| 2015-11-18 Andy Estes <aestes@apple.com> |
| |
| [Content Filtering] Crash in DocumentLoader::notifyFinished() when allowing a media document to load |
| https://bugs.webkit.org/show_bug.cgi?id=151433 |
| rdar://problem/23506594 |
| |
| Reviewed by Alexey Proskuryakov. |
| |
| When the main resource of a media document commits, WebKit cancels its load since the plug-in or media engine |
| will do its own loading. If content filtering is enabled, and the filter waits allow the load until the entire |
| resource is downloaded, then ContentFilter will attempt to call DocumentLoader::notifyFinished() immediately |
| after delivering the buffered resource data to DocumentLoader. However, delivering the data will have nulled out |
| DocumentLoader's m_mainResource when the load was cancelled, leading to a crash in notifyFinished(). |
| |
| To resolve this, add a new Stopped state to ContentFilter. Set this state if DocumentLoader clears its main |
| resource or detaches from its frame. If ContentFilter is in the Stopped state after calling |
| DocumentLoader::dataReceived(), do not proceed to call DocumentLoader::notifyFinished(). |
| |
| Test: contentfiltering/allow-media-document.html |
| |
| * loader/ContentFilter.cpp: |
| (WebCore::ContentFilter::stopFilteringMainResource): Set m_state to Stopped. If m_mainResource is non-null, |
| removed ContentFilter as a client and set m_mainResource to null. |
| (WebCore::ContentFilter::notifyFinished): Stopped calling DocumentLoader::notifyFinished() if m_state is Stopped |
| after calling DocumentLoader::dataReceived(). |
| * loader/ContentFilter.h: |
| * loader/DocumentLoader.cpp: |
| (WebCore::DocumentLoader::detachFromFrame): Called ContentFilter::stopFilteringMainResource() instead of setting |
| m_contentFilter to null. |
| (WebCore::DocumentLoader::clearMainResource): Ditto. |
| |
| 2015-11-25 Pranjal Jumde <pjumde@apple.com> |
| |
| Checks for buffer-overflows when reading characters from textRun |
| https://bugs.webkit.org/show_bug.cgi?id=151055 |
| <rdar://problem/23251789> |
| |
| Reviewed by Myles C. Maxfield. |
| |
| Prevents an off by one error when adding the last font data to the GlyphBuffer. |
| |
| * Source/WebCore/platform/graphics/WidthIterator.cpp: |
| * Source/WebCore/platform/graphics/FontCascade.cpp: |
| |
| 2015-11-22 Andy Estes <aestes@apple.com> |
| |
| Teach MiniBrowser how to enable the mock content filter |
| https://bugs.webkit.org/show_bug.cgi?id=151540 |
| |
| Reviewed by Andreas Kling. |
| |
| Moved the implementation of MockContentFilterEnabler from TestWebKitAPI to here, renamed it to |
| WebMockContentFilterEnabler, and made it compatible with the legacy Objective-C runtime. Renamed Decision and |
| DecisionPoint to WebMockContentFilterDecision and WebMockContentFilterDecisionPoint, and changed them from enum |
| classes to CF_ENUMs so that they can be used by both C++ and Objective-C source files. |
| |
| * WebCore.xcodeproj/project.pbxproj: |
| * bindings/js/JSMockContentFilterSettingsCustom.cpp: |
| (WebCore::JSMockContentFilterSettings::decisionPoint): |
| (WebCore::JSMockContentFilterSettings::setDecisionPoint): |
| (WebCore::toDecision): |
| (WebCore::JSMockContentFilterSettings::decision): |
| (WebCore::JSMockContentFilterSettings::setDecision): |
| (WebCore::JSMockContentFilterSettings::unblockRequestDecision): |
