Upgrade WebKit 2025/12/07 #118
Draft
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
https://bugs.webkit.org/show_bug.cgi?id=302851 Reviewed by Ryosuke Niwa. Since attachShadow() no longer looks at the host's custom element registry, but rather the host document registry, we should do something similar when serializing. WPT PR: web-platform-tests/wpt#56419 Canonical link: https://commits.webkit.org/303841@main
rdar://165712922 https://bugs.webkit.org/show_bug.cgi?id=303410 Unreviewed test gardening Applying a new rebaseline for viewport tests in iOS. * LayoutTests/platform/ios/fast/viewport/ios/responsive-viewport-with-minimum-width-after-changing-view-scale-expected.txt: * LayoutTests/platform/ios/fast/viewport/ios/shrink-to-fit-for-page-without-viewport-meta-expected.txt: * LayoutTests/platform/ios/fast/viewport/ios/viewport-fit-auto-expected.txt: Added. * LayoutTests/platform/ios/fast/viewport/ios/viewport-fit-contain-expected.txt: Added. * LayoutTests/platform/ios/fast/viewport/ios/viewport-fit-cover-expected.txt: Added. * LayoutTests/platform/ios/fast/viewport/ios/width-is-device-width-expected.txt: * LayoutTests/platform/ios/fast/viewport/ios/width-is-device-width-overflowing-body-overflow-hidden-expected.txt: * LayoutTests/platform/ios/fast/viewport/ios/width-is-device-width-overflowing-body-overflow-hidden-tall-expected.txt: * LayoutTests/platform/ios/fast/viewport/ios/width-is-device-width-overflowing-expected.txt: * LayoutTests/platform/ios/fast/viewport/ios/width-is-device-width-overflowing-no-shrink-to-fit-expected.txt: Canonical link: https://commits.webkit.org/303842@main
…tems. https://bugs.webkit.org/show_bug.cgi?id=303441 rdar://165728471 Reviewed by Brandon Stewart and Alan Baradlay. The automatic sizes of grid items is defined in: https://drafts.csswg.org/css-grid-1/#grid-item-sizing This patch adds some initial code to start supporting the automatic *inline* sizes of grid items that we can build upon. In particular, this is for content in which the grid item: 1.) is non-replaced element 2.) does not have a preferred aspect ratio 3.) has non-auto margins 4.) has an alignment position of normal This ends up placing us into the "Fill grid area," portion of this section of the spec. "Fill grid area," seems to refer to "the grid item is sized as for align-self: stretch," which after following a couple of spec links leads to: https://www.w3.org/TR/css-align-3/#justify-self-property This is the core logic that is implemented as part of usedInlineSizeForGridItem. * Source/WebCore/layout/formattingContexts/grid/GridLayout.cpp: (WebCore::Layout::GridLayout::performInlineAxisSelfAlignment): Since "Normal," is handled by grid item sizing there is nothing to do here as part of alignment. * Source/WebCore/layout/formattingContexts/grid/GridLayoutUtils.cpp: (WebCore::Layout::GridLayoutUtils::usedInlineSizeForGridItem): This is the core of the logic that is written to implement the aforementioned portion of the spec. Basically we need to determine the total size of the columns the grid item spans and then stretch it to fit within that size while still respecting the min and max sizes specified by style. Note that there was was small change made when for the fixed inline size case. Now we add in the border and padding within this function instead of it being done on top of what we returned from this function. This is because we need the border and padding in order to properly stretch the grid item. Canonical link: https://commits.webkit.org/303843@main
https://bugs.webkit.org/show_bug.cgi?id=299999 rdar://162244690 Reviewed by Tim Nguyen. This patch aligns WebKit with Gecko / Firefox and Blink / Chromium. This patch fixes an issue in serialization where in the past, we used to not serialize `border` property and were not canonical. Now it aligns with other browsers and returns canonical reference. CSS spec: https://drafts.csswg.org/cssom/#serialize-a-css-declaration-block * LayoutTests/fast/css/remove-shorthand-expected.txt: * LayoutTests/imported/w3c/web-platform-tests/css/css-sizing/stretch/block-height-008-expected.txt: * LayoutTests/imported/w3c/web-platform-tests/css/cssom/shorthand-values-expected.txt: * LayoutTests/platform/glib/imported/w3c/web-platform-tests/css/css-sizing/stretch/block-height-008-expected.txt: * Source/WebCore/css/StyleProperties.cpp: (WebCore::canUseShorthandForLonghand): Canonical link: https://commits.webkit.org/303844@main
…-attribute-basic.html is constant failure https://bugs.webkit.org/show_bug.cgi?id=302741 rdar://164998514 Unreviewed test gardening Removing extra lines from the expectation file. * LayoutTests/platform/ios/imported/w3c/web-platform-tests/html/semantics/popovers/popover-attribute-basic-expected.txt: Canonical link: https://commits.webkit.org/303845@main
