* perf: prefer NewFromUtf8Literal() over NewFromUtf8() for string literals
the string length is known at compile time and no need to call ToLocalChecked()
Co-authored-by: Charles Kerr <charles@charleskerr.com>
* perf: string length is known when calling NewFromUtf8(), so use it
Co-authored-by: Charles Kerr <charles@charleskerr.com>
* perf: remove unnecessary calls to c_str()
these just force the code being called to have to recalculate the string length
Co-authored-by: Charles Kerr <charles@charleskerr.com>
---------
Co-authored-by: trop[bot] <37223003+trop[bot]@users.noreply.github.com>
Co-authored-by: Charles Kerr <charles@charleskerr.com>
fix: respect setAlwaysOnTop before showInactive on Linux under X11 (#44078)
Co-authored-by: trop[bot] <37223003+trop[bot]@users.noreply.github.com>
Co-authored-by: John Beutner <31829097+cptpcrd@users.noreply.github.com>
fixup e8948397 really fix the warning this time
Co-authored-by: trop[bot] <37223003+trop[bot]@users.noreply.github.com>
Co-authored-by: Charles Kerr <charles@charleskerr.com>
perf: use ArrayBuffer::Data() instead of GetBackingStore()->Data() (#44067)
perf: use ArrayBuffer::Data() API
Replace our `GetBackingStore()->Data()` calls with this instead.
Explained by the V8 docs, ArrayBuffer.Data() is
> More efficient shortcut for GetBackingStore()->Data(). The
> returned pointer is valid as long as the ArrayBuffer is alive.
notify_notification_set_hint_string() is deprecated, so let's use
notify_notification_set_hint() instead.
Xref: 2fe1748295
Co-authored-by: trop[bot] <37223003+trop[bot]@users.noreply.github.com>
Co-authored-by: Charles Kerr <charles@charleskerr.com>
perf: build GettextPackage name at compile time
Co-authored-by: trop[bot] <37223003+trop[bot]@users.noreply.github.com>
Co-authored-by: Charles Kerr <charles@charleskerr.com>
* fix: -Wunsafe-buffer-usage warnings in GdkPixbufFromSkBitmap()
Co-authored-by: Charles Kerr <charles@charleskerr.com>
* refactor: don't change previous behavior for 0-height images
Is a 0x0 image even a thing? I'm not sure; but just in case, let's
treat it the same way the previous implementation did.
Co-authored-by: Charles Kerr <charles@charleskerr.com>
---------
Co-authored-by: trop[bot] <37223003+trop[bot]@users.noreply.github.com>
Co-authored-by: Charles Kerr <charles@charleskerr.com>
test: expect a `sender-pid` hint in Linux notifications.
This PR ensures that the `sender-pid` hint is set for new notifications.
It also updates the spec to confirm that DBus receives the hint and that
it has the correct value.
This fixes a spec failure when running libnotify >= 0.7.12 (2022-05-05).
Starting with that version, libnotify started injecting `sender-pid` if
not provided by the client. So our tests received a slightly different
DBus payload depending on what version of libnotify was installed,
causing our deep-equals tests to fail.
By always providing and testing the `sender-pid` hint, our behavior and
tests should be consistent across distros.
Co-authored-by: trop[bot] <37223003+trop[bot]@users.noreply.github.com>
Co-authored-by: Charles Kerr <charles@charleskerr.com>
Right now DelayedNativeViewHost attaches its underlying native view
when it's being attached to a widget but it doesn't detach it when
it's being detached. It may lead to use-after-free and crash.
Co-authored-by: trop[bot] <37223003+trop[bot]@users.noreply.github.com>
Co-authored-by: Cezary Kulakowski <cezary@openfin.co>
* fix: -Wunsafe-buffer-usage warnings in Clipboard::WriteBuffer()
Co-authored-by: Charles Kerr <charles@charleskerr.com>
* chore: add a DCHECK to confirm the BigBuffer is full
Co-authored-by: Charles Kerr <charles@charleskerr.com>
---------
Co-authored-by: trop[bot] <37223003+trop[bot]@users.noreply.github.com>
Co-authored-by: Charles Kerr <charles@charleskerr.com>
fix: Emit() should not leak converted arg handles into caller's HandleScope
Co-authored-by: trop[bot] <37223003+trop[bot]@users.noreply.github.com>
Co-authored-by: Charles Kerr <charles@charleskerr.com>
Closes https://github.com/electron/electron/issues/43714.
Fixes an issue where the resizing border was not being handled correctly on Linux WCO
caption buttons. This is now taken into account as a part of the NonClientHitTest.
Co-authored-by: trop[bot] <37223003+trop[bot]@users.noreply.github.com>
Co-authored-by: Shelley Vohr <shelley.vohr@gmail.com>
* fix: restore Chromium default Content-Disposition header parsing
Co-authored-by: Milan Burda <milan.burda@gmail.com>
* Update api-web-request-spec.ts
Co-authored-by: Milan Burda <milan.burda@gmail.com>
---------
Co-authored-by: trop[bot] <37223003+trop[bot]@users.noreply.github.com>
Co-authored-by: Milan Burda <milan.burda@gmail.com>
* perf: use v8::Object* as direct keys instead of using hash + a linked list
* refactor: use v8::Local<v8::Object> as the key
Co-authored-by: Charles Kerr <charles@charleskerr.com>
refactor: use an EscapableHandleScope
Co-authored-by: trop[bot] <37223003+trop[bot]@users.noreply.github.com>
Co-authored-by: Charles Kerr <charles@charleskerr.com>
reassign the uv_handle_t of the source
Co-authored-by: trop[bot] <37223003+trop[bot]@users.noreply.github.com>
Co-authored-by: Charles Kerr <charles@charleskerr.com>
refactor: avoid redundant Promise.GetContext calls
Several Promise methods call `GetContext()` multiple times. From looking
at the assembly in obj/electron/electron_lib/promise.o, these redundant
calls are actually being made -- they aren't optmized out.
This PR keeps the return value in a local variable to avoid extra calls.
Co-authored-by: trop[bot] <37223003+trop[bot]@users.noreply.github.com>
Co-authored-by: Charles Kerr <charles@charleskerr.com>
refactor: take a uint8_t span in ValidateIntegrityOrDie()
Doing some groundwork for fixing unsafe base::File() APIs:
- Change ValidateIntegrityOrDie() to take a span<const uint8_t> arg.
We'll need this to migrate asar's base::File API calls away from the
ones tagged `UNSAFE_BUFFER_USAGE` because the safe counterparts use
span<uint8_t> too.
- Simplify ValidateIntegrityOrDie()'s implementation by using
crypto::SHA256Hash() instead of reinventing the wheel.
Co-authored-by: trop[bot] <37223003+trop[bot]@users.noreply.github.com>
Co-authored-by: Charles Kerr <charles@charleskerr.com>
fix: confirm a v8::Value is a v8::Object before casting it
Co-authored-by: trop[bot] <37223003+trop[bot]@users.noreply.github.com>
Co-authored-by: Charles Kerr <charles@charleskerr.com>
* fix: free UvTaskRunner timers only after they are closed
Co-authored-by: Charles Kerr <charles@charleskerr.com>
* refactor: UvTaskRunner now holds UvHandles
Co-authored-by: Charles Kerr <charles@charleskerr.com>
---------
Co-authored-by: trop[bot] <37223003+trop[bot]@users.noreply.github.com>
Co-authored-by: Charles Kerr <charles@charleskerr.com>