| (WebCore::JSMockContentFilterSettings::setUnblockRequestDecision): |
| (WebCore::toJSValue): Deleted. |
| * testing/MockContentFilter.cpp: |
| (WebCore::MockContentFilter::willSendRequest): |
| (WebCore::MockContentFilter::responseReceived): |
| (WebCore::MockContentFilter::addData): |
| (WebCore::MockContentFilter::finishedAddingData): |
| (WebCore::MockContentFilter::unblockHandler): |
| (WebCore::MockContentFilter::maybeDetermineStatus): |
| * testing/MockContentFilter.h: |
| * testing/MockContentFilterEnabler.h: Added. |
| * testing/MockContentFilterEnabler.mm: Added. |
| (-[WebMockContentFilterEnabler initWithDecision:decisionPoint:blockedString:]): |
| (-[WebMockContentFilterEnabler initWithCoder:]): |
| (-[WebMockContentFilterEnabler encodeWithCoder:]): |
| (-[WebMockContentFilterEnabler enable]): |
| (-[WebMockContentFilterEnabler dealloc]): |
| (+[WebMockContentFilterEnabler supportsSecureCoding]): |
| (-[WebMockContentFilterEnabler copyWithZone:]): |
| * testing/MockContentFilterSettings.h: |
| (WebCore::MockContentFilterSettings::decisionPoint): |
| (WebCore::MockContentFilterSettings::setDecisionPoint): |
| (WebCore::MockContentFilterSettings::decision): |
| (WebCore::MockContentFilterSettings::setDecision): |
| (WebCore::MockContentFilterSettings::unblockRequestDecision): |
| (WebCore::MockContentFilterSettings::setUnblockRequestDecision): |
| |
| 2015-11-24 Xabier Rodriguez Calvar <calvaris@igalia.com> |
| |
| [Streams API] Implement pipeTo method in readable Stream |
| https://bugs.webkit.org/show_bug.cgi?id=151588 |
| |
| Reviewed by Darin Adler. |
| |
| Implemented pipeTo method according to the reference implementation in the spec as the spec is not written |
| yet. It can be found at https://github.com/whatwg/streams/blob/632b26a05f3106650b1ec91239ad5b012e6c64af/reference-implementation/lib/readable-stream.js#L75. |
| |
| Tests: streams/pipe-to.html |
| streams/reference-implementation/brand-checks.html |
| streams/reference-implementation/pipe-through.html |
| streams/reference-implementation/pipe-to.html |
| streams/reference-implementation/pipe-to-options.html |
| streams/reference-implementation/readable-stream-templated |
| |
| * Modules/streams/ReadableStream.js: |
| (doPipe): Internal function of pipeTo. |
| (closeDest): Internal function of pipeTo. |
| (abortDest): Internal function of pipeTo. |
| (pipeTo): Implemented as per spec with some other internal functions as helpers. |
| |
| 2015-11-24 Antti Koivisto <antti@apple.com> |
| |
| REGRESSION (r190983): Non-element, non-text nodes should not be distributed to slots |
| https://bugs.webkit.org/show_bug.cgi?id=151566 |
| rdar://problem/23430177 |
| |
| Reviewed by Zalan Bujtas. |
| |
| We don't invalidate slot assignments except for text or element children. Fix by not |
| not assigning other nodes to slots as it is not useful. |
| |
| Test: fast/html/details-comment-crash.html |
| |
| * dom/SlotAssignment.cpp: |
| (WebCore::slotNameFromSlotAttribute): |
| (WebCore::SlotAssignment::findAssignedSlot): |
| (WebCore::SlotAssignment::assignSlots): |
| |
| 2015-11-23 David Kilzer <ddkilzer@apple.com> |
| |
| Hardening against CSSSelector double frees |
| <http://webkit.org/b/56124> |
| <rdar://problem/9119036> |
| |
| Reviewed by Antti Koivisto. |
| |
