* chore: migrate from base::StringPiece to std::string_view in keyboard_util.cc
* chore: migrate from base::StringPiece to std::string_view in error_thrower.cc
* chore: migrate from base::StringPiece to std::string_view in electron_api_web_contents.cc
* chore: migrate from base::StringPiece to std::string_view in gin_helper/dictionary.h
* chore: migrate from base::StringPiece to std::string_view in electron_api_url_loader.cc
* chore: phase out internal use of base:::StringPiece
`base::StringPiece` is being phased out upstream. Its code has been
removed upstream and it's just a typedef for `std::string_view`.
They haven't removed the typedef yet, so this PR tries to get ahead
of future breakage by migrating "internal" use (i.e. leaving alone the
places where the `base::StringPiece` name is coming from an upstream
method that we override).
Xref: https://bugs.chromium.org/p/chromium/issues/detail?id=691162
Xref: 4294483
Xref: https://docs.google.com/document/d/1d4RnD1uAE2t4iANR0nXy82ASIPGsPuw2mpO6v6T7JKs
* chore: initial prototype of net api from utility process
* chore: update url loader to work on both browser and utility processes
* chore: add net files to utility process bundle
* chore: re-add app ready check but only on main process
* chore: replace browser thread dcheck's with sequence checker
* refactor: move url loader from browser to common
* refactor: move net-client-request.ts from browser to common
* docs: add utility process to net api docs
* refactor: move net module app ready check to browser only
* refactor: switch import from main to common after moving to common
* test: add basic net module test for utility process
* refactor: switch browser pid with utility pid
* refactor: move electron_api_net from browser to common
* chore: add fetch to utility net module
* chore: add isOnline and online to utility net module
* refactor: move net spec helpers into helper file
* refactor: break apart net module tests
Adds two additional net module test files: `api-net-session-spec.ts` for
tests that depend on a session being available (aka depend on running on
the main process) and `api-net-custom-protocols-spec.ts` for custom
protocol tests. This enables running `api-net-spec.ts` in the utility
process.
* test: add utility process mocha runner to run net module tests
* docs: add utility process to net module classes
* refactor: update imports in lib/utility to use electron/utility
* chore: check browser context before using in main process
Since the browser context supplied to the SimpleURLLoaderWrapper can now
be null for use in the UtilityProcess, adding a null check for the main
process before use to get a more sensible error if something goes wrong.
Co-authored-by: Cheng Zhao <github@zcbenz.com>
* chore: remove test debugging
* chore: remove unnecessary header include
* docs: add utility process net module limitations
* test: run net module tests in utility process individually
* refactor: clean up prior utility process net tests
* chore: add resolveHost to utility process net module
* chore: replace resolve host dcheck with sequence checker
* test: add net module tests for net.resolveHost
* docs: remove utility process limitation for resolveHost
---------
Co-authored-by: deepak1556 <hop2deep@gmail.com>
Co-authored-by: Cheng Zhao <github@zcbenz.com>
* feat: add blinkUtils module with getPathForFile method
This is designed to replace the File.path augmentation
we currently have in place to allow apps to get the filesystem
path for a file that blink has a representation of.
File.path is non-standard and messes with certain websites, using
a method like this is effectively 0-cost and removes one of the final
deviations we have with web standards.
* add error
* refactor: update per PR feedback
* chore: update patches
* oops
* chore: update patches
* chore: update patches
* feat: add blinkUtils module with getPathForFile method
This is designed to replace the File.path augmentation
we currently have in place to allow apps to get the filesystem
path for a file that blink has a representation of.
File.path is non-standard and messes with certain websites, using
a method like this is effectively 0-cost and removes one of the final
deviations we have with web standards.
