This is a speculative fix for a crash we are seeing in `menuDidClose`. We
can't repro the crash but the traces have it happening in this method
and just by reading through the impl the only part that jumps out as
Might Crash is this `model_` call. Other methods in the menu controller
check `model_` before using it so it probably makes sense to do that here
as well.
* add notice to files being removed
* add gin version of function_template.h
* rename callback.h to avoid confliction
* add gin version of callback_converter
* add gin converter for OnceCallback
* remove callback_converter_gin_adapter.h
* remove gin_util.h and gin_utils.h
* fix lint warning
* add helper for setting methods
* move Destroyable utilities out of native_mate
* do not set "destroy" in ObjectTemplateBuilder
* remove ObjectTemplateBuilder::MakeDestroyable
* do not pollute gin namespace
* add more comments
* remove hack of Arguments
On multi-monitor setups where the monitors are not all origined at 0 on
the Y coordinate (E.g. vertical stacked monitors) the maximize
calculation was incorrect as it assumed top was "0". This instead
adjusts the math to calculate the correct top value.
* refactor: don't walk maps twice to remove elements
* refactor: don't walk maps twice to read elements
* refactor: don't walk maps twice to insert elements
* refactor: don't walk map 3x on UvTaskRunner timeout
* refactor: more don't-walk-maps-twice cleanup
* fixup! refactor: don't walk maps twice to insert elements
* refactor: don't walk containers twice when erasing
* refactor: omit excess lookups in RemoteObjectFreer
* fix: remove WM_GETMINMAXINFO workaround since it's no longer needed
* fix: handle WM_GETMINMAXINFO ourselves
* fix: remove part of the chromium WM_GETMINMAXINFO handler
Sometimes it's necessary to convey more information about the window to screen reader users only (simply putting everything to the window title might be unnecessarily noisy).
For example, Chromium uses that technique to tell screen reader users that the window is in incognito mode (the incognito window looks differently and doesn't have «incognito» in the title, but for blind users the screen reader will announce that it's incognito).
* no need to get WebContents for URLLoaderFactory
* consult embedder for network_factory created in net module
* set disable_web_security to false
* re-enable webRequest tests in net module
* fix: i18n of gtk msgbox buttons
similar to #19756 (12df0e8) but for messageboxes
* refactor: DRY the gtk+ button mnemonics
* fix: don't compile gtk_util on non-Linux platforms
rename from `gtk_util.[cc,h]` to `util_gtk.[cc,h]` so that it gets
picked up by the `extra_source_filters` rule in `BUILD.gn`.
* fix: make linter happy
It really shows that I cannot build locally atm... :P
* chore: bump chromium in DEPS to f3bf493731e868e1f5f48e7e1adc02ea5eccfbbd
* chore: bump chromium in DEPS to 4db0c87d4aa6f27ffa0b5fc77d20e10047962484
* chore: bump chromium in DEPS to d933a504c264dc8fe85267f47aef3588531875b5
* chore: bump chromium in DEPS to 34afdb68980f581ae911b85b727bc17e126cf5f9
* update disable-redraw-lock.patch
1600387
* update desktop_media_list.patch
1729156
* update notification_provenance.patch
1742779
* update printing.patch
1646772
* update verbose_generate_bpad_syms.patch
1745986
* update patch metadata
* remove printing_compositor manifests
1742734
* update for URLLoaderFactoryType enum
1754716
* remove gin string16 converter
1750093
* ClearCompositorFrame() has been removed
1746301
* message_loop -> message_loop_current
1738552
* include resource_response header
* pdf compositor no longer uses service manager
1742734
* chore: bump chromium in DEPS to 00d5933101d8d8dc9546eadbe7ee1b41077e6db1
* pane focus fns aren't pure virtual anymore
1708767
* fix: make std::hash value-non-const
broken by 1711202
* update swiftshader in zip_manifests
https://swiftshader-review.googlesource.com/c/SwiftShader/+/34911
* address feedback from @deepak1556
* don't enable kLegacyWindowsDWriteFontFallback
1753006
* chore: bump chromium in DEPS to 84497314005e1968da06804f8fde539d9872310e
* update printing.patch
remove bottom diff owing to 1678182 and update for 1678182
* convert CookieChangeListener to new Mojo types
1753371
* rename ui::ClipboardType -> ui::ClipboardBuffer
1758730
* logging::LoggingSettings log_file -> log_file_path
1699477
* roll DEPS to latest lkgr
* fix: override GetFontLookupTableCacheDir()
When Chromium goes to use its fallback font table creation code paths,
it creates the cache directory it uses by calling
GetFontLookupTableCacheDir() with a path that doesn't exist in Electron.
To ensure that a legitimate file path is created, we need to override it
with Electron's DIR_USER_DATA so it doesn't use chrome::DIR_USER_DATA.
* chore: bump chromium in DEPS to 6758a0879931bc4df630a80a36c82d7855ae3155
* update pthread_fchdir patch
1759149
* update printing patch
* update cookie usage and fn signatures
1758437
* chore: bump chromium in DEPS to bdaca97e1cc27fb977e56f30f74cdb906da9527e
* remove fix_make_std_hash_value-non-const.patch
1762335
* Convert enum to enum class for FocusManager::FocusChangeReason
1767281
* roll DEPS to latest lkgr
* update dom_storage_limits.patch
1767556
* refactor: make util::Promise type safe when chaining in native
* fixup! refactor: make util::Promise type safe when chaining in native
* chore: remove spare brackets
* feat: enable picture in picture mode for video tags
* test: add test to verify picture in picture support
* lint: fix indent
* fix: clean up after rebase
* test: update test with 16:9 test video
* fix: .paches after rebase