…View` https://bugs.webkit.org/show_bug.cgi?id=303431 rdar://165724968 Reviewed by Wenson Hsieh. `-[UIScreen mainScreen]` was deprecated in iOS 26. Work towards removing its use in WebKit. * Source/WebKit/UIProcess/Inspector/ios/WKInspectorHighlightView.mm: (-[WKInspectorHighlightView update:scale:frame:]): * Source/WebKit/UIProcess/ios/WKContentView.mm: (-[WKContentView _showInspectorHighlight:]): `-intrinsicDeviceScaleFactor` gets the display scale from the view's trait collection and falls back to using `UIScreen` if the value is unspecified. Canonical link: https://commits.webkit.org/303846@main
https://bugs.webkit.org/show_bug.cgi?id=303442 Reviewed by Jonathan Bedard. We don't need to enumerate all branches and then check whether the default branch name exists in each source remote: we can instead just rely on merge-base to fail when the ref doesn't exist. With a very large number of branches (mostly from numerous remotes), this is a 3.9x speedup for `git-webkit pickable origin/main~10..origin/main`. * Tools/Scripts/libraries/webkitscmpy/webkitscmpy/local/git.py: (Git._is_on_default_branch): Canonical link: https://commits.webkit.org/303847@main
https://bugs.webkit.org/show_bug.cgi?id=303439 rdar://165728567 Reviewed by Darin Adler. libc++ marks std::make_unique's return value as [[nodiscard]]. Apply the same to WTF::makeUnique and variants to warn against discarding return values. LLVM/libc++ docs: https://libcxx.llvm.org/CodingGuidelines.html#apply-nodiscard-where-relevant * Source/WTF/wtf/StdLibExtras.h: (WTF::makeUnique): (WTF::makeUniqueWithoutRefCountedCheck): (WTF::makeUniqueWithoutFastMallocCheck): * Source/WTF/wtf/UniqueArray.h: (WTF::makeUniqueArray): * Source/WTF/wtf/UniqueRef.h: (WTF::makeUniqueRefWithoutFastMallocCheck): (WTF::makeUniqueRefWithoutRefCountedCheck): (WTF::makeUniqueRef): Canonical link: https://commits.webkit.org/303848@main
…itmapTexture https://bugs.webkit.org/show_bug.cgi?id=303312 Reviewed by Carlos Garcia Campos. Introduce m_allocatedSize to BitmapTexture to track the actual allocated buffer size, which may differ from the logical m_size when using specific tiled texture formats that require e.g. 64x64 alignment. When rendering textures with allocated size differences, adjust UV coordinates via drawTextureWithPhysicalSize() to sample only the logical region, preventing rendering artifacts from the padding area. Add a new ClampUVBounds shader option that clamps texture coordinates to stay within the logical texture region. This prevents GPU bilinear filtering from sampling texels in the padding area when the allocated texture is larger than the logical size, which would otherwise cause visible artifacts at texture edges. * Source/WebCore/platform/graphics/gbm/MemoryMappedGPUBuffer.cpp: (WebCore::MemoryMappedGPUBuffer::allocatedSize const): * Source/WebCore/platform/graphics/gbm/MemoryMappedGPUBuffer.h: * Source/WebCore/platform/graphics/texmap/BitmapTexture.cpp: (WebCore::BitmapTexture::allocatedSize const): * Source/WebCore/platform/graphics/texmap/BitmapTexture.h: * Source/WebCore/platform/graphics/texmap/TextureMapper.cpp: (WebCore::TextureMapper::drawTexture): (WebCore::TextureMapper::drawTextureWithPhysicalSize): * Source/WebCore/platform/graphics/texmap/TextureMapper.h: * Source/WebCore/platform/graphics/texmap/TextureMapperShaderProgram.cpp: (WebCore::TextureMapperShaderProgram::create): * Source/WebCore/platform/graphics/texmap/TextureMapperShaderProgram.h: Canonical link: https://commits.webkit.org/303849@main
https://bugs.webkit.org/show_bug.cgi?id=303314 Reviewed by Alex Christensen. The tag deduping test cases are failing for multiple reasons: * Invalid tag rules are causing the entire Speculation Rules script to fail, rather than just the relevant rule. * `null` rules are not being sorted properly, as they are treated as strings. * Consecutive speculationrules scripts are being considered individually, rather than wait for the next microtask checkpoint. This PR fixes all of the above, and with it most of the test cases of the relevant test. No new tests, but this progresses a currently failing test. * LayoutTests/imported/w3c/web-platform-tests/speculation-rules/speculation-tags/deduped-and-sorted-tags.https_type=prefetch-expected.txt: Progression. * LayoutTests/imported/w3c/web-platform-tests/speculation-rules/speculation-tags/prefetch-eagerness-pointer-down.https-expected.txt: Progression. * LayoutTests/imported/w3c/web-platform-tests/speculation-rules/speculation-tags/prefetch-eagerness-pointer-hover.https-expected.txt: Expectation change. * Source/WebCore/dom/Document.cpp: (WebCore::Document::considerSpeculationRules): Align to spec. Queue a microtask. (WebCore::Document::processSpeculationRules): Moved the actual logic here. Aligned to spec when it comes to tag deduping. * Source/WebCore/dom/Document.h: Add m_speculationRulesConsiderationScheduled. * Source/WebCore/loader/DocumentPrefetcher.cpp: (WebCore::makePrefetchRequest): Handle null string as a tag. * Source/WebCore/loader/SpeculationRules.cpp: (WebCore::parseSingleRule): Dedupe tag if identical to ruleset tag. Append null instead of "null". (WebCore::parseRules): Don't terminate speculation rules processing due to an invalid rule. (WebCore::SpeculationRules::parseSpeculationRules): Add spec comments. Canonical link: https://commits.webkit.org/303850@main
https://bugs.webkit.org/show_bug.cgi?id=302986 rdar://165243794 Reviewed by Sihui Liu and Alex Christensen. This patch fixes issues with the `UITextSearching` protocol in site isolation. It fixes `replaceFoundTextInRange`, `decorateFoundTextRange`, `scrollRangeToVisible`, `clearAllDecoratedFoundText`, and `requestRectForFoundTextRange` in site isolation. The main problem is that a client can search for string in one WebView and then highlight or replace or scroll in another that is hosting the same content. We address this issue by storing the path from the mainFrame to the frame in which the match was found on each WebFoundTextRange. This way, we can traverse the frame tree for each WebFoundTextRange and find, positionally, which frame the range was found in. This means we do not have to store the frameIdentifier, which will be different between WebViews. * Source/WebCore/page/Frame.cpp: (WebCore::Frame::indexInFrameTreeSiblings const): (WebCore::Frame::pathToFrame const): * Source/WebCore/page/Frame.h: * Source/WebCore/page/Page.cpp: (WebCore::Page::findFrameByPath const): * Source/WebCore/page/Page.h: * Source/WebKit/Shared/WebFoundTextRange.cpp: (WebKit::WebFoundTextRange::operator== const): (WebKit::operator<<): * Source/WebKit/Shared/WebFoundTextRange.h: (WTF::HashTraits<WebKit::WebFoundTextRange>::constructDeletedValue): (WTF::HashTraits<WebKit::WebFoundTextRange>::isDeletedValue): * Source/WebKit/Shared/WebFoundTextRange.serialization.in: * Source/WebKit/UIProcess/WebFrameProxy.cpp: (WebKit::WebFrameProxy::childFrame const): * Source/WebKit/UIProcess/WebFrameProxy.h: * Source/WebKit/UIProcess/WebPageProxy.cpp: (WebKit::WebPageProxy::replaceFoundTextRangeWithString): (WebKit::WebPageProxy::scrollTextRangeToVisible): (WebKit::WebPageProxy::clearAllDecoratedFoundText): (WebKit::WebPageProxy::didBeginTextSearchOperation): (WebKit::WebPageProxy::requestRectForFoundTextRange): * Source/WebKit/UIProcess/ios/WKContentViewInteraction.mm: (+[WKFoundTextRange foundTextRangeWithWebFoundTextRange:]): (-[WKFoundTextRange dealloc]): (-[WKFoundDOMTextRange webFoundTextRange]): (-[WKFoundPDFTextRange webFoundTextRange]): * Source/WebKit/WebProcess/WebPage/WebFoundTextRangeController.cpp: (WebKit::WebFoundTextRangeController::findTextRangesForStringMatches): (WebKit::WebFoundTextRangeController::rectsForTextMatchesInRect): (WebKit::WebFoundTextRangeController::frameForFoundTextRange const): * Tools/TestRunnerShared/spi/UIKitSPIForTesting.h: * Tools/TestWebKitAPI/Tests/WebKitCocoa/FindInPage.mm: (TEST(WebKit, FindAndReplace)): (swizzledIsEmbeddedScreen): Deleted. (-[TestScrollViewDelegate init]): Deleted. (-[TestScrollViewDelegate scrollViewDidEndScrollingAnimation:]): Deleted. (-[TestFindDelegate setDidAddLayerForFindOverlayHandler:]): Deleted. (-[TestFindDelegate didAddLayerForFindOverlayHandler]): Deleted. (-[TestFindDelegate setDidRemoveLayerForFindOverlayHandler:]): Deleted. (-[TestFindDelegate didRemoveLayerForFindOverlayHandler]): Deleted. (-[TestFindDelegate _webView:didAddLayerForFindOverlay:]): Deleted. (-[TestFindDelegate _webViewDidRemoveLayerForFindOverlay:]): Deleted. * Tools/TestWebKitAPI/Tests/WebKitCocoa/FindInPageUtilities.h: * Tools/TestWebKitAPI/Tests/WebKitCocoa/FindInPageUtilities.mm: (swizzledIsEmbeddedScreen): (-[TestScrollViewDelegate init]): (-[TestScrollViewDelegate scrollViewDidEndScrollingAnimation:]): (-[TestFindDelegate setDidAddLayerForFindOverlayHandler:]): (-[TestFindDelegate didAddLayerForFindOverlayHandler]): (-[TestFindDelegate setDidRemoveLayerForFindOverlayHandler:]): (-[TestFindDelegate didRemoveLayerForFindOverlayHandler]): (-[TestFindDelegate _webView:didAddLayerForFindOverlay:]): (-[TestFindDelegate _webViewDidRemoveLayerForFindOverlay:]): * Tools/TestWebKitAPI/Tests/WebKitCocoa/SiteIsolation.mm: (TestWebKitAPI::(SiteIsolation, FindStringInFrameAndReplaceIOS)): (TestWebKitAPI::(SiteIsolation, DecorateFoundTextRangeIOS)): (TestWebKitAPI::(SiteIsolation, ScrollTextRangeToVisibleIOS)): (TestWebKitAPI::(SiteIsolation, ClearAllDecoratedFoundTextIOS)): (TestWebKitAPI::(SiteIsolation, RequestRectForFoundTextRangeIOS)): Canonical link: https://commits.webkit.org/303851@main