| Add some security assertions to catch this issue if it ever |
| happens in Debug builds, and make changes in |
| CSSSelector::~CSSSelector() and |
| CSSSelectorList::deleteSelectors() to prevent obvious issues if |
| they're ever called twice in Release builds. |
| |
| No new tests because we don't know how to reproduce this. |
| |
| * css/CSSSelector.cpp: |
| (WebCore::CSSSelector::CSSSelector): Initialize |
| m_destructorHasBeenCalled. |
| * css/CSSSelector.h: |
| (WebCore::CSSSelector::m_destructorHasBeenCalled): Add bitfield. |
| (WebCore::CSSSelector::CSSSelector): Initialize |
| m_destructorHasBeenCalled. |
| (WebCore::CSSSelector::~CSSSelector): Add security assertion |
| that this is never called twice. Clear out any fields that |
| would have caused us to dereference an object twice. |
| |
| * css/CSSSelectorList.cpp: |
| (WebCore::CSSSelectorList::deleteSelectors): Clear |
| m_selectorArray when freeing the memory to which it was |
| pointing. This prevents re-entrancy issues or calling this |
| method twice on the same thread. Also restructure the for() |
| loop to prevent calling CSSSelector::isLastInSelectorList() |
| after CSSSelector::~CSSSelector() has been called (via CRBug |
| 241892). |
| |
| 2015-11-23 Brian Burg <bburg@apple.com> |
| |
| Web Inspector: inspector settings should not be shared between different inspection levels |
| https://bugs.webkit.org/show_bug.cgi?id=151151 |
| |
| Reviewed by Timothy Hatcher. |
| |
| InspectorController and InspectorFrontendClient should know how to compute their |
| inspection levels, so that this level can be exposed to the Inspector frontend for |
| preference disambiguation by inspection level. |
| |
| - A Page with normal web content has an inspection level of 0. |
| - The first Web Inspector has an inspection level of 1. |
| - The second Web Inspector that inspects the Inspector has an inspection level of 2. |
| - And so forth... |
| |
| For local frontend clients (WK1 and the ProtocolTestStub), the inspection level of |
| a Page computed as follows: |
| |
| - If the page's inspector controller has no frontend client, then the Page |
| is normal web content (level 0). |
| - If the page's inspector controller has a frontend client, ask for its inspection level. |
| A frontend client looks at the inspection level reported by inspected page's inspector |
| controller and increments by one to account for the crossed inspection boundary. |
| |
| No new tests, this is blocked by test infrastructure fixes (https://webkit.org/b/151573). |
| |
| * inspector/InspectorController.cpp: |
| (WebCore::InspectorController::inspectionLevel): Added. |
| (WebCore::InspectorController::hasInspectorFrontendClient): Deleted. |
| * inspector/InspectorController.h: |
| * inspector/InspectorFrontendClient.h: |
| * inspector/InspectorFrontendClientLocal.cpp: |
| (WebCore::InspectorFrontendClientLocal::canAttachWindow): Check inspectionLevel(). |
| (WebCore::InspectorFrontendClientLocal::inspectionLevel): Added. |
| * inspector/InspectorFrontendClientLocal.h: |
| * inspector/InspectorFrontendHost.cpp: |
| (WebCore::InspectorFrontendHost::inspectionLevel): Added. |
| * inspector/InspectorFrontendHost.h: |
| * inspector/InspectorFrontendHost.idl: |
| * page/ContextMenuController.cpp: |
| (WebCore::ContextMenuController::populate): Check inspectionLevel(). |
| |
