ranged symbols that label functions. Empty symbols are discarded since
at least RISC-V ELF files contain quite a few of those. Zero-sized
symbols and ranged symbols are handled almost the same, but this way we
can make sure that zero-sized symbols don't interfere with ranged
symbol's newline separation logic. For zero-sized symbols, the "symbol
contains address" logic is updated so they actually contain the one
address they're pointing at, fixing the bug with many "dangling"
zero-sized symbols after a function that contained them. Zero-sized
labels are also no longer printed as a start-end range, since that is
unnecessary visual noise.
We previously were considering Float and Doubles as non-supported types.
But this was done in a sneaky way, by letting them hit the default case
in the `read_type` method. So, when I ported this function to the
generator we started to make this types flow into the system without a
proper support there. Since 3124c161, we would have crashes on images
containing tags with a floating point value.
A lot of images format use Exif to store there metadata. As Exif is
based on the TIFF structure, the TIFF decoder can, without modification
be able to decode the metadata. We only need a new API to explicitly
mention that we only need the metadata.
Support for JPEGs embedded in TIFF images was introduced with TIFF 6.0.
However, this implementation had major issues. It was so problematic
that they decided to reimplement it from scratch in 1995, three years
later. The two incarnations are obviously incompatible.
For more details see:
https://www.awaresystems.be/imaging/tiff/specification/TIFFTechNote2.txt
When present, the alpha channel is also affected by the horizontal
differencing predictor.
The test case was generated with GIMP with the following steps:
- Open an RGB image
- Add a transparency layer
- Export as TIFF with the LZW compression scheme
Remove all functions with platform #if's from fenv, and
add arch dependent implementations instead. The build
system now selects the implementation based on the platform.
Doesn't matter if both profiles are sRGB since
inv(A) * A == A * inv(A), but when converting e.g. a P3 image to
sRGB, the colors are very off if the matrices are the wrong way round
here.
perform_call() wants a ReadonlySpan<Value>, so just grab a slice of the
current register window instead of making a MarkedVector.
10% speed-up on this function call microbenchmark:
function callee(a, b, c) { }
function caller(callee) {
for (let i = 0; i < 10_000_000; ++i)
callee(1, 2, 3)
}
caller(callee)
This is an option supported by coreutils, so we might as well support
it too.
It allows users to wrap their encoded output after the "column" value
they provide.
This commit also has the Markdown look more like what we see
when running ArgsParser::print_usage_markdown() (and it fixes some
of the examples).
Value::Value(double) already converts double to int when it is safe, no
need to check for this here explicitly. While this technically removes
an optimization, I doubt that it will regress performance in any
measurable way.
This method (unlike can_read_line) ensures that the delimiter is present
in the buffer, and doesn't return true after eof when the delimiter is
absent.
For parameters that exist strictly as "locals", we can save time and
space by not adding them to the function environment.
This is a speed-up across the board on basically every test.
For example, ~11% on Octane/typescript.js :^)
Spec defines `[LegacyNullToEmptyString]` on `value` argument of
`setProperty` but since `internal_set` calls `setProperty` directly
instead of using IDL generated binding, we need to make sure that null
is treated as empty string.
Fixes items grid loading on https://d.rsms.me/stuff/
Destructor of AccelGfxContext needs to make sure that correct OpenGL
context is active so that destructors of its members could proceed
destroying they resources (for example framebuffer owned by
AccelGfx::Canvas).
Fixes https://github.com/SerenityOS/serenity/issues/22879
The layout system can't currently answer the question "what height does
this Label want to be, if it has a certain width available?" Instead it
relies on counting newlines, which doesn't work in a lot of cases. This
made the notification windows look and behave in a funky way when their
text wraps onto multiple lines.
This patch uses TextLayout to measure how many lines we need, and then
manually sets the Label and Window heights to match. It's a bit hacky,
hence the FIXME, but it does make things behave the way they are
supposed to.
FontDatabase.h with its includes add up to quite a lot of code. In the
next commit, compiled GML files are going to need to access the
FontWeight enum, so let's allow them to do that without pulling in lots
of other things.
Also, change users to include FontWeight.h instead of FontDatabase.h
where appropriate.
