* fix: delete desktop capturers when they're not needed
Delete desktop capturer objects by resetting the DesktopMediaList
objects that own them after the sources have been collected. Capturers
that are not delegated are already being reset via a patch on
NativeDesktopMediaList. That is not safe for delegated capturers as
thumbnail generation depends on user events. Deleting the
DesktopMediaList operation is safe for all capturers and releases OS
capture resources as soon as possible.
* fix: add a patch to clean up PipeWire resources
Adding a patch to workaround a Chromium issue:
https://bugs.chromium.org/p/chromium/issues/detail?id=1467060
The patch can be removed when the issue is resolved.
Screensharing with PipeWire via XDG Desktop Portal requires explicit
user permission via permission dialogs. Chromium has separate tabs for
screens and windows and thus its portal implementation requests
permissions separately for each. However, the screencast portal has no
such limitation and supports both screens and windows in a single
request.
WebRTC now supports this type of capture in a new method called
called `CreateGenericCapturer`. The `desktopCapturer` implementation has
been modified to use it. Additionally, Chromium has been patched to use
same generic capturer to ensure that the source IDs remain valid for
`getUserMedia`.
* chore: bump chromium in DEPS to 117.0.5866.0
* chore: bump chromium in DEPS to 117.0.5868.0
* chore: update mas_no_private_api.patch
Xref: https://chromium-review.googlesource.com/c/chromium/src/+/4634925
Minor manual patch syncing due to upstream code shear
* chore: update mas_disable_remote_layer.patch
Xref: https://chromium-review.googlesource.com/c/chromium/src/+/4647191
Manually sync patch to minor upstream code shear
* chore: update mas_disable_remote_accessibility.patch
Xref: https://chromium-review.googlesource.com/c/chromium/src/+/4641746
No manual changes; patch applied with fuzz
* chore: update mas_avoid_usage_of_private_macos_apis.patch
Xref: https://chromium-review.googlesource.com/c/chromium/src/+/4634925
Manually sync base/process/launch_mac.cc to minor upstream shear
Manually sync base/mac/foundation_util.mm to upstream changes:
_CFIsObjC use has been removed upstream, so we no longer need
to remove it 🎉
* chore: update printing.patch
Xref: https://chromium-review.googlesource.com/c/chromium/src/+/4658496
Manually sync patch to minor upstream code shear
* chore: update disable_color_correct_rendering.patch
Xref: https://chromium-review.googlesource.com/c/chromium/src/+/4625254
Manually sync patch to minor upstream code shear
* chore: update feat_expose_raw_response_headers_from_urlloader.patch
Xref: services/network/public/cpp/resource_request.cc
No manual changes; patch applied with fuzz
* chore: update add_electron_deps_to_license_credits_file.patch
Xref: https://chromium-review.googlesource.com/c/chromium/src/+/4634961
No manual changes; patch applied with fuzz
* chore: update build_only_use_the_mas_build_config_in_the_required_components.patch
Xref: https://chromium-review.googlesource.com/c/chromium/src/+/4648411
No manual changes; patch applied with fuzz
* chore: update patches
* fixup! chore: update add_electron_deps_to_license_credits_file.patch
chore: license files must be an array
* chore: bump chromium in DEPS to 117.0.5870.0
* chore: update patches
* chore: run ./script/gen-libc++-filenames.js
* chore: update json_parse_errors_made_user-friendly.patch
Xref: https://chromium-review.googlesource.com/c/v8/v8/+/4652014
v8 error message changed upstream; update Node test to match it
* chore: bump chromium in DEPS to 117.0.5872.0
* chore: update patches
* chore: explicitly cast x11::Window to unsigned int
Xref: https://chromium-review.googlesource.com/c/chromium/src/+/4661049
This is an `enum class Window : uint32_t` defined in ui/gfx/x/xproto.h.
Previous versions of clang let this implicit cast happen,
but it generates a warning in the new clang roll.
