Context Navigation

← Previous Period
Next Period →

Timeline

Dec 7, 2021:

11:27 PM Changeset in webkit [286645] by mmaxfield@apple.com

2 edits in trunk/Source/ThirdParty/ANGLE

[MacCatalyst] Fix build after r286603
https://bugs.webkit.org/show_bug.cgi?id=233977

Unreviewed.

Previously, TARGET_OS_IOS and TARGET_OS_MACCATALYST were never on at the same time.
Now, they're on at the same time, and ANGLE doesn't expect that. This patch turns
off TARGET_OS_IOS when TARGET_OS_MACCATALYST is on, just to get ANGLE to build.
The long-term fix is to either update all the sites to use correct macros, or to
use the ANGLE_PLATFORM_* defines instead.

src/common/platform.h:

10:43 PM Changeset in webkit [286644] by youenn@apple.com

2 edits in trunk/Source/WebCore

Add logging to RealtimeOutgoingVideoSource creation
https://bugs.webkit.org/show_bug.cgi?id=233929

Reviewed by Eric Carlson.

This will allow to identify when trying to start sending video.
No change of behavior.

platform/mediastream/RealtimeOutgoingVideoSource.cpp:

(WebCore::RealtimeOutgoingVideoSource::RealtimeOutgoingVideoSource):
(WebCore::RealtimeOutgoingVideoSource::setSource):

10:34 PM Changeset in webkit [286643] by graouts@webkit.org

2 edits in trunk/Source/WebCore

[Model] Only expose APIs on the IDL interface when the feature is enabled
https://bugs.webkit.org/show_bug.cgi?id=233942

Reviewed by Dean Jackson.

Modules/model-element/HTMLModelElement.idl:

10:28 PM Changeset in webkit [286642] by graouts@webkit.org

5 edits in trunk/Source/WebCore

ActiveDOMObject::suspendIfNeeded() should not be called in the WebAnimation constructor
https://bugs.webkit.org/show_bug.cgi?id=233932

Reviewed by Chris Dumez.

It is not safe to call ActiveDOMObject::suspendIfNeeded() in a constructor. Since WebAnimation has several subclasses
(DeclarativeAnimation, and then CSSTransition and CSSAnimation deriving from DeclarativeAnimation) each with their own
create() methods, we add a new protected initialize() method which refactors code from the WebAnimation constructor
and is called from the two WebAnimation::create() methods.

DeclarativeAnimation already had an initialize method for common setup for the create() methods in CSSTransition and
CSS Animation, so we can simply call WebAnimation::initialize() from DeclarativeAnimation::initialize() and this guarantees
all create() methods for WebAnimation and all of its subclasses are correctly calling ActiveDOMObject::suspendIfNeeded().

animation/DeclarativeAnimation.cpp:

(WebCore::DeclarativeAnimation::initialize):

animation/DeclarativeAnimation.h:
animation/WebAnimation.cpp:

(WebCore::WebAnimation::create):
(WebCore::WebAnimation::initialize):
(WebCore::WebAnimation::WebAnimation):

animation/WebAnimation.h:

9:30 PM Changeset in webkit [286641] by Aditya Keerthi

17 edits in trunk

[iOS] Add initial support for find-in-page SPI
https://bugs.webkit.org/show_bug.cgi?id=233915
rdar://86140501

Reviewed by Wenson Hsieh.

Source/WebKit:

Expose new find-in-page SPI for use by clients.

Platform/spi/ios/UIKitSPI.h:
UIProcess/API/Cocoa/WKWebViewPrivate.h:
UIProcess/API/ios/WKWebViewIOS.mm:

(-[WKWebView selectedTextRange]):
(-[WKWebView offsetFromPosition:toPosition:inDocument:]):
(-[WKWebView performTextSearchWithQueryString:usingOptions:resultAggregator:]):
(-[WKWebView decorateFoundTextRange:inDocument:usingStyle:]):
(-[WKWebView clearAllDecoratedFoundText]):

UIProcess/WebPageProxy.cpp:

(WebKit::WebPageProxy::findRectsForStringMatches):
(WebKit::WebPageProxy::hideFindIndicator):

UIProcess/WebPageProxy.h:
UIProcess/ios/WKContentViewInteraction.h:
UIProcess/ios/WKContentViewInteraction.mm:

(-[WKContentView offsetFromPosition:toPosition:]):

The implementation here is needed to determine the relative ordering of
two search results.