* add error
* refactor: update per PR feedback
* chore: update patches
* oops
* chore: update patches
* chore: update patches
* chore: update patches
* fix: provide isolate to WebBlob::FromV8Value
* chore: add tests
* build: fix depshash mismatch on arm64 macOS
---------
Co-authored-by: PatchUp <73610968+patchup[bot]@users.noreply.github.com>
* fix: allow ESM loads from within ASAR files
* fix: ensure that ESM entry points finish loading before app ready
* fix: allow loading ESM entrypoints via default_app
* fix: allow ESM loading for renderer preloads
* docs: document current known limitations of esm
* chore: add patches to support blending esm handlers
* refactor: use SetDefersLoading instead of JoinAppCode in renderers
Blink has it's own event loop so pumping the uv loop in the renderer is not enough, luckily in blink we can suspend the loading of the frame while we do additional work.
* chore: add patch to expose SetDefersLoading
* fix: use fileURLToPath instead of pathname
* chore: update per PR feedback
* fix: fs.exists/existsSync should never throw
* fix: convert path to file url before importing
* fix: oops
* fix: oops
* Update docs/tutorial/esm-limitations.md
Co-authored-by: Jeremy Rose <jeremya@chromium.org>
* windows...
* windows...
* chore: update patches
* spec: fix tests and document empty body edge case
* Apply suggestions from code review
Co-authored-by: Daniel Scalzi <d_scalzi@yahoo.com>
Co-authored-by: Jeremy Rose <jeremya@chromium.org>
* spec: add tests for esm
* spec: windows
* chore: update per PR feedback
* chore: update patches
* Update shell/common/node_bindings.h
Co-authored-by: Jeremy Rose <jeremya@chromium.org>
* chore: update patches
* rebase
* use cjs loader by default for preload scripts
* chore: fix lint
* chore: update patches
* chore: update patches
* chore: fix patches
* build: debug depshash
* ?
* Revert "build: debug depshash"
This reverts commit 0de82523fb93f475226356b37418ce4b69acdcdf.
* chore: allow electron as builtin protocol in esm loader
* Revert "Revert "build: debug depshash""
This reverts commit ff86b1243ca6d05c9b3b38e0a6d717fb380343a4.
* chore: fix esm doc
* chore: update node patches
---------
Co-authored-by: Jeremy Rose <jeremya@chromium.org>
Co-authored-by: electron-patch-conflict-fixer[bot] <83340002+electron-patch-conflict-fixer[bot]@users.noreply.github.com>
Co-authored-by: PatchUp <73610968+patchup[bot]@users.noreply.github.com>
Co-authored-by: Daniel Scalzi <d_scalzi@yahoo.com>
perf: prefer Sorted variant of MakeFixedFlatSet()
4660000
says that the sorted version is simpler at compile time because it
can skip MakeFixedFlatSet()'s compile-time dynamic sorting.
We're currently building these on the heap with `std::set<std::string>`
but this can be a very small compile-time container instead.
Marking as 'refactor' rather than 'perf' since this isn't called often,
but moving from heap to compile-time is good and using this container
makes the code more readable.