* chore: remove unused #include
Xref: https://chromium-review.googlesource.com/c/chromium/src/+/4650453
header was removed upstream, so FTBFS unless removed here
* chore: add include guard patch
Xref: https://chromium-review.googlesource.com/c/chromium/src/+/4628373
h/t @jkleinsc
* chore: bump chromium in DEPS to 117.0.5874.0
* chore: update render_widget_host_view_mac.patch
Xref: https://chromium-review.googlesource.com/c/chromium/src/+/4661244
Manually sync patch to minor upstream code
* chore: update mas_disable_remote_accessibility.patch
Xref: https://chromium-review.googlesource.com/c/chromium/src/+/4653209
Manually sync patch to upstream code shear
* chore: update build_only_use_the_mas_build_config_in_the_required_components.patch
Xref: https://chromium-review.googlesource.com/c/chromium/src/+/4653209
Manually sync patch to minor upstream code shear
* chore: update GetInitiatorProcessId()
Xref: https://chromium-review.googlesource.com/c/chromium/src/+/4641991
trivial upstream naming change: s/ProcessID/ProcessId/
* chore: sync to upstream SetInputRegion() changes
Xref: https://chromium-review.googlesource.com/c/chromium/src/+/4665245
Simple upstream chang: SetInputRegion() used to take a gfx::Rect* where
`nullptr` meant "no opaque region". The function signature changed to
absl::optional<gfx::Rect> w/the same meaning.
* chore: sync to upstream SetOpaqueRegion() changes
Xref: https://chromium-review.googlesource.com/c/chromium/src/+/4656738
Simple upstream chang: SetOpaqueRegion() used to take a vector<Rect>* where
`nullptr` meant "no opaque region". The function signature changed to
absl::optional<std::vector<gfx::Rect>> w/the same meaning.
* chore: update patches
* chore: bump chromium in DEPS to 117.0.5876.0
* chore: update mas_disable_remote_accessibility.patch
Xref: https://chromium-review.googlesource.com/c/chromium/src/+/4658375
We no longer need to patch out a field that's now removed upstream.
RenderWidgetHostNSViewBridgeOwner.remote_accessibility_element_
* chore: update feat_filter_out_non-shareable_windows_in_the_current_application_in.patch
Xref: https://chromium-review.googlesource.com/c/chromium/src/+/4658680
Manually sync patch to upstream code shear (ARC adoption).
* chore: update patches
* fix: -Werror,-Wshadow error in Node.js
* chore: bump chromium in DEPS to 117.0.5878.0
* chore: bump chromium in DEPS to 117.0.5880.0
* chore: bump chromium in DEPS to 117.0.5880.4
* chore: update patches
* 4658680: Convert /content/browser to use ARC
https://chromium-review.googlesource.com/c/chromium/src/+/4658680
* 4669995: Remove CFToNSCast and NSToCFCast
https://chromium-review.googlesource.com/c/chromium/src/+/4669995
* WIP: 4658680: Convert /content/browser to use ARC
https://chromium-review.googlesource.com/c/chromium/src/+/4658680
* chore: update printing patch after rebase
* chore: bump chromium in DEPS to 117.0.5882.0
* Revert "WIP: 4658680: Convert /content/browser to use ARC"
This reverts commit c8559ea448ad93123865a6fa8af6fe2767e43396.
* Revert "4669995: Remove CFToNSCast and NSToCFCast"
This reverts commit 38e145f33d844e077b09e1fb119fae9d5d7bbc3e.
* chore: bump chromium in DEPS to 117.0.5874.0
* 4661244: Convert /content/browser/renderer_host to use ARC
https://chromium-review.googlesource.com/c/chromium/src/+/4661244
* 4653209: Convert /ui/views:views to use ARC
https://chromium-review.googlesource.com/c/chromium/src/+/4653209
* chore: fixup mas_disable_remote_accessibility.patch
* chore: fixup render_widget_host_view_mac.patch
* chore: update to newer clang
* chore: bump chromium in DEPS to 117.0.5884.1
(cherry picked from commit a3879acfde8f3d962d58d6ad0632164ccb88ee63)
* 4669995: Remove CFToNSCast and NSToCFCast
https://chromium-review.googlesource.com/c/chromium/src/+/4669995
(cherry picked from commit 38e145f33d844e077b09e1fb119fae9d5d7bbc3e)
(cherry picked from commit b1224ab7e00aedbd0e5dc78ebb3c6162573b28c4)
* 4658680: Convert /content/browser to use ARC
https://chromium-review.googlesource.com/c/chromium/src/+/4658680
(cherry picked from commit 77039a323b9ebb3e8edb3a92a3b94a8d7d026a32)
* Rename and cleanup ExtensionsBrowserClient functions
| https://chromium-review.googlesource.com/c/chromium/src/+/4665670
(cherry picked from commit 677a2e646a8d82da60dc252b08b320c3ddff6be6)
* chore: bump chromium in DEPS to 117.0.5886.0
* chore: update patches
* Revert "chore: update patches"
This reverts commit db9294f9446492d4a3d63f5f3ecc551aeeeb2e26.