…utes instead of sending across an opaque NSDictionary https://bugs.webkit.org/show_bug.cgi?id=301693 rdar://163710069 Reviewed by Alex Christensen. Replaces FontPlatformDataAttributes attributes dictionary with the existing FontPlatformSerializedAttributes. * Source/WebCore/platform/graphics/FontPlatformData.h: * Source/WebCore/platform/graphics/coretext/FontPlatformDataCoreText.cpp: (WebCore::FontPlatformDataAttributes::serializableAttributes const): (WebCore::FontPlatformDataAttributes::FontPlatformDataAttributes): * Source/WebKit/Scripts/webkit/opaque_ipc_types.tracking.in: * Source/WebKit/Shared/WebCoreFont.serialization.in: Canonical link: https://commits.webkit.org/303852@main
https://bugs.webkit.org/show_bug.cgi?id=303487 rdar://165771567 Reviewed by Dan Hecht. Temporarily disabling this feature due to stability concerns while we continue development. Canonical link: https://commits.webkit.org/303853@main
…anes orientation switch" https://bugs.webkit.org/show_bug.cgi?id=303424 rdar://problem/165722322 Reviewed by Elika Etemad. The upstream WPT change adds grid-auto-flow in places that are missing. Upstream commit: web-platform-tests/wpt@e2ce61e * LayoutTests/TestExpectations: * LayoutTests/imported/w3c/web-platform-tests/css/css-grid/grid-lanes/tentative/grid-placement/row-explicit-placement-001.html: * LayoutTests/imported/w3c/web-platform-tests/css/css-grid/grid-lanes/tentative/grid-placement/row-explicit-placement-002.html: * LayoutTests/imported/w3c/web-platform-tests/css/css-grid/grid-lanes/tentative/grid-placement/row-explicit-placement-003.html: * LayoutTests/imported/w3c/web-platform-tests/css/css-grid/grid-lanes/tentative/grid-placement/row-explicit-placement-004.html: * LayoutTests/imported/w3c/web-platform-tests/css/css-grid/grid-lanes/tentative/grid-placement/row-explicit-placement-005.html: * LayoutTests/imported/w3c/web-platform-tests/css/css-grid/grid-lanes/tentative/grid-placement/row-explicit-placement-006.html: * LayoutTests/imported/w3c/web-platform-tests/css/css-grid/grid-lanes/tentative/grid-placement/row-explicit-placement-007.html: * LayoutTests/imported/w3c/web-platform-tests/css/css-grid/grid-lanes/tentative/item-placement/row-negative-margin-001.html: Canonical link: https://commits.webkit.org/303854@main
https://bugs.webkit.org/show_bug.cgi?id=303492 rdar://165780580 Reverting 303842@main Reverted change: [iOS26]: 12x fast/viewport/ios tests are constant text failures rdar://165712922 https://bugs.webkit.org/show_bug.cgi?id=303410 303842@main (a415a5d) Canonical link: https://commits.webkit.org/303855@main
https://bugs.webkit.org/show_bug.cgi?id=303245 Reviewed by Yusuke Suzuki. Implement basic support for PlainYearMonth, without its methods. Test: JSTests/stress/temporal-plainyearmonth.js * JSTests/stress/temporal-plainyearmonth.js: Added. (shouldBe): (shouldThrow): (const.yearMonth.new.Temporal.PlainYearMonth): * JSTests/test262/config.yaml: * JSTests/test262/expectations.yaml: * Source/JavaScriptCore/CMakeLists.txt: * Source/JavaScriptCore/DerivedSources-input.xcfilelist: * Source/JavaScriptCore/DerivedSources-output.xcfilelist: * Source/JavaScriptCore/DerivedSources.make: * Source/JavaScriptCore/Sources.txt: * Source/JavaScriptCore/runtime/ISO8601.h: * Source/JavaScriptCore/runtime/JSGlobalObject.cpp: (JSC::JSGlobalObject::init): (JSC::JSGlobalObject::visitChildrenImpl): * Source/JavaScriptCore/runtime/JSGlobalObject.h: (JSC::JSGlobalObject::plainYearMonthStructure): * Source/JavaScriptCore/runtime/TemporalCalendar.cpp: (JSC::TemporalCalendar::isoDateFromFields): (JSC::TemporalCalendar::isoDateAdd): (JSC::TemporalCalendar::calendarDateUntil): * Source/JavaScriptCore/runtime/TemporalObject.cpp: (JSC::createPlainYearMonthConstructor): * Source/JavaScriptCore/runtime/TemporalPlainYearMonth.cpp: Added. (JSC::TemporalPlainYearMonth::create): (JSC::TemporalPlainYearMonth::createStructure): (JSC::TemporalPlainYearMonth::TemporalPlainYearMonth): (JSC::TemporalPlainYearMonth::finishCreation): (JSC::TemporalPlainYearMonth::visitChildrenImpl): (JSC::TemporalPlainYearMonth::tryCreateIfValid): (JSC::TemporalPlainYearMonth::monthCode const): * Source/JavaScriptCore/runtime/TemporalPlainYearMonth.h: Added. * Source/JavaScriptCore/runtime/TemporalPlainYearMonthConstructor.cpp: Added. (JSC::TemporalPlainYearMonthConstructor::create): (JSC::TemporalPlainYearMonthConstructor::createStructure): (JSC::TemporalPlainYearMonthConstructor::TemporalPlainYearMonthConstructor): (JSC::TemporalPlainYearMonthConstructor::finishCreation): (JSC::JSC_DEFINE_HOST_FUNCTION): * Source/JavaScriptCore/runtime/TemporalPlainYearMonthConstructor.h: Added. * Source/JavaScriptCore/runtime/TemporalPlainYearMonthPrototype.cpp: Added. (JSC::TemporalPlainYearMonthPrototype::create): (JSC::TemporalPlainYearMonthPrototype::createStructure): (JSC::TemporalPlainYearMonthPrototype::TemporalPlainYearMonthPrototype): (JSC::TemporalPlainYearMonthPrototype::finishCreation): (JSC::JSC_DEFINE_CUSTOM_GETTER): * Source/JavaScriptCore/runtime/TemporalPlainYearMonthPrototype.h: Added. Canonical link: https://commits.webkit.org/303856@main
… parent https://bugs.webkit.org/show_bug.cgi?id=303417 rdar://163028025 Reviewed by Sammy Gill. When a frameset enters fullscreen while being a grid item of a parent with container-type: size, container query interleaving (StyleTreeResolver.cpp:1382-1403) defers descendant style resolution, preventing StyleAdjuster from setting position:absolute on the top layer element. This creates a broken invariant where isGridItem() returns true but containingBlock() returns RenderView (per top layer rules), causing an invalid downcast from RenderView to RenderGrid in computeInlineDirectionMargins. The fix changes the downcast to dynamicDowncast, returning nullptr when the containing block is RenderView and safely avoiding the crash. This is a defensive workaround; the deeper architectural issue of container query interleaving blocking StyleAdjuster on top layer elements is tracked separately in https://bugs.webkit.org/show_bug.cgi?id=303414#. This specific test/crash requires several specific conditions. display: grid on the parent makes the frameset a grid item. container-type: size triggers both the container query interleaving and the shouldApplySizeContainment() check that causes a second track sizing pass in RenderGrid. content: test on the frameset makes it a RenderBlockFlow instead of a RenderFrameSet, allowing it to participate in grid layout. And the fullscreen request triggers the layout that exposes the type confusion. Test: fullscreen/fullscreen-grid-item-container-type-crash.html * LayoutTests/TestExpectations: * LayoutTests/fullscreen/fullscreen-grid-item-container-type-crash-expected.txt: Added. * LayoutTests/fullscreen/fullscreen-grid-item-container-type-crash.html: Added. * LayoutTests/platform/glib/TestExpectations: * Source/WebCore/rendering/RenderBox.cpp: (WebCore::RenderBox::computeInlineDirectionMargins const): Canonical link: https://commits.webkit.org/303857@main
https://bugs.webkit.org/show_bug.cgi?id=271360 rdar://problem/125529396 Reviewed by Brent Fulgham. This patch aligns WebKit with Gecko / Firefox, Blink / Chromium and Web Specification [1]: [1] https://drafts.csswg.org/css-highlight-api/#highlightregistry This removes `constructor` since it is not in web specification. Additionally, this patch addresses two issues with the HighlightRegistry tampered prototype test: 1. Removes reliance on the non-standard HighlightRegistry constructor and instead uses CSS.highlights to access the standard HighlightRegistry instance. 2. Properly restores Map.prototype to its original state after the test completes. Since Map.prototype is frozen during the test, restoration is done by creating a new prototype object with the saved descriptors and replacing Map's prototype, rather than attempting to modify the frozen object directly. NOTE: Both of above changes were suggested by Alexey Shvaika (original test author). * LayoutTests/highlight/highlight-interfaces-expected.txt: * LayoutTests/highlight/highlight-interfaces.html: * LayoutTests/imported/w3c/web-platform-tests/css/css-highlight-api/idlharness.window-expected.txt: * Source/WebCore/Modules/highlight/HighlightRegistry.idl: * LayoutTests/imported/w3c/web-platform-tests/css/css-highlight-api/HighlightRegistry-maplike-tampered-Map-prototype.html: * LayoutTests/imported/w3c/web-platform-tests/css/css-highlight-api/HighlightRegistry-maplike-expected.txt: Canonical link: https://commits.webkit.org/303858@main
https://bugs.webkit.org/show_bug.cgi?id=303486 rdar://165770860 Reviewed by Geoffrey Garen. 303784@main wasn't sufficient. Something strange is going on, and it's unclear why it doesn't happen on Tahoe. Skip the tests on debug until we investigate. Test: Tools/TestWebKitAPI/Tests/WebKitCocoa/SiteIsolation.mm * Source/WebKit/Platform/IPC/MessageReceiverMap.cpp: (IPC::MessageReceiverMap::addMessageReceiver): * Tools/TestWebKitAPI/Tests/WebKitCocoa/SiteIsolation.mm: Canonical link: https://commits.webkit.org/303859@main