(-[WKContentView performTextSearchWithQueryString:usingOptions:resultAggregator:]):

Call into the WebProcess to perform the search, and asynchronously
mark the search as complete.

A search result is represented as a WKFoundTextRange, and contains
two pieces of information.

A rect – so that clients can determine whether or not the result is visible.

An index – to allow for highlighting of a result and to enable determining relative ordering between two results.

(-[WKContentView decorateFoundTextRange:usingStyle:]):
(-[WKContentView clearAllDecoratedFoundText]):
(+[WKFoundTextRange foundTextRangeWithRect:index:]):
(-[WKFoundTextRange start]):
(-[WKFoundTextRange end]):
(-[WKFoundTextRange isEmpty]):
(+[WKFoundTextPosition textPositionWithIndex:]):

WebProcess/WebPage/FindController.cpp:

(WebKit::FindController::findString):

Prevent Editor from revealing the selection on all platforms, as the
existing call to didFindString is now responsible for revealing
selection on all platforms.

(WebKit::FindController::findRectsForStringMatches):

Gather rects for all search results and return them to the UIProcess.

(WebKit::FindController::indicateFindMatch):

This method was previously unused on iOS, and required some changes
to work correctly.

Wrap the call to select the current match with calls to
{will|did}FindString to account for iOS specific selection behavior.
See the existing comment in didFindString for more details.

No behavior change on other platforms, since willFindString is empty,
and didFindString now reveals the selection.

(WebKit::FindController::didFindString):

WebProcess/WebPage/FindController.h:
WebProcess/WebPage/WebPage.cpp:

(WebKit::WebPage::findRectsForStringMatches):
(WebKit::WebPage::hideFindIndicator):

WebProcess/WebPage/WebPage.h:
WebProcess/WebPage/WebPage.messages.in:

Added two new WebPage messages.

findRectsForStringMatches returns an vector of string match rects for use in the UIProcess.

Expose hideFindIndicator via IPC so that the UIProcess can hide the indicator without dismissing the overlay entirely.

Source/WTF:

wtf/PlatformHave.h:

Tools:

TestWebKitAPI/Tests/WebKitCocoa/FindInPage.mm:

(-[TestSearchAggregator initWithCompletionHandler:]):
(-[TestSearchAggregator foundRange:forSearchString:inDocument:]):
(-[TestSearchAggregator finishedSearching]):
(TEST):

Added an API test to verify that search results are found correctly.

8:24 PM Changeset in webkit [286640] by Wenson Hsieh

33 edits
9 adds in trunk

Add support for navigator.requestCookieConsent() behind a disabled feature flag
https://bugs.webkit.org/show_bug.cgi?id=233890
rdar://86117718

Reviewed by Darin Adler.

Source/WebCore:

Add plumbing to support a new (internal-for-the-time-being) navigator API, requestCookieConsent(). This
returns a promise indicating whether or not the user (or the UA, on behalf of the user) consents to usage of
cookies that are not strictly necessary for page performance.

Paired with this web-facing API, we also introduce corresponding private WKUIDelegate methods that are invoked
as a result of calling navigator.requestCookieConsent();, which a WebKit embedder may implement in order to
present UI or otherwise respond appropriately to the request. See WebKit ChangeLog for more details.

Tests: CookieConsent.BasicDecisionHandler

CookieConsent.NotEnabled
CookieConsent.ThrowsExceptionWithoutDelegate

CMakeLists.txt:
DerivedSources-input.xcfilelist:
DerivedSources-output.xcfilelist:
DerivedSources.make:
Headers.cmake:
Modules/cookie-consent/CookieConsentDecisionResult.h: Added.
Modules/cookie-consent/Navigator+CookieConsent.idl: Added.
Modules/cookie-consent/NavigatorCookieConsent.cpp: Added.

Add a new partial interface on Navigator that implements the new API by calling into the chrome client hook (see
WebKit changes for more details). Note that in the case where the consent delegate method is not implemented by
the client, we will reject the promise with a NotSupportedError.

(WebCore::NavigatorCookieConsent::requestCookieConsent):
(WebCore::NavigatorCookieConsent::from):

Modules/cookie-consent/NavigatorCookieConsent.h: Added.