* Revert "chore: bump chromium in DEPS to 117.0.5886.0"
This reverts commit a7de0276e6e0eeecfdbb17f9bbc27c01b4bbfc24.
* build: use built toolchains instead of locally installed VSCode
* fixup: build: use built toolchains instead of locally installed VSCode
* chore: add missing mojo dep
---------
Co-authored-by: electron-roller[bot] <84116207+electron-roller[bot]@users.noreply.github.com>
Co-authored-by: Charles Kerr <charles@charleskerr.com>
Co-authored-by: PatchUp <73610968+patchup[bot]@users.noreply.github.com>
Co-authored-by: John Kleinschmidt <jkleinsc@electronjs.org>
Co-authored-by: Shelley Vohr <shelley.vohr@gmail.com>
* fix: use StartUpdating method for PipeWire capturer
Fixed a crash related to PipeWire capturer by adapting to Chromium's
interface changes. Chromium expects a call to
`NativeDesktopMediaList::StartUpdating` with an implementation of
`DesktopMediaListObserver` for delegated capturers like PipeWire. This
interface allows listening to user permission events and listing
sources only after the user has made a choice on the permission dialog.
The interface has been implemented by an inner class to allow listening
to screen and window capture permissions concurrently using two
instances of the class. A patch that was resetting the capturer on the
first refresh has been changed to exclude PipeWire. PipeWire capturer
object will follow the lifecycle of `NativeDesktopMediaList`, as is the
case in Chromium.
Fixes#37463
* fix: wait for thumbnails from PipeWire when necessary
The PipeWire stream starts after the dialog is dismissed. If the sources
are listed immediately afterwards, the thumbnail may not have been
generated by that time. Explicitly wait for both thumbnail generation
and a selection on the source dialog before listing sources.
* fix: menu border being created properly on Windows 11
* chore: update patches
---------
Co-authored-by: PatchUp <73610968+patchup[bot]@users.noreply.github.com>
* chore: bump chromium in DEPS to 117.0.5846.0
* chore: update patches
* 4628901: Bump the macOS deployment target to 10.15
https://chromium-review.googlesource.com/c/chromium/src/+/4628901
* 4593350: [Private Network Access] Trigger Permission Prompt
https://chromium-review.googlesource.com/c/chromium/src/+/4593350
* 4631011: Remove unlaunched "InstallReplacementAndroidApp" Platform App APIs
https://chromium-review.googlesource.com/c/chromium/src/+/4631011
* chore: disable API deprecation warnings in NSKeyedArchiver
* chore: update libcxx filenames
* chore: bump chromium in DEPS to 117.0.5848.2
* chore: update feat_add_set_theme_source_to_allow_apps_to.patch
Xref: https://chromium-review.googlesource.com/c/chromium/src/+/4629743
No manual changes; patch succeeded with fuzz
* chore: update process_singleton.patch
Xref: https://chromium-review.googlesource.com/c/chromium/src/+/4605398
Trivial manual patch adjustments to account for code shear.
* chore: remove electron::BrowserContext::GetMediaDeviceIDSalt()
Xref: https://chromium-review.googlesource.com/c/chromium/src/+/4608130
upstream tldr:
- content::BrowserContext::GetMediaDeviceIDSalt()
- content::ContentBrowserClient::ArePersistentMediaDeviceIDsAllowed()
+ content::ContentBrowserClient::GetMediaDeviceIDSalt()
This commit leaves ElectronBrowserContext::GetMediaDeviceIDSalt() in
place (now non-virtual, non-override). It is now called by the new
function ElectronBrowserClient::GetMediaDeviceIDSalt().