…pport https://bugs.webkit.org/show_bug.cgi?id=303453 rdar://165741135 Reviewed by Tim Nguyen. * LayoutTests/TestExpectations: * Source/WTF/Scripts/Preferences/UnifiedWebPreferences.yaml: * Source/WebKit/Shared/WebPreferencesDefaultValues.h: Canonical link: https://commits.webkit.org/303860@main
…ive-outside-range-test.html` fails with "Threaded Scroll-driven Animations" enabled https://bugs.webkit.org/show_bug.cgi?id=303407 rdar://165710235 Reviewed by Simon Fraser. We need to clamp the scroll offset for scroll timelines such that the current time does not account for rubber-banding. Note that there is no test change in this patch since the flag is not yet enabled on bots. This was caught in preparation of that running animation tests locally using `--experimental-feature ThreadedScrollDrivenAnimationsEnabled=true`. * Source/WebCore/page/scrolling/ScrollingTreeScrollingNode.cpp: (WebCore::ScrollingTreeScrollingNode::clampedCurrentScrollOffset const): * Source/WebCore/page/scrolling/ScrollingTreeScrollingNode.h: * Source/WebKit/UIProcess/RemoteLayerTree/RemoteProgressBasedTimeline.cpp: (WebKit::RemoteProgressBasedTimeline::updateCurrentTime): Canonical link: https://commits.webkit.org/303861@main
…64301290) https://bugs.webkit.org/show_bug.cgi?id=303488 rdar://165774936 Reviewed by Aditya Keerthi and Abrar Rahman Protyasha. Fix the build after the changes in rdar://164301290; we avoid importing `<UIFoundation/NSTextTable.h>` on Catalyst, since the header does not declare `NSTextBlock` when `TARGET_OS_MACCATALYST` is defined (so we need to fall back on forward declarations). * Source/WebCore/PAL/pal/spi/ios/UIKitSPI.h: * Source/WebCore/editing/cocoa/AttributedString.mm: (WebCore::extractParagraphStyle): * Source/WebCore/editing/cocoa/NodeHTMLConverter.mm: (HTMLConverter::_exitElement): Canonical link: https://commits.webkit.org/303862@main
…k that message comes from process containing frame rdar://165728342 https://bugs.webkit.org/show_bug.cgi?id=303437 Reviewed by Matt Woodrow. WebPageProxy::{broadcastFrameTreeSyncData,broadcastAllFrameTreeSyncData} contains a MESSAGE_CHECK that supposedly checks that the message is coming from the process that owns the frame, according to the comment. But the MESSAGE_CHECK itself actually checks if the process sending the message is the main process. Fix the check to follow the comment. * Source/WebKit/UIProcess/WebPageProxy.cpp: (WebKit::WebPageProxy::broadcastFrameTreeSyncData): (WebKit::WebPageProxy::broadcastAllFrameTreeSyncData): Canonical link: https://commits.webkit.org/303863@main
https://bugs.webkit.org/show_bug.cgi?id=303449 rdar://165738264 Reviewed by Brent Fulgham. Currently, we check the column and positioning style independently when determining if we can take GFC. This is fine, but it seems like we will want to check the start and end values together in certain scenarios as we continue to relax these restrictions. For example, if we have an auto column-start, we may only want to allow this if the column-end is an explicit position or auto. This patch refactors this code to better handle this dependency between the start and end positioning properties. This is done by: 1.) Switching on the start (e.g. grid-column-start) property 2.) If we have a valid start, then calling a helper function to check the corresponding end property based upon the type of start. For example, now if we have an explicit grid-column-start and we meet all other criteria, we will check the column-end by calling hasValidColumnEnd. Currently, we have a hasValidColumnEnd/hasValidRowEnd that takes in a Style::GridPositionExplicit, but as we continue to allow more types of content, we can create more overloads of these functions to check the column/row end corresponding to a specific type of start. Canonical link: https://commits.webkit.org/303864@main
rdar://165607857 https://bugs.webkit.org/show_bug.cgi?id=303306 Reviewed by Chris Dumez. We finalize the support of end of iteration by having ReadableStream::Iterator implement a isFinished getter. This is usd by JSDOMAsyncIteratorBase to identify when the next promise resolves whether the iteration is done. We then mark the IsFinish object as finished according ReadableStream::Iterator isFinished getter. This is done in getNextIterationResult. We also add a way to iterate asynchronously via DOMPromise: - ReadableStream::Iterator::next returns a Ref<DOMPromise> - JSDOMAsyncIteratorBase::getNextIterationResult will return the promise. This allows ReadableStream::Iterator to forward any error that may be given to it. We use IsAsyncIteratorNextReturningPromise concept to detect whether ReadableStream::Iterator::next is using the new code path. We keep the old code path and plan to rework it in a follow-up. All ReadableStream async iterator tests are now passing. Covered by existing tests. Canonical link: https://commits.webkit.org/303865@main
…YearMonth https://bugs.webkit.org/show_bug.cgi?id=303413 Reviewed by Darin Adler. Implement these methods. Co-authored-by: Darin Adler <[email protected]> * JSTests/stress/temporal-plainyearmonth.js: (shouldThrow): * JSTests/test262/config.yaml: * JSTests/test262/expectations.yaml: * Source/JavaScriptCore/runtime/ISO8601.cpp: (JSC::ISO8601::temporalYearMonthToString): * Source/JavaScriptCore/runtime/ISO8601.h: * Source/JavaScriptCore/runtime/TemporalPlainYearMonth.cpp: (JSC::TemporalPlainYearMonth::toString const): * Source/JavaScriptCore/runtime/TemporalPlainYearMonth.h: * Source/JavaScriptCore/runtime/TemporalPlainYearMonthPrototype.cpp: (JSC::JSC_DEFINE_HOST_FUNCTION): Canonical link: https://commits.webkit.org/303866@main
…g to start https://bugs.webkit.org/show_bug.cgi?id=283546 Reviewed by Adrian Perez de Castro. Third time is the charm! We have previously landed two failed attempts to fix frequent crashing in WebPageProxy::sendWheelEvent. Turns out Lauro Moura had already debugged this and proposed a fix, which had never landed. An AuxiliaryProcessProxy does not initially have a valid IPC::Connection; instead it stores IPC messages into a queue, to be processed and sent when the auxiliary process has finished launching. We simply need to use that functionality instead of trying to access the connection manually. This commit is a rebased version of Lauro's work, except I use WebPageProxy::sendToProcessContainingFrame to avoid breaking site isolation, and have also changed WebPageProxy::sendWheelEventScrollingAccelerationCurveIfNecessary, which has the same bug. Finally, I also revert the unnecessary changes that I previously landed in 302030@main. Canonical link: https://commits.webkit.org/303867@main
rdar://165741968 https://bugs.webkit.org/show_bug.cgi?id=303455 Reviewed by Eric Carlson. The -legibleMenuController:didRequestCaptionPreviewForProfileID: delegate callback was not being called on macOS because we reset the NSMenu's delegate to ourselves after creation; this overrides AVKit's own delegate, and subsequently, their own delegate methods were not called when menu items are hovered. Take this opportunity to refactor the base classes of WKCaptionStyleMenuController so that behavior when hovering can be shared between the base class and the AVKit subclass. And add support for additional delegate callbacks like -_WKCaptionStyleMenuControllerInternal. * Source/WebKit/UIProcess/Cocoa/_WKCaptionStyleMenuControllerInternal.h: * Source/WebKit/UIProcess/ios/_WKCaptionStyleMenuControllerAVKit.mm: (-[_WKCaptionStyleMenuControllerAVKit legibleMenuController:didRequestCaptionPreviewForProfileID:]): (-[_WKCaptionStyleMenuControllerAVKit legibleMenuControllerDidRequestStoppingSubtitleCaptionPreview:]): (-[_WKCaptionStyleMenuControllerAVKit findAndDismissContextMenus]): Deleted. (-[_WKCaptionStyleMenuControllerAVKit searchForContextMenuInteractionsInView:]): Deleted. * Source/WebKit/UIProcess/ios/_WKCaptionStyleMenuControllerIOS.mm: (+[WKCaptionStyleMenuController menuController]): (-[WKCaptionStyleMenuController setPreviewProfileID:]): * Source/WebKit/UIProcess/mac/_WKCaptionStyleMenuControllerAVKitMac.mm: (-[_WKCaptionStyleMenuControllerAVKitMac rebuildMenu]): (-[_WKCaptionStyleMenuControllerAVKitMac legibleMenuController:didRequestCaptionPreviewForProfileID:]): (-[_WKCaptionStyleMenuControllerAVKitMac legibleMenuControllerDidRequestStoppingSubtitleCaptionPreview:]): (-[_WKCaptionStyleMenuControllerAVKitMac findAndDismissPopoverMenus]): Deleted. (-[_WKCaptionStyleMenuControllerAVKitMac searchForMenuInteractionsInWindow:]): Deleted. * Source/WebKit/UIProcess/mac/_WKCaptionStyleMenuControllerMac.mm: (-[WKCaptionStyleMenuController profileMenuItemHighlighted:]): (-[WKCaptionStyleMenuController setPreviewProfileID:]): Canonical link: https://commits.webkit.org/303868@main
… if they should be disallowed (fast/images/page-wide-animation-toggle.html is timing out after 302278@main). https://bugs.webkit.org/show_bug.cgi?id=301655 rdar://163668078 Unreviewed test gardening * LayoutTests/platform/mac-wk2/TestExpectations: Canonical link: https://commits.webkit.org/303869@main
…irectiveGeneration.VerifyFragmentRanges is a flaky timeout rdar://165630504 Unreviewed test gardening. * Tools/Scripts/webkitpy/api_tests/allowlist.txt: Canonical link: https://commits.webkit.org/303870@main
|
🔄 Preview build started Release tag: |
|
✅ Preview build completed Release: autobuild-preview-pr-118-11651087 |
Member
Author
|
/build-preview |
|
🔄 Preview build started Release tag: |
|
✅ Preview build completed Release: autobuild-preview-pr-118-3cb865b0 |
Member
Author
|
/build-preview |
|
🔄 Preview build started Release tag: |
|
✅ Preview build completed Release: autobuild-preview-pr-118-6ebc1d6e |
Member
Author
|
/build-preview |
|
🔄 Preview build started Release tag: |
|
✅ Preview build completed Release: autobuild-preview-pr-118-93c3b800 |
Member
Author
|
/build-preview |
|
🔄 Preview build started Release tag: |
|
✅ Preview build completed Release: autobuild-preview-pr-118-15810d00 |
Member
Author
|
/build-preview |
|
🔄 Preview build started Release tag: |
|
❌ Preview build failed Check the workflow run for details. |
|
✅ Preview build completed Release: autobuild-preview-pr-118-578fb0ce |
Member
Author
|
/build-preview |
|
🔄 Preview build started Release tag: |
|
✅ Preview build completed Release: autobuild-preview-pr-118-740bfcf2 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
20 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Upgrades WebKit
WebKit/JavaScriptCore Changes Summary
This document summarizes the significant changes in JavaScriptCore between the previous and current WebKit versions used by Bun.
Major Changes
Promise System Rewrite (Most Impactful)
[JSC] Move most of Promise operations to C++ (7598b7e)
JSPromise::reject(),JSPromise::fulfill(),JSPromise::rejectAsHandled(), andJSPromise::performPromiseThen()now require aVM¶meterJSPromiseReactionchanged fromJSInternalFieldObjectImplto a plainJSCellPromise.$resolveandPromise.$rejectstatic properties - replaced with@promiseResolveand@promiseRejectlink-time constant functions[JSC] Make all of Microtask defined via InternalMicrotask ID (8844dcd)
InternalMicrotaskenum values for different promise operation types[JSC] Always associate JSGlobalObject to microtask (a69a3f6)
[JSC] Support
Promise.racein async stack trace (eb9fff6)WebAssembly IPInt SIMD Support
Multiple commits added SIMD support to the IPInt (Interpreter) tier:
Temporal API Implementation Progress
Significant progress on the Temporal API (ECMAScript proposal):
isoDateDifference()withcalendarDateUntil()(f1cf8d2)VM and Runtime Improvements
Implement a Stop the World mechanism for JSC (2e80a20)
[JSC] Implement
Array.ofin C++ (8d5b7e4)Array.offrom JavaScript builtins to C++ for performance[JSC] Do not read Structure's classInfo field in the fast path for isThenFastAndNonObservable (c1b92cb)
[JSC] Remove @getSetSizeAsInt in SetPrototype.js (c1ecefa)
Bug Fixes
Conditional writes are incorrectly handled in Array allocation sinking (934b1e2)
ASSERTION FAILED: !needsSlowPutIndexing() at ensureArrayStorageSlow (716536c)
ASSERTION FAILED: constructor.isObject() when OSR from an inlined function (8b9fc1b)
[JSC] Fix instanceof metadata fields in LLIntPrototypeLoadAdaptiveStructureWatchpoint (7563333)
[JSC] Add stack overflow checks to LiteralParser::parse (06f13ed)
[JSC] Math.random JIT path uses arithmetic instead of logical right shifts (cf0e17e)
Fix IPInt catch handler bounds generation (09be147)
OMG stack slots should be positioned at the beginning of the OSR buffer when IPInt OSR layout is used (899a38c)
Code Quality / SaferCPP
Multiple commits addressing C++ safety warnings:
JSC::ArrayBuffer(3ea1357)JSC::Watchpoint(5ff7a5d)LIFETIME_BOUNDforWTF::RefPtr(74dea89)Other Notable Changes
new(dabbab2)tryMakeStringfor error message on invalid destructuring (ff79c41)Bun-Specific Patches Applied
The following Bun-specific changes were applied on top of upstream:
BunPerformMicrotaskJobandBunInvokeJobWithArgumentsBreaking Changes for Bun Integration
reject(),fulfill(),rejectAsHandled(),performPromiseThen()now requireVM&as first parameterPromise.$resolveandPromise.$rejectremoved - must use$promiseResolve()and$promiseReject()functions insteadJSInternalFieldObjectImplto plainJSCellMigration Notes
When updating Bun bindings for this WebKit version:
promise->reject(globalObject, value)calls topromise->reject(vm, globalObject, value)promise->rejectAsHandled(globalObject, value)calls topromise->rejectAsHandled(vm, globalObject, value)Promise.$resolve(value)with$promiseResolve(Promise, value)in JavaScript builtinsPromise.$reject(value)with$promiseReject(Promise, value)in JavaScript builtins