| 2015-11-20 Brian Burg <bburg@apple.com> |
| |
| Web Inspector: RemoteInspector should track targets and connections for remote automation |
| https://bugs.webkit.org/show_bug.cgi?id=151042 |
| |
| Reviewed by Joseph Pecoraro. |
| |
| Use the new RemoteControllableTarget API method names. |
| |
| No new tests, no behavior change. |
| |
| * page/PageDebuggable.cpp: |
| (WebCore::PageDebuggable::connect): |
| (WebCore::PageDebuggable::dispatchMessageFromRemote): |
| (WebCore::PageDebuggable::dispatchMessageFromRemoteFrontend): Deleted. |
| * page/PageDebuggable.h: |
| |
| 2015-11-23 Brady Eidson <beidson@apple.com> |
| |
| Modern IDB: Unskip all indexeddb/mozilla tests that pass. |
| https://bugs.webkit.org/show_bug.cgi?id=151568 |
| |
| Reviewed by Alex Christensen. |
| |
| No new tests (Lots of skipped existing tests now pass). |
| |
| * Modules/indexeddb/shared/IDBError.h: |
| |
| 2015-11-23 Nikos Andronikos <nikos.andronikos-webkit@cisra.canon.com.au> |
| |
| Add WebCore namespace for generated bindings for supplemental method calls |
| https://bugs.webkit.org/show_bug.cgi?id=151431 |
| |
| Reviewed by Alex Christensen. |
| |
| Address build failures for generated supplemental method calls which are in the WebCore namespace. |
| The bindings are outside the WebCore namespace - so add WebCore namespace specifier where required. |
| |
| * bindings/scripts/CodeGeneratorJS.pm: |
| (GenerateHeader): |
| (GenerateImplementation): |
| (GenerateParametersCheck): |
| * bindings/scripts/CodeGeneratorObjC.pm: |
| (GenerateImplementation): |
| * bindings/scripts/test/JS/JSTestInterface.cpp: |
| (WebCore::jsTestInterfaceConstructorSupplementalStaticReadOnlyAttr): |
| (WebCore::jsTestInterfaceConstructorSupplementalStaticAttr): |
| (WebCore::jsTestInterfaceSupplementalStr1): |
| (WebCore::jsTestInterfaceSupplementalStr2): |
| (WebCore::jsTestInterfaceSupplementalNode): |
| (WebCore::setJSTestInterfaceConstructorSupplementalStaticAttr): |
| (WebCore::setJSTestInterfaceSupplementalStr2): |
| (WebCore::setJSTestInterfaceSupplementalNode): |
| (WebCore::jsTestInterfacePrototypeFunctionSupplementalMethod1): |
| (WebCore::jsTestInterfacePrototypeFunctionSupplementalMethod2): |
| (WebCore::jsTestInterfaceConstructorFunctionSupplementalMethod4): |
| * bindings/scripts/test/ObjC/DOMTestInterface.mm: |
| (-[DOMTestInterface supplementalStr1]): |
| (-[DOMTestInterface supplementalStr2]): |
| (-[DOMTestInterface setSupplementalStr2:]): |
| (-[DOMTestInterface supplementalStr3]): |
| (-[DOMTestInterface setSupplementalStr3:]): |
| (-[DOMTestInterface supplementalNode]): |
| (-[DOMTestInterface setSupplementalNode:]): |
| (-[DOMTestInterface builtinAttribute]): |
| (-[DOMTestInterface setBuiltinAttribute:]): |
| (-[DOMTestInterface supplementalMethod1]): |
| (-[DOMTestInterface supplementalMethod2:objArg:]): |
| (-[DOMTestInterface supplementalMethod3]): |
| (-[DOMTestInterface supplementalMethod4]): |
| (-[DOMTestInterface builtinFunction]): |
| |
| 2015-11-23 Brady Eidson <beidson@apple.com> |
| |
| Modern IDB: When a transaction is aborted, call onerror handlers for all in-progress requests. |
| https://bugs.webkit.org/show_bug.cgi?id=151550 |
| |
| Reviewed by Alex Christensen. |
| |
| Test: storage/indexeddb/modern/abort-requests-cancelled.html |
| storage/indexeddb/modern/idbtransaction-objectstore-failures.html (with changes) |
