- Capture PrototypeChainValidity before invoking `internal_get()`. A
getter may mutate the prototype chain (e.g., delete itself). Capturing
earlier ensures such mutations invalidate the cached entry and prevent
stale GetById hits.
- When caching, take PrototypeChainValidity from the base object
(receiver), not from the prototype where the property was found.
Otherwise, changes to an intermediate prototype between the base
object and the cached prototype object go unnoticed, leading to
incorrect cache hits.
Specifically, NotificationOptions has this:
```webidl
dictionary NotificationOptions {
// ...
boolean? silent = null;
// ...
};
```
https://notifications.spec.whatwg.org/#dictdef-notificationoptions
Without this patch, we generate this code, which isn't valid:
```c++
silent_value_11 = static_cast<bool>(null);
```
Treating `null` the same as no default value seems like the best option,
as they're equivalent in our C++ types.
This lets us avoid each UI needing to handle link clicks directly, and
lets actions stored in LibWebView avoid awkwardly going through the link
click callbacks to open URLs.
When building with REGEX_DEBUG or ENABLE_ALL_THE_DEBUG_MACROS there are
two issues with linking of bin/TestRegex
- Libraries/LibRegex/RegexDebug.h:76 with undefined reference
regex::OpCode_Compare::variable_arguments_to_byte_string(
AK::Optional<regex::MatchInput const&>) const
- Libraries/LibRegex/RegexByteCode.h:672 with undefined reference
regex::OpCode::name(regex::OpCodeId)
Add REGEX_API on regex::OpCode and regex::OptCode_Compare to allow
access to the classes in bin/TestRegex
When building with DNS_DEBUG or ENABLE_ALL_THE_DEBUG_MACROS there is a
linking issue with Libraries/LibDNS/Resolver.h:1003 with
undefined reference DNS::Messages::Records::SIG::to_string() const
Add DNS_API on DNS:Messages::Records::SIG definition to allow access
to the class in bin/dns, bin/TestDNSResolver, libexec/RequestServer
We often use Optional<T> in a cache pattern such as:
if (!m_cache.has_value())
m_cache = slow_thing();
return m_cache.value();
The new ::ensure() makes it a bit simpler:
return m_cache.ensure([&] { return slow_thing(); });
We already had IC support in PutById for the following cases:
- Changing an existing own property
- Calling a setter located in the prototype chain
This was enough to speed up code where structurally identical objects
(same shape) are processed in a loop:
```js
const arr = [{ a: 1 }, { a: 2 }, { a: 3 }];
for (let obj of arr) {
obj.a += 1;
}
```
However, creating structurally identical objects in a loop was still
slow:
```js
for (let i = 0; i < 10_000_000; i++) {
const o = {};
o.a = 1;
o.b = 2;
o.c = 3;
}
```
This change addresses that by adding a new IC type that caches both the
source and target shapes, allowing property additions to be fast-pathed
by directly jumping to the shape that already includes the new property.
On macOS Tahoe, it is now recommended to show menu item icons. We use
system symbols for this now. Symbols do not have constant variable names
and must be found via the SF Symbols app.
The symbols chosen here were to match Safari as close as possible.
On macOS Tahoe, the zoom button's border is visible when the button
itself is hidden. This feels like a macOS bug, but for now let's hide
the borders as well.
We currently have the issue that our macOS builds are being hosted by
both Apple M1 and M4 runners. By default, these target `-mcpu=apple-m1`
and `-mcpu=apple-m3` respectively, causing their ccache artifacts to be
incompatible and effectively thrashing each other's ccache cache on each
run.
Add a new CMake option `ENABLE_CI_BASELINE_CPU` that toggles a baseline
`-mcpu/-march` compile option for each of the supported build platforms.
When disabled (the default), we now default to `-march=native` which on
Linux could theoretically lead to better performance.
The new CPU baseline is used by the `lagom-template.yml` and
`js-and-wasm-artifacts.yml` workflows, since they both produce artifacts
or caches that might be picked up by other runners. We also enable it
for Flatpak builds, so they target a fixed architecture instead of
whatever architecture the action runner that picked up the job has.
Previously if the directory returned by `downloads_directory()` didn't
exist (or wasn't a directory) when taking a screnshot in headless mode
we try to ask the user for the download directory and fail with the
unhelpful error: QWidget: Must construct a QApplication before a QWidget
While editing, we need to consider whether removing a <br> has any
effect on layout to determine whether its extraneous. This new condition
finds most cases for extraneous <br>s inside block elements.
Global Privacy Control aims to be a replacement for Do Not Track. DNT
ended up not being a great solution, as it wasn't enforced by law. This
actually resulted in the DNT header serving as an extra fingerprinting
data point.
GPC is becoming enforced by law in USA states such as California and
Colorado. CA is further working on a bill which requires that browsers
implement such an opt-out preference signal (OOPS):
https://cppa.ca.gov/announcements/2025/20250911.html
This patch replaces DNT with GPC and hooks up the associated settings.
The extra representations of a video element are unnecessary to specify
in VideoPaintable, as the playback system itself should take care of
all the details of the representation as specified.
According to RFC 2046, the BNF of the form data body is:
multipart-body := [preamble CRLF]
dash-boundary transport-padding CRLF
body-part *encapsulation
close-delimiter transport-padding
[CRLF epilogue]
Where "epilogue" is any text that "may be ignored or discarded". So we
should stop parsing the body once we encounter the terminating delimiter
("--").
Note that our parsing function is from an attempt to standardize the
grammar in the spec: https://andreubotella.github.io/multipart-form-data
This proposal hasn't been updated in ~4 years, and the fetch spec still
does not have a formal definition of the body string.