(WebCore::RequestCookieConsentOptions::RequestCookieConsentOptions):

Modules/cookie-consent/RequestCookieConsentOptions.h: Added.
Modules/cookie-consent/RequestCookieConsentOptions.idl: Added.
Modules/filesystemaccess/WorkerFileSystemStorageConnection.cpp:
Modules/filesystemaccess/WorkerFileSystemStorageConnection.h:

Add some missing includes to fix the unified source build on Windows.

Sources.txt:
WebCore.xcodeproj/project.pbxproj:
loader/EmptyClients.cpp:

(WebCore::EmptyChromeClient::requestCookieConsent):

loader/EmptyClients.h:
page/ChromeClient.h:

Source/WebKit:

Add plumbing for an async requestCookieConsent() method on ChromeClient that plumbs through the WebKit2 client
layer, into a new private (for the time being) UI delegate method. If the new delegate method is not implemented
(or there is no UI delegate at all), we immediately invoke the decision handler with a result of NotSupported,
allowing us to reject the promise with a NotSupportedError.

See WebCore ChangeLog for more details.

UIProcess/API/APIUIClient.h:

(API::UIClient::requestCookieConsent):

UIProcess/API/Cocoa/WKUIDelegatePrivate.h:
UIProcess/Cocoa/UIDelegate.h:
UIProcess/Cocoa/UIDelegate.mm:

(WebKit::UIDelegate::setDelegate):
(WebKit::UIDelegate::UIClient::requestCookieConsent):

UIProcess/WebPageProxy.cpp:

(WebKit::WebPageProxy::requestCookieConsent):

UIProcess/WebPageProxy.h:
UIProcess/WebPageProxy.messages.in:
WebProcess/WebCoreSupport/WebChromeClient.cpp:

(WebKit::WebChromeClient::requestCookieConsent):

WebProcess/WebCoreSupport/WebChromeClient.h:

Source/WebKitLegacy/mac:

Add an implementation stub for the new chrome client hook, requestCookieConsent().

WebCoreSupport/WebChromeClient.h:
WebCoreSupport/WebChromeClient.mm:

(WebChromeClient::requestCookieConsent):

Source/WebKitLegacy/win:

Add an implementation stub for the new chrome client hook, requestCookieConsent().

WebCoreSupport/WebChromeClient.cpp:

(WebChromeClient::requestCookieConsent):

WebCoreSupport/WebChromeClient.h:

Source/WTF:

Add the new feature flag (off by default). Note that this is also only an internal feature flag for the time
being (rather than an experimental feature flag), such that it cannot (yet) be turned on by non-Apple internal
users.

Scripts/Preferences/WebPreferencesInternal.yaml:

Tools:

Add several new API tests that enable the feature flag and exercise the API.

TestWebKitAPI/TestWebKitAPI.xcodeproj/project.pbxproj:
TestWebKitAPI/Tests/WebKit/CookieConsent.mm: Added.

(-[CookieConsentDelegate _webView:requestCookieConsentWithMoreInfoHandler:decisionHandler:]):
(TestWebKitAPI::createWebViewForTestingCookieConsent):
(TestWebKitAPI::checkForString):
(TestWebKitAPI::TEST):

TestWebKitAPI/Tests/WebKit/cookie-consent-basic.html: Added.

7:59 PM Changeset in webkit [286639] by sbarati@apple.com

6 edits
3 adds in trunk

TypedArray prototype set should go down the fast path when using non clamped integer types of the same byte size
https://bugs.webkit.org/show_bug.cgi?id=233905

Reviewed by Keith Miller.

JSTests:

microbenchmarks/typed-array-prototype-set.js: Added.
microbenchmarks/typed-array-prototype-set-order.js: Added.
stress/typed-array-prototype-set.js: Added.

Source/JavaScriptCore:

We can use memmove in this scenario because the bitpattern of the
data between the signed and unsigned values will be the same.

This patch also fixes a bug where we were looking at the wrong
pointer when determining to do a forward or backwards loop in
our memmove. We were looking at the vector instead of vector+offset.

runtime/JSGenericTypedArrayViewInlines.h:

(JSC::JSGenericTypedArrayView<Adaptor>::set):

LayoutTests:

js/script-tests/typedarray-set-overlapping-elements-of-same-size.js:
js/typedarray-set-overlapping-elements-of-same-size-expected.txt:

6:49 PM Changeset in webkit [286638] by ysuzuki@apple.com

3 edits in trunk/Source/ThirdParty/ANGLE

Unreviewed, these files are modified after building ANGLE
https://bugs.webkit.org/show_bug.cgi?id=220896
<rdar://problem/73539682>

src/libANGLE/renderer/metal/shaders/mtl_default_shaders_src_autogen.inc:
src/libANGLE/renderer/metal/shaders/mtl_default_shaders_src_autogen.metal:

6:11 PM Changeset in webkit [286637] by Cameron McCormack

4 edits in trunk/Source/WebCore

Remove expandAroundIdeographs static variables
https://bugs.webkit.org/show_bug.cgi?id=233750

Reviewed by Myles Maxfield.