| storage/indexeddb/modern/index-5.html (with changes) |
| Various (currently skipped) legacy IDB tests. |
| |
| * Modules/indexeddb/client/IDBDatabaseImpl.cpp: |
| (WebCore::IDBClient::IDBDatabase::transaction): |
| (WebCore::IDBClient::IDBDatabase::startVersionChangeTransaction): |
| (WebCore::IDBClient::IDBDatabase::didStartTransaction): |
| (WebCore::IDBClient::IDBDatabase::willCommitTransaction): |
| (WebCore::IDBClient::IDBDatabase::didCommitTransaction): |
| (WebCore::IDBClient::IDBDatabase::willAbortTransaction): |
| (WebCore::IDBClient::IDBDatabase::didAbortTransaction): |
| (WebCore::IDBClient::IDBDatabase::didCommitOrAbortTransaction): |
| |
| * Modules/indexeddb/client/IDBTransactionImpl.cpp: |
| (WebCore::IDBClient::IDBTransaction::abort): |
| (WebCore::IDBClient::IDBTransaction::abortOnServerAndCancelRequests): |
| (WebCore::IDBClient::IDBTransaction::didCreateObjectStoreOnServer): |
| (WebCore::IDBClient::IDBTransaction::didCreateIndexOnServer): |
| (WebCore::IDBClient::IDBTransaction::didGetRecordOnServer): |
| (WebCore::IDBClient::IDBTransaction::didDeleteObjectStoreOnServer): |
| (WebCore::IDBClient::IDBTransaction::didDeleteIndexOnServer): |
| (WebCore::IDBClient::IDBTransaction::immediateAbort): Deleted. |
| (WebCore::IDBClient::IDBTransaction::abortOnServer): Deleted. |
| * Modules/indexeddb/client/IDBTransactionImpl.h: |
| |
| * Modules/indexeddb/server/MemoryIDBBackingStore.cpp: |
| (WebCore::IDBServer::MemoryIDBBackingStore::abortTransaction): |
| (WebCore::IDBServer::MemoryIDBBackingStore::commitTransaction): |
| |
| * Modules/indexeddb/shared/IDBError.cpp: |
| (WebCore::idbErrorName): |
| (WebCore::idbErrorDescription): |
| * Modules/indexeddb/shared/IDBError.h: |
| |
| * Modules/indexeddb/shared/IDBResourceIdentifier.cpp: |
| (WebCore::IDBResourceIdentifier::loggingString): |
| * Modules/indexeddb/shared/IDBResourceIdentifier.h: |
| |
| 2015-11-23 Brady Eidson <beidson@apple.com> |
| |
| Modern IDB: Unskip storage/indexeddb/mozilla/global-data.html. |
| https://bugs.webkit.org/show_bug.cgi?id=151557 |
| |
| Reviewed by Alex Christensen. |
| |
| No new tests (Unskipping existing test storage/indexeddb/mozilla/global-data.html). |
| |
| - Reworking some invalid ASSERTS |
| - Actually opening pending open-database-requests after a version change transaction completes |
| - Allow starting new transactions when the version change transaction has *started* finishing, |
| but before it finishes finishing. |
| |
| * Modules/indexeddb/client/IDBDatabaseImpl.cpp: |
| (WebCore::IDBClient::IDBDatabase::transaction): |
| |
| * Modules/indexeddb/client/IDBTransactionImpl.h: |
| |
| * Modules/indexeddb/server/MemoryObjectStore.cpp: |
| (WebCore::IDBServer::MemoryObjectStore::~MemoryObjectStore): |
| |
| * Modules/indexeddb/server/MemoryObjectStoreCursor.cpp: |
| (WebCore::IDBServer::MemoryObjectStoreCursor::keyAdded): Deleted. |
| |
| * Modules/indexeddb/server/UniqueIDBDatabase.cpp: |
| (WebCore::IDBServer::UniqueIDBDatabase::UniqueIDBDatabase): |
| (WebCore::IDBServer::UniqueIDBDatabase::handleOpenDatabaseOperations): |
| (WebCore::IDBServer::UniqueIDBDatabase::commitTransaction): |
| * Modules/indexeddb/server/UniqueIDBDatabase.h: |
| |
| 2015-11-23 Youenn Fablet <youenn.fablet@crf.canon.fr> |
| |
| Remove DOMPromiseWithCallback |