As a followup, we might want to consider using the new upstream
media_device_salt::MediaDeviceSaltService and removing our
electron::MediaDeviceIDSalt code. CC @MarshallOfSound for 2nd
opinion since he has done the most work on MediaDeviceIDSalt and
may have more context.
* chore: fix iwyu breakage
Xref: https://chromium-review.googlesource.com/c/chromium/src/+/4629624
electron_browser_main_parts.cc uses ui::ColorProviderManager but didn't
include it. Things worked anyway because we got it indirectly from
content/public/browser/web_contents.h until 4629624.
* chore: remove call to base::mac::IsAtLeastOS10_14
upstream has bumped minimum version to 10.15 so this call is moot?
* chore: remove obsolete API_AVAILABLE calls in IAP
upstream has bumped minimum version to 10.15 so this call is moot?
* chore: remove obsolete API_AVAILABLE calls in electron_application_delegate
upstream has bumped minimum version to 10.15 so this call is moot?
* chore: remove broken-before-macOS-10.15 patch in mas_avoid_usage_of_private_macos_apis.patch
Upstream has bumped minimum to macOS 10.15
* chore: remove @available(macOS 10.14) check
Upstream minimum requirement for macOS is now 10.15
* chore: update patches
* chore: bump chromium in DEPS to 117.0.5850.0
* chore: update patches
* chore: bump chromium in DEPS to 117.0.5852.0
* chore: update patches
* Move two params from NetworkContextParams to NetworkContextFilePaths.
https://chromium-review.googlesource.com/c/chromium/src/+/4615930
* WebUSB: Add exclusionFilters to USBRequestDeviceOptions
https://chromium-review.googlesource.com/c/chromium/src/+/4614682
* Convert /chrome/browser/ui to use ARC
https://chromium-review.googlesource.com/c/chromium/src/+/4615920
* fixup! Bump the macOS deployment target to 10.15
* fixup! Bump the macOS deployment target to 10.15
* chore: update libcxx files
* win: Remove 10Glass from Windows10Glass function and var names
https://chromium-review.googlesource.com/c/chromium/src/+/4641314
* chore: revert 392e5f43 from chromium
* Add an ExecutionContext to ScriptState
https://chromium-review.googlesource.com/c/chromium/src/+/4609446
* fixup! Add an ExecutionContext to ScriptState
* chore: fix header
* Revert "chore: revert 392e5f43 from chromium"
This reverts commit b7f782943e4ce83cae8cd35780d8d3618cf0772c.
* fix: return correct min/max sizes in WinFrameView
* fixup! Revert chore: revert 392e5f43 from chromium
* fixup! Add an ExecutionContext to ScriptState
* Revert "fixup! Revert chore: revert 392e5f43 from chromium"
This reverts commit 7e2c7281abfc4f309255339fdba073d90a9ae3eb.
* Revert "fix: return correct min/max sizes in WinFrameView"
This reverts commit 3f418b1ab5155686730e087ae6cabe4a21b4bb61.
* Revert "Revert "chore: revert 392e5f43 from chromium""
This reverts commit 56296d8b7c434147e032e3c3b08c0e371b6c27ba.
---------
Co-authored-by: electron-roller[bot] <84116207+electron-roller[bot]@users.noreply.github.com>
Co-authored-by: PatchUp <73610968+patchup[bot]@users.noreply.github.com>
Co-authored-by: Shelley Vohr <shelley.vohr@gmail.com>
Co-authored-by: Charles Kerr <charles@charleskerr.com>
Co-authored-by: deepak1556 <hop2deep@gmail.com>
Co-authored-by: Cheng Zhao <zcbenz@gmail.com>
fix: linker error missing uv__strtok
This symbol is referenced inside what seems to be dead code
in `uv__search_path` in third_party/electron_node/deps/uv/src/unix/core.c
When compiling in LTO mode, the reference is removed,
but not during a non-LTO build.