With OffscreenCanvas, we can call into
ComplexTextController::adjustGlyphsAndAdvances,
FontCascade::expansionOpportunityCountInternal, and
WidthIterator::calculateAdditionalWidth from worker threads.
These all have a static variable variable with an initializer
that calls FontCascade::canExpandAroundIdeographsInComplexText,
and such initializers are not safe under -fno-threadsafe-statics.

canExpandAroundIdeographsInComplexText is a simple enough function in
all ports (it just returns a constant bool) that it's not worth caching
the result in a static.

platform/graphics/ComplexTextController.cpp:

(WebCore::ComplexTextController::adjustGlyphsAndAdvances):

platform/graphics/FontCascade.cpp:

(WebCore::FontCascade::expansionOpportunityCountInternal):

platform/graphics/WidthIterator.cpp:

(WebCore::WidthIterator::calculateAdditionalWidth const):

6:09 PM Changeset in webkit [286636] by Lauro Moura

2 edits in trunk/Source/ThirdParty/ANGLE

Avoid noisy Wcomment warnings after ANGLE update in r286603
https://bugs.webkit.org/show_bug.cgi?id=233964

Unreviewed warning removal.

src/compiler/translator/tree_util/IntermTraverse.h:

5:43 PM Changeset in webkit [286635] by Ross Kirsling

8 edits in trunk/Source/JavaScriptCore

[JSC] Add LLInt IC for try_get_by_id of own cacheable value
https://bugs.webkit.org/show_bug.cgi?id=233830

Reviewed by Yusuke Suzuki.

This patch adds an LLInt IC for the "own cacheable value" path of try_get_by_id;
this is the simplest case and basically the same as get_by_id_direct.

Performance is neutral with JIT enabled as well as on current uses of try_get_by_id in JSC
(e.g. hasObservableSideEffectsForRegexpSplit), but microbenchmarks of try_get_by_id itself see a 2x speedup:

Before After

try-get-by-id-polymorphic 123.8361+-0.4562 ^{61.7586+-0.3770} definitely 2.0052x faster
try-get-by-id-basic 124.4437+-0.6091 ^{61.0340+-0.1924} definitely 2.0389x faster

<geometric> 124.1207+-0.3130 ^{61.3865+-0.2019} definitely 2.0220x faster

bytecode/BytecodeList.rb:
bytecode/CodeBlock.cpp:
bytecode/GetByStatus.cpp:
llint/LLIntSlowPaths.cpp:
llint/LowLevelInterpreter.asm:
llint/LowLevelInterpreter32_64.asm:
llint/LowLevelInterpreter64.asm:

5:35 PM Changeset in webkit [286634] by Cameron McCormack

3 edits in trunk/Source/WebCore

Move shouldAutoActivateFontIfNeeded knownFamilies cache to FontCache
https://bugs.webkit.org/show_bug.cgi?id=233749

Reviewed by Myles Maxfield.

With OffscreenCanvas, we can call shouldAutoActivateFontIfNeeded on a
worker thread, and the knownFamilies HashSet is not thread safe. Also,
the AtomStrings it stores are thread-specific. Move it to FontCache, so
each thread has its own set.

platform/graphics/FontCache.h:
platform/graphics/cocoa/FontCacheCoreText.cpp:

(WebCore::FontCache::shouldAutoActivateFontIfNeeded):
(WebCore::shouldAutoActivateFontIfNeeded): Deleted.

5:32 PM Changeset in webkit [286633] by Yousuke.Kimoto@sony.com

2 edits in trunk/Tools

[webkitcorepy] Fail to install python module when a working directory is not on C drive on Windows
https://bugs.webkit.org/show_bug.cgi?id=233936

Reviewed by Jonathan Bedard.