| https://bugs.webkit.org/show_bug.cgi?id=151565 |
| |
| Reviewed by Eric Carlson. |
| |
| DOMPromiseWithCallback is no longer used as WebCore JS built-ins are used to the cases where DOMPromiseWithCallback made sense. |
| |
| No change in behavior. |
| |
| * Modules/mediastream/MediaDevices.h: Moved from DOMPromiseWithCallback to DOMPromise since the API is now called by JS built-ins. |
| * bindings/js/JSDOMPromise.h: |
| (WebCore::DOMPromiseWithCallback::DOMPromiseWithCallback): Deleted. |
| (WebCore::DOMPromiseIteratorWithCallback::DOMPromiseIteratorWithCallback): Deleted. |
| (WebCore::Error>::resolve): Deleted. |
| (WebCore::Error>::reject): Deleted. |
| (WebCore::Error>::resolveEnd): Deleted. |
| |
| 2015-11-23 Brian Burg <bburg@apple.com> |
| |
| Web Inspector: when inspecting the inspector, add the inspection level to the title bar |
| https://bugs.webkit.org/show_bug.cgi?id=151555 |
| |
| Reviewed by Timothy Hatcher. |
| |
| * English.lproj/Localizable.strings: add new localized string for alternate inspector title. |
| |
| 2015-11-23 Zan Dobersek <zdobersek@igalia.com> |
| |
| [GStreamer] No need to assert the pipeline's bus presence in MediaPlayerPrivateGStreamerBase dtor |
| https://bugs.webkit.org/show_bug.cgi?id=151558 |
| |
| Reviewed by Carlos Garcia Campos. |
| |
| In the MediaPlayerPrivateGStreamerBase destructor, there's some leftover code |
| that acquires a reference to the GStreamer pipeline's bus object and then |
| just asserts that it's not null. Not very useful, so remove it. |
| |
| Also use nullptr to null out the m_player member. |
| |
| * platform/graphics/gstreamer/MediaPlayerPrivateGStreamerBase.cpp: |
| (WebCore::MediaPlayerPrivateGStreamerBase::~MediaPlayerPrivateGStreamerBase): |
| |
| 2015-11-17 Sergio Villar Senin <svillar@igalia.com> |
| |
| ASSERTION FAILED: freeSpace >= 0 in WebCore::RenderGrid::computeTrackSizesForDirection |
| https://bugs.webkit.org/show_bug.cgi?id=151254 |
| |
| Reviewed by Darin Adler. |
| |
| When in quirks mode, abnormally huge margins could lead to |
| negative computations of available logical sizes. We could add |
| an !document.inQuirksMode() check in the assertion but since |
| negative freeSpace values are valid (i.e. the condition is not |
| really part of the contract and everything will work as |
| expected) I've decided to simply remove it. |
| |
| Test: fast/css-grid-layout/grid-quirks-mode-huge-margin-crash.html |
| |
| * rendering/RenderGrid.cpp: |
| (WebCore::RenderGrid::computeTrackSizesForDirection): Deleted. |
| |
| 2015-11-22 Myles C. Maxfield <mmaxfield@apple.com> |
| |
| Font selection should not consult font-variant property |
| https://bugs.webkit.org/show_bug.cgi?id=151537 |
| |
| Reviewed by Simon Fraser. |
| |
| In section 4.7 of the CSS Fonts Level 3 spec, it says "[The font-variant and |
| font-feature-settings] do not affect font selection." |
| |
| All the other browsers (Chrome, Firefox, and Edge) all obey the spec here. We |
| are the only one who misbehaves. This patch aligns our behavior with the other |
| browsers. |
| |
| Test: fast/text/font-selection-font-variant.html |
| |
| * css/CSSFontSelector.cpp: |
| (WebCore::computeTraitsMask): Deleted. |
| (WebCore::compareFontFaces): Deleted. |
| (WebCore::CSSFontSelector::getFontFace): Deleted. |
| * css/CSSParser.cpp: |
| (WebCore::isValidKeywordPropertyAndValue): |
| (WebCore::isKeywordPropertyID): |
| (WebCore::CSSParser::createFontFaceRule): |
| (WebCore::CSSParser::CSSParser): Deleted. |
| (WebCore::CSSParser::parseValue): Deleted. |
| (WebCore::CSSParser::parseDeclaration): Deleted. |
| (WebCore::CSSParser::clearProperties): Deleted. |
| (WebCore::CSSParser::parseFontVariant): Deleted. |
| (WebCore::CSSParser::createStyleRule): Deleted. |
| (WebCore::CSSParser::deleteFontFaceOnlyValues): Deleted. |
| * css/CSSParser.h: |
| * platform/graphics/FontDescription.cpp: |
| (WebCore::FontDescription::traitsMask): Deleted. |
| * platform/graphics/win/FontCacheWin.cpp: |
| (WebCore::traitsInFamilyEnumProc): |
| * platform/graphics/cocoa/FontCacheCoreText.cpp: |
| (WebCore::toTraitsMask): |
| * platform/text/TextFlags.h: |
| |
| 2015-11-22 Carlos Garcia Campos <cgarcia@igalia.com> |
| |
| Unreviewed, rolling out r192727. |
| |
| It made the selections transparent again and broke |
| /webkit2/WebKitWebView/snapshot |
| |
| Reverted changeset: |
| |
| "[GTK] RenderThemeGtk::platformActiveSelectionBackgroundColor, |
| et. al. should not clobber state of cached GtkStyleContexts" |
| https://bugs.webkit.org/show_bug.cgi?id=151533 |
| http://trac.webkit.org/changeset/192727 |
| |
| 2015-11-22 Michael Catanzaro <mcatanzaro@igalia.com> |
| |
| [GTK] RenderThemeGtk::platformActiveSelectionBackgroundColor, et. al. should not clobber state of cached GtkStyleContexts |
| https://bugs.webkit.org/show_bug.cgi?id=151533 |
| |
| Reviewed by Carlos Garcia Campos. |
| |
| platformActiveSelectionBackgroundColor(), platformInactiveSelectionBackgroundColor(), etc. |
| are const functions intended only to return a color used for painting, but since r174929 |
| they also change the state of the cached style contexts we use for GTK_TYPE_ENTRY and |
| GTK_TYPE_TREE_VIEW. That's wrong; those style contexts should not have any state set. This |
| could cause theme colors returned by those GtkStyleContexts to change unexpectedly, |
| depending on whether the state is explicitly set before each use, or whether the theme |
| actually uses the states. |
| |
| This didn't cause any regression only because every place using these style contexts |
| explicitly sets the state of the style contexts before use. In fact, the GtkTreeView style |
| context is not used anywhere else, and the GtkEntry style context is only used in |
| paintTextField, which does set the state before use (and then reverts it using |
| save/restore), so this cannot have broken anything in practice. But it's a landmine waiting |
| for the next programmer to trip it. |
| |
| Fix this with a gtk_style_context_save()/gtk_style_context_restore() pair. |
| |
| * rendering/RenderThemeGtk.cpp: |
| (WebCore::styleColor): |
| |
| 2015-11-21 Myles C. Maxfield <mmaxfield@apple.com> |
| |
| Tiny cleanup in ComplexTextController::collectComplexTextRuns() |
| https://bugs.webkit.org/show_bug.cgi?id=151534 |
| |
| Reviewed by Zalan Bujtas. |
| |
| The isMissingGlyph boolean is completely unnecessary. Its entire |
| responsiblity is duplicated by the "font" pointer. |
| |
| No new tests because there is no behavior change. |
| |
| * platform/graphics/mac/ComplexTextController.cpp: |
| (WebCore::ComplexTextController::collectComplexTextRuns): |
| |
| == Rolled over to ChangeLog-2015-11-21 == |