* chore: bump chromium in DEPS to 111.0.5522.0
* chore: bump chromium in DEPS to 111.0.5524.0
* chore: bump chromium in DEPS to 111.0.5526.0
* chore: bump chromium in DEPS to 111.0.5528.0
* chore: update patches/chromium/mas_avoid_usage_of_private_macos_apis.patch
Xref: 4132807
Fix simple code shear
* chore: update patches/chromium/unsandboxed_ppapi_processes_skip_zygote.patch
Xref: 4130675
Fix simple code shear
* chore: update patches/chromium/hack_plugin_response_interceptor_to_point_to_electron.patch
Xref: 4144281
Fix simple code shear; applied cleanly w/patch-fuzz
* chore: update patches/chromium/disable_unload_metrics.patch
Xref: 4126173
Fix simple code shear; applied cleanly w/patch-fuzz
* chore: update patches/chromium/feat_add_data_parameter_to_processsingleton.patch
Xref: 4144281
Fix simple code shear; applied cleanly w/patch-fuzz
* chore: update patches/chromium/preconnect_manager.patch
4144281
Fix simple code shear; applied cleanly w/patch-fuzz
* chore: update patches/v8/force_cppheapcreateparams_to_be_noncopyable.patch
3533019
Fix simple code shear; applied cleanly w/patch-fuzz
* chore: update patches
* chore: update patches/chromium/add_maximized_parameter_to_linuxui_getwindowframeprovider.patch
Xref: 4128765
Upstream added a new call to HeaderContext(), whose signature we have patched
* chore: bump chromium in DEPS to 111.0.5530.0
* chore: update patches
* Move ChildProcessHost* from content/common to content/browser
Xref: Move ChildProcessHost* from content/common to content/browser
* Remove RenderViewHostChanged
Xref: 4134103
[upstream removal of RenderViewHostChanged]
Xref: 4092763
Xref: 4093234
Xref: 4133892
Xref: 4134103
[examples of upstream code adjusting to the change]
Upstream handles this change in roughly two approaches:
1. Move the code over to RenderFrameHostChanged(old_host, new_host)
but test for new_host->IsInPrimaryMainFrame() before acting
2. Migrate to the PrimaryPageChanged(page) API and use
page.GetMainDocument() to get the RenderFrameHost.
I've chosen 1. because electron_api_web_contents needed that pointer
to old_host to call RemoveInputEventListener(), but I may be missing
some context & would appreciate review on this commit.
* Make electron/shell/browser/relauncher_win.cc use <winternl.h>
Xref: 4129135
Many internal Windows types are now available in winternl.h
so upstrem no longer defines the types themselves.
* Move ChildProcessHost* from content/common to content/browser
Xref: 4134795
* fixup! Make electron/shell/browser/relauncher_win.cc use <winternl.h>
winternl.h does not define the field we need, so clone the struct Chromium was using into unnamed namespace
* fixup! Move ChildProcessHost* from content/common to content/browser
chore: update #includes too
* chore: bump chromium in DEPS to 111.0.5532.0
* chore: sync patches/chromium/pepper_plugin_support.patch
Xref: 4133323
manually reync patch; no code changes
* chore: sync patches/chromium/mas_no_private_api.patch
Xref: 4143865
the content/common/pseudonymization_salt.cc patch is no longer needed
* chore: sync patches/chromium/mas_disable_remote_accessibility.patch
patch-fuzz update; no manual changes
* chore: sync patches/chromium/build_do_not_depend_on_packed_resource_integrity.patch
Xref: 4111725
manually reync patch; no code changes
* chore: sync patches/chromium/create_browser_v8_snapshot_file_name_fuse.patch
Xref: 4133323
manually reync patch; no code changes
* chore: sync patches/v8/fix_build_deprecated_attribute_for_older_msvc_versions.patch
Xref: 4127230
patch-fuzz update; no manual changes
* chore: rebuild patches
* fixup! Remove RenderViewHostChanged
Use PrimaryPageChanged()
* chore: remove unused method TabsUpdateFunction::OnExecuteCodeFinished()
Xref: 4133991
This private, already-unused function showed up as a FTBFS because it
took a base::ListValue parameter and ListValue was removed upstream.
* task posting v3: remove includes of runner handles and IWYU task runners
Xref: 4133323
* chore: lint
* chore: more lint
* fixup! task posting v3: remove includes of runner handles and IWYU task runners
macOS, too
* fixup! task posting v3: remove includes of runner handles and IWYU task runners
* chore: bump chromium in DEPS to 111.0.5534.0
* chore: sync patches/chromium/allow_new_privileges_in_unsandboxed_child_processes.patch
Xref: 4141862
patch-fuzz update; no manual changes
* chore: sync patches/chromium/logging_win32_only_create_a_console_if_logging_to_stderr.patch
Xref: 4153110
Sync to minor upstream changes. Add const correctness.
* chore: sync electron/patches/chromium/feat_configure_launch_options_for_service_process.patch
4141862
patch-fuzz update; no manual changes
* chore: patches/v8/fix_build_deprecated_attribute_for_older_msvc_versions.patch
sync 4147787
patch-fuzz update; no manual changes
* chore: update patches
* chore: bump chromium in DEPS to 111.0.5536.0
* chore: sync patches/chromium/allow_new_privileges_in_unsandboxed_child_processes.patch
Xref: 4141863
Sync with upstream code changes. Minor code golf for readability.
Note: upstream is laying groundwork for being able to work off of env vars
instead of switches. Doesn't affect us yet but worth being aware of.
> + // Environment variables could be supported in the future, but are not
> + // currently supported when launching with the zygote.
* chore: update patches/chromium/feat_expose_raw_response_headers_from_urlloader.patch
Xref: 4126836
patch-fuzz update; no manual changes
* chore: sync electron/patches/chromium/feat_configure_launch_options_for_service_process.patch
Xref: 4141863
manual sync
* chore: sync electron/patches/v8/fix_build_deprecated_attribute_for_older_msvc_versions.patch
4147788
fuzz-patch
* chore: rebuild patches
* chore: bump chromium in DEPS to 111.0.5538.0
* chore: bump chromium in DEPS to 111.0.5540.0
* chore: update patches
* Remove sdk_forward_declarations
4166680
* task posting v3: Remove task runner handles from codebase entirely
Refs 4150928
* Cleanup child_process_launcher_helper*
Refs 4141863
* fix: utilityprocess spec on macOS
* fix: build on windows
Refs 4141863
* chore: fix lint
* chore: bump chromium 111.0.5544.3
* chore: gen filenames.libcxx.gni
* Add check for Executable+Writable handles in renderer processes.
Refs 3774416
* fixup! Add check for Executable+Writable handles in renderer processes.
* 4143761: [110] Disable SwiftShader for WebGL on M1 Macs.
4143761
(cherry picked from commit 2f74db3c2139424c416f92d9169aeaa8a2f9c1ec)
* chore: bump chromium to 111.0.5555.0
* 56085: Remove hmac.h include from ssl.h.
https://boringssl-review.googlesource.com/c/boringssl/+/56085
* 4167020: Remove forwarding headers
4167020
* chore: bump chromium to 111.0.5559.0
* 4181044: Restrict WebCursor usage to RenderWidgetHostViewAura
4181044
* 4189437: views: rename ink_drop_host_view to ink_drop_host
4189437
* chore: bump chromium to 111.0.5560.0
* 4167016: win7dep: remove non aeroglass code
4167016
* fixup after rebase: Remove forwarding header
s 4167020
* 4125755: Reland "Reject getDisplayMedia calls without user activation"
4125755
* test: add workaround
* chore: update patches
* fix: alter coreModuleRegExp to prevent arm crash
* Revert "fix: alter coreModuleRegExp to prevent arm crash"
This reverts commit 7e50630c98137831a711c5abdbc8809e60cf1d73.
* 4218354: Disable the use of preserve_most on arm64 Windows
4218354
* chore: review changes
---------
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: deepak1556 <hop2deep@gmail.com>
Co-authored-by: John Kleinschmidt <jkleinsc@electronjs.org>
Co-authored-by: Shelley Vohr <shelley.vohr@gmail.com>
* refactor: change defined(MAS_BUILD) to IS_MAS_BUILD()
This is missing-definition safe and thus allows us to move the definition of this macro away from "all compilation targets" to "just the compilation targets that depend on this macro".
In turn this makes the rebuild time changing from mas <-> darwin only 80 seconds on my machine, instead of the 12-15 minutes it used to take. This will also allow us in the future to build both MAS and darwin on the same CI machine. Costing us ~2 minutes on one machine but saving us anywhere from 30 minutes to an hour of CI time on other parts of the matrix.
* build: always define IS_MAS_BUILD even on non-mac builds
* build: use extra_configs