On Windows, autoinstall.py installs python modules to C drive even if
your working directory is on other drive (e.g. D drive). To install
python modules correctly on Windows, a drive letter where your working
directory exists needs to be given to "--root" option of "setup.py"
in autoinstall.py.

Scripts/libraries/webkitcorepy/webkitcorepy/autoinstall.py:

(Package.install): Added root_location to detect a right drive letter to isntall modules on Windows.

4:55 PM Changeset in webkit [286632] by Devin Rousso

40 edits in trunk/Source

Add a DOMPasteAccessCategory to control which pasteboard the WebProcess is granted access to when pasting
https://bugs.webkit.org/show_bug.cgi?id=233939

Reviewed by Wenson Hsieh.

Implementing -pasteFont: <https://webkit.org/b/191379> requires that we read from the font
pasteboard instead of the general pasteboard. In order to allow web content to trigger this
(e.g. document.execCommand("PasteFont"))) we must have a way for the WebProcess to signal
that it wishes to read from the font pasteboard instead of the general pasteboard. As such,
create and pass along a DOMPasteAccessCategory to requestDOMPasteAccess.

No change in behavior.

Source/WebCore:

dom/DOMPasteAccess.h:

Create DOMPasteAccessCategory.

page/EditorClient.h:
page/Frame.h:
page/Frame.cpp:

(WebCore::Frame::requestDOMPasteAccess):

loader/EmptyClients.cpp:

(WebCore::EmptyEditorClient::requestDOMPasteAccess):
Pass along the DOMPasteAccessCategory.

Source/WebKit:

Scripts/webkit/messages.py:

(headers_for_type):
Indicate which header is associated with DOMPasteAccessCategory.

UIProcess/Cocoa/WebViewImpl.h:
UIProcess/Cocoa/WebViewImpl.mm:

(-[WKDOMPasteMenuDelegate initWithWebViewImpl:pasteAccessCategory:]): Added.
(-[WKDOMPasteMenuDelegate menuDidClose:]):
(-[WKDOMPasteMenuDelegate _web_grantDOMPasteAccess]): Added.
(WebKit::WebViewImpl::handleProcessSwapOrExit):
(WebKit::pasteboardNameForAccessCategory): Added.
(WebKit::pasteboardForAccessCategory): Added.
(WebKit::WebViewImpl::requestDOMPasteAccess):
(WebKit::WebViewImpl::handleDOMPasteRequestForCategoryWithResult):
(WebKit::WebViewImpl::hideDOMPasteMenuWithResult): Added.
(-[WKDOMPasteMenuDelegate initWithWebViewImpl:]): Deleted.
(WebKit::WebViewImpl::handleDOMPasteRequestWithResult): Deleted.

UIProcess/API/mac/WKView.mm:

(-[WKView _web_grantDOMPasteAccess]): Deleted.

UIProcess/API/mac/WKWebViewMac.mm:

(-[WKWebView _web_grantDOMPasteAccess]): Deleted.
Pass the DOMPasteAccessCategory to the WKDOMPasteMenuDelegate and use it as the target
of the NSMenuItem so that we can pass along the DOMPasteAccessCategory in the action.

UIProcess/ios/WKContentViewInteraction.h:
UIProcess/ios/WKContentViewInteraction.mm:

(-[WKContentView cleanUpInteraction]):
(-[WKContentView resignFirstResponderForWebView]):
(-[WKContentView _webTouchEventsRecognized:]):
(-[WKContentView _willHideMenu:]):
(-[WKContentView pasteForWebView:]):
(pasteboardNameForAccessCategory): Added.
(pasteboardForAccessCategory): Added.
(-[WKContentView _hidePasteMenuWithResult:]): Added.
(-[WKContentView _handleDOMPasteRequestForCategory:result:]): Added.
(-[WKContentView _willPerformAction:sender:]):
(-[WKContentView _didPerformAction:sender:]):
(-[WKContentView handleKeyWebEvent:withCompletionHandler:]):
(-[WKContentView _requestDOMPasteAccessForCategory:elementRect:originIdentifier:completionHandler:]): Added.
(-[WKContentView _handleDOMPasteRequestWithResult:]): Deleted.
(-[WKContentView _requestDOMPasteAccessWithElementRect:originIdentifier:completionHandler:]): Deleted.

UIProcess/PageClient.h:
UIProcess/API/gtk/PageClientImpl.h:
UIProcess/API/gtk/PageClientImpl.cpp: