Timeline
Sep 16, 2013:
- 9:23 PM Changeset in webkit [155929] by
-
- 3 edits in trunk/Tools
[Windows] Clean up WinLauncher by using smart pointers
https://bugs.webkit.org/show_bug.cgi?id=121467
Reviewed by Anders Carlsson.
- WinLauncher/PrintWebUIDelegate.cpp:
(PrintWebUIDelegate::webViewPrintingMarginRect): Switch to smart pointers.
- WinLauncher/WinLauncher.cpp:
(WinLauncherWebHost::updateAddressBar): Use _bstr_t and smart pointers.
(WinLauncherWebHost::didFailProvisionalLoadWithError): Ditto.
(showLastVisitedSites): Smart pointer updates.
(WinLauncherWebHost::didFinishLoadForFrame): Ditto.
(setToDefaultPreferences): Ditto.
(dllLauncherEntryPoint): Ditto.
(PrintView): Ditto.
(LaunchInspector): Ditto.
(NavigateToHistory): Ditto.
(MyEditProc): Ditto.
(loadURL): Ditto.
- 9:22 PM Changeset in webkit [155928] by
-
- 69 edits in trunk/Source/WebCore
CTTE: InputType should store its HTMLInputElement back pointer as a reference
https://bugs.webkit.org/show_bug.cgi?id=121466
Reviewed by Anders Carlsson.
- Converts InputType::create() and InputType::createText() to take a HTMLInputElement& (as well as all the subclasses).
- Change InputType::element() to return an HTMLInputElement& and update all the callers.
- 9:15 PM Changeset in webkit [155927] by
-
- 3 edits in trunk/Source/WebKit2
Fix EFL build after r155888
https://bugs.webkit.org/show_bug.cgi?id=121465
Patch by Sergio Correia <Sergio Correia> on 2013-09-16
Reviewed by Gyuyoung Kim.
Update WebView to follow the changes in PageClient in r155888.
- UIProcess/CoordinatedGraphics/WebView.cpp:
(WebKit::WebView::createDrawingAreaProxy): Use createOwned (addded in
r155407) to return the DrawingAreaProxy from DrawingAreaProxyImpl.
- UIProcess/CoordinatedGraphics/WebView.h: Update createDrawingAreaProxy()
signature to match PageClient.h.
- 6:49 PM Changeset in webkit [155926] by
-
- 4 edits in trunk/Source/WTF
USE(WEB_THREAD): More explicit WebThread initialization
https://bugs.webkit.org/show_bug.cgi?id=121454
Patch by Joseph Pecoraro <Joseph Pecoraro> on 2013-09-16
Reviewed by Benjamin Poulain.
Be more explicit when initializing WebThread only data.
- wtf/MainThread.h:
- wtf/MainThread.cpp:
(WTF::initializeWebThreadOnce):
(WTF::initializeWebThread):
Ensure one time initialization.
- wtf/mac/MainThreadMac.mm:
(WTF::initializeMainThreadPlatform):
(WTF::initializeWebThreadPlatform):
Move WebThread value initialization to its own function.
(WTF::isMainThread):
Remove no longer invalid assert if WebThread was not initialized.
- 6:39 PM Changeset in webkit [155925] by
-
- 5 edits2 copies in branches/safari-537.60-branch/Source/WebCore
Merged r155226. <rdar://problem/14949946>
- 6:31 PM Changeset in webkit [155924] by
-
- 9 edits in trunk/Source
Make InspectorTimelineAgent use an enum for the record type instead of a string.
https://bugs.webkit.org/show_bug.cgi?id=121461
Reviewed by Joseph Pecoraro.
Source/WebCore:
- inspector/Inspector.json:
- inspector/InspectorTimelineAgent.cpp:
(WebCore::InspectorTimelineAgent::addRecordToTimeline):
(WebCore::toProtocol):
(WebCore::InspectorTimelineAgent::innerAddRecordToTimeline):
(WebCore::InspectorTimelineAgent::didCompleteCurrentRecord):
(WebCore::InspectorTimelineAgent::appendRecord):
(WebCore::InspectorTimelineAgent::pushCurrentRecord):
- inspector/InspectorTimelineAgent.h:
(WebCore::InspectorTimelineAgent::TimelineRecordEntry::TimelineRecordEntry):
Source/WebInspectorUI:
- UserInterface/InspectorBackendCommands.js:
- UserInterface/Legacy/6.0/InspectorBackendCommands.js:
- UserInterface/TimelineManager.js:
(WebInspector.TimelineManager.prototype.eventRecorded.processRecord):
(WebInspector.TimelineManager.prototype.eventRecorded):
- Versions/Inspector-iOS-6.0.json:
- 6:26 PM Changeset in webkit [155923] by
-
- 4 edits1 copy1 delete in branches/safari-537.60-branch/Source/WebCore
Merged r155215. <rdar://problem/14949946>
- 6:26 PM Changeset in webkit [155922] by
-
- 6 edits in trunk/Tools
[Windows] Add rudimentary history to WinLauncher.
https://bugs.webkit.org/show_bug.cgi?id=121463
Reviewed by Anders Carlsson.
- WinLauncher/WinLauncher.cpp:
(updateMenuItemForHistoryItem): Added
(showLastVisitedSites): Added
(WinLauncherWebHost::didFinishLoadForFrame): Add history item once
a new page has been visited.
(dllLauncherEntryPoint): Deallocate history on exit.
(NavigateToHistory): Added
(WndProc): Add handler for history entries
- WinLauncher/WinLauncher.vcxproj/WinLauncherLib.rc: Add menu items to hold
up to 10 history slots.
- WinLauncher/WinLauncher.vcxproj/WinLauncherLibCommon.props: Link against
Windows utility library to allow use of _bstr_t type.
- WinLauncher/WinLauncher.vcxproj/WinLauncherLibResource.h: Add menu items
to hold up to 10 history slots.
- WinLauncher/stdafx.h: Add <comutil.h> to precompiled headers.
- 6:21 PM Changeset in webkit [155921] by
-
- 3 edits in branches/safari-537.60-branch/Source/WebKit/win
Merged r155212. <rdar://problem/14949951>
- 6:18 PM Changeset in webkit [155920] by
-
- 5 edits in branches/safari-537.60-branch/Source/WebCore
Merged r155130. <rdar://problem/14932481>
- 6:16 PM Changeset in webkit [155919] by
-
- 2 edits in branches/safari-537.60-branch/Source/WebCore
Merged r155127. <rdar://problem/14932481>
- 6:08 PM Changeset in webkit [155918] by
-
- 13 edits3 adds in trunk
[GTK] Make symbol export filter more strict, and disable for dev/test builds
https://bugs.webkit.org/show_bug.cgi?id=120586
Reviewed by Martin Robinson.
.:
- GNUmakefile.am: only include test-related automake files when developer
mode is on.
- Source/autotools/ReadCommandLineArguments.m4: add --enable-developer-mode,
defaults to no, and to yes for debug builds.
- Source/autotools/SetupAutomake.m4: add ENABLE_DEVELOPER_MODE conditional.
- Source/autotools/symbols.filter: make the exported symbols list much shorter,
covering only the public ABI and a few symbols required by WebKit2 processes.
Source/WebKit/gtk:
- GNUmakefile.am: only use the version script when in developer mode..
Source/WebKit2:
- GNUmakefile.am: only use the version script if developer mode is disabled.
Tools:
- GNUmakefile.am: moved GtkLauncher and documentation builds to their own
makefiles, so they can be included unconditionally.
- GtkLauncher/GNUmakefile.am: added.
- Scripts/webkitdirs.pm:
(runAutogenForAutotoolsProjectIfNecessary): always pass --enable-developer-mode
to configure, since we consider builds done through build-webkit to be development
and test builds.
- gtk/GNUmakefile.am: added.
- gtk/common.py: do not use the Scripts directory as a reference to the top path;
it's not necessary and it will not work now that the Scripts directory is not
shipped in the tarball.
- 6:07 PM Changeset in webkit [155917] by
-
- 4 edits in branches/safari-537.60-branch/Source/WebCore
Merged r154915. <rdar://problem/14932478>
- 6:04 PM Changeset in webkit [155916] by
-
- 9 edits in branches/safari-537.60-branch/Source/WebCore
Merged r154914. <rdar://problem/14932478>
- 6:01 PM Changeset in webkit [155915] by
-
- 2 edits in trunk/Source/WTF
Fix WebKit1 build after r155910
Reviewed by Anders Carlsson.
- wtf/Platform.h: Some files in WebKit include system headers before
WTF headers, causing the macro to be defined twice. #undef the macro
to support that case.
- 6:00 PM Changeset in webkit [155914] by
-
- 2 edits in branches/safari-537.60-branch/Source/WebCore
Merged r154890. <rdar://problem/14932478>
- 5:31 PM Changeset in webkit [155913] by
-
- 2 edits in trunk/Source/WebKit2
Unreviewed build fix after 155905.
- Platform/CoreIPC/unix/ConnectionUnix.cpp:
(CoreIPC::Connection::sendOutgoingMessage): use OwnPtr instead of PassOwnPtr.
- 4:39 PM Changeset in webkit [155912] by
-
- 2 edits in trunk/Source/WTF
Disable OS X's unprefixed debug macro
https://bugs.webkit.org/show_bug.cgi?id=121460
Reviewed by Anders Carlsson.
- wtf/Platform.h: OS X defines a series of platform macros for debugging.
Some of them are really annoying because they use common names (e.g. check()).
Disable those macros so that we are not limited in how we name methods and functions.
- 4:32 PM Changeset in webkit [155911] by
-
- 8 edits in trunk/Source
Add RunLoop::isMain and use it in WebKit2
https://bugs.webkit.org/show_bug.cgi?id=121459
Reviewed by Anders Carlsson.
Source/WebCore:
- WebCore.exp.in:
- platform/RunLoop.cpp:
(WebCore::RunLoop::isMain):
- platform/RunLoop.h:
Source/WebKit2:
- UIProcess/API/gtk/WebKitFaviconDatabase.cpp:
(getIconSurfaceSynchronously):
(iconDataReadyForPageURLCallback):
(webkit_favicon_database_get_favicon_uri):
- UIProcess/API/mac/WKPrintingView.mm:
(-[WKPrintingView _delayedResumeAutodisplayTimerFired]):
(-[WKPrintingView _adjustPrintingMarginsForHeaderAndFooter]):
(pageDidDrawToImage):
(pageDidDrawToPDF):
(-[WKPrintingView _preparePDFDataForPrintingOnSecondaryThread]):
(pageDidComputePageRects):
(-[WKPrintingView _askPageToComputePageRects]):
(prepareDataForPrintingOnSecondaryThread):
(-[WKPrintingView knowsPageRange:]):
(-[WKPrintingView _drawPreview:]):
(-[WKPrintingView drawRect:]):
(-[WKPrintingView _drawPageBorderWithSizeOnMainThread:]):
(-[WKPrintingView drawPageBorderWithSize:]):
- UIProcess/WebProcessProxy.cpp:
(WebKit::globalPageMap):
- 4:30 PM Changeset in webkit [155910] by
-
- 2 edits in trunk/Source/WTF
Disable OS X's unprefixed debug macro
https://bugs.webkit.org/show_bug.cgi?id=121460
Reviewed by Anders Carlsson.
- wtf/Platform.h: OS X defines a serie of platform macros for debugging.
Some of them are really annoying because they use common names (e.g. check()).
Disable those macros so that we are not limited how we name methods and functions.
- 3:57 PM Changeset in webkit [155909] by
-
- 3 edits2 adds in trunk
AX: ARIA tablist is disabled, but VoiceOver does not speak the tabs as dimmed
https://bugs.webkit.org/show_bug.cgi?id=121408
Reviewed by Darin Adler.
Source/WebCore:
ARIA says that aria-disabled should propagate to children nodes.
Test: accessibility/aria-disabled-propagated-to-children.html
- accessibility/AccessibilityNodeObject.cpp:
(WebCore::AccessibilityNodeObject::isEnabled):
LayoutTests:
- accessibility/aria-disabled-propagated-to-children-expected.txt: Added.
- accessibility/aria-disabled-propagated-to-children.html: Added.
- 3:20 PM Changeset in webkit [155908] by
-
- 167 edits in trunk/Source/WebCore
CTTE: Element::createRenderer() should take references
https://bugs.webkit.org/show_bug.cgi?id=121449
Reviewed by Anders Carlsson.
*Many file elided*
- 3:16 PM Changeset in webkit [155907] by
-
- 2 edits in trunk/Source/WebCore
iOS build fix after r155638.
Unreviewed.
- editing/Editor.cpp:
- 3:01 PM Changeset in webkit [155906] by
-
- 5 edits in trunk/Source/WebCore
Make FloatingObjects own it's FloatingObject instances
https://bugs.webkit.org/show_bug.cgi?id=121323
Reviewed by Alexandru Chiculita.
As part of decoupling FloatingObjects from RenderBlock, change
FloatingObjects to properly manage the FloatingObject instances it
contains.
No new tests, no behavior change.
- rendering/FloatingObjects.cpp:
(WebCore::FloatingObject::FloatingObject): Make the constructors
private so that FloatingObjects can only be created with an OwnPtr.
Also make a RenderBox required to create a FloatingObject.
(WebCore::FloatingObject::create): Factory method to create a vanilla
FloatingObject.
(WebCore::FloatingObject::copyToNewContainer): Factory method to copy
an existing FloatingObject in the case it is overhanging or intruding
and needs to be copied to the block that it overhangs or intrudes
into.
(WebCore::FloatingObject::unsafeClone): Rename this method so it is
more obvious that it really shouldn't be used, and to make it more
obvious that one should use the copyToNewContainer method for all
normal FloatingObject copies.
(WebCore::FloatingObjects::clear): Delete all the FloatingObjects in
the set before clearing it.
(WebCore::FloatingObjects::moveAllToFloatInfoMap): Move all of the
FloatingObjects in the set to a RendererToFloatInfoMap. This is used
in RenderBlockFlow::clearFloats to when it is readding floats after
clearing the set.
(WebCore::FloatingObjects::add): Take an OwnPtr.
(WebCore::FloatingObjects::remove): Delete the removed FloatingObject.
- rendering/FloatingObjects.h: Remove FloatingObject::setRenderer(),
since the RenderBox must be set in the constructor.
- rendering/RenderBlock.cpp:
(WebCore::RenderBlock::moveAllChildrenIncludingFloatsTo): Rename clone
to unsafeClone.
(WebCore::RenderBlock::removeFloatingObjects): Don't delete anymore,
since clear does it.
(WebCore::RenderBlock::insertFloatingObject): Handle OwnPtr properly.
(WebCore::RenderBlock::removeFloatingObject): Don't delete anymore,
since remove does it.
(WebCore::RenderBlock::removeFloatingObjectsBelow): Ditto.
(WebCore::RenderBlock::addOverhangingFloats): Use copyToNewContainer
and OwnPtr.
(WebCore::RenderBlock::addIntrudingFloats): Ditto.
- rendering/RenderBlockFlow.cpp:
(WebCore::RenderBlockFlow::clearFloats): Use exportToFloatInfoMap.
- 3:00 PM Changeset in webkit [155905] by
-
- 4 edits in trunk/Source/WebKit2
Replace more uses of PassOwnPtr with OwnPtr in WebKit2
https://bugs.webkit.org/show_bug.cgi?id=121453
Reviewed by Sam Weinig.
- Platform/CoreIPC/Connection.cpp:
(CoreIPC::Connection::SecondaryThreadPendingSyncReply::SecondaryThreadPendingSyncReply):
(CoreIPC::Connection::createSyncMessageEncoder):
(CoreIPC::Connection::sendMessage):
(CoreIPC::Connection::sendSyncReply):
(CoreIPC::Connection::waitForMessage):
(CoreIPC::Connection::sendSyncMessage):
(CoreIPC::Connection::sendSyncMessageFromSecondaryThread):
(CoreIPC::Connection::waitForSyncReply):
(CoreIPC::Connection::processIncomingSyncReply):
(CoreIPC::Connection::processIncomingMessage):
(CoreIPC::Connection::enqueueIncomingMessage):
(CoreIPC::Connection::dispatchMessage):
- Platform/CoreIPC/Connection.h:
(CoreIPC::Connection::PendingSyncReply::PendingSyncReply):
- Platform/CoreIPC/mac/ConnectionMac.cpp:
(CoreIPC::Connection::sendOutgoingMessage):
- 2:59 PM Changeset in webkit [155904] by
-
- 2 edits in trunk/LayoutTests
Flaky Test: svg/batik/masking/maskRegions.svg
https://bugs.webkit.org/show_bug.cgi?id=114120
Fails in the same way as many other batik tests.
- platform/mac/TestExpectations: Marking as flaky.
- 2:46 PM Changeset in webkit [155903] by
-
- 2 edits in trunk/Source/WebCore
Fix creation of embedded JS and CSS files on cmake based ports.
https://bugs.webkit.org/show_bug.cgi?id=121448
Reviewed by Joseph Pecoraro.
Just a build fix, layout tests already test this issue.
- CMakeLists.txt:
- 2:35 PM Changeset in webkit [155902] by
-
- 2 edits in trunk/LayoutTests
Marked a few tests that are too slow in debug builds, and frequently fail.
- platform/mac/TestExpectations:
- 2:30 PM Changeset in webkit [155901] by
-
- 3 edits in trunk/Source/WebKit/win
[Windows] Rendering stops when mouse is moving for some types of animations.
https://bugs.webkit.org/show_bug.cgi?id=121329
Reviewed by Anders Carlsson.
- WebView.cpp:
(WebView::WebView): Add new display flag.
(WebView::repaint): Mark view as needing a display operation.
(WebView::addToDirtyRegion): Ditto
(WebView::scrollBackingStore): Ditto
(WebView::sizeChanged): Ditto
(WebView::updateBackingStore): Ditto
(WebView::performLayeredWindowUpdate): Mark the WebView as having been
moved to screen.
(WebView::paintIntoWindow): Ditto
(WebView::WebViewWndProc): If the view needs display, and it was not done during
this message loop iteration, draw the screen.
- WebView.h:
(WebView::needsDisplay): Added
- 2:09 PM Changeset in webkit [155900] by
-
- 7 edits in trunk/Source
MessageQueue should use a Deque of OwnPtrs
https://bugs.webkit.org/show_bug.cgi?id=121450
Reviewed by Andreas Kling.
Source/WebCore:
- Modules/webdatabase/DatabaseThread.cpp:
(WebCore::SameDatabasePredicate::operator()):
- dom/default/PlatformMessagePortChannel.h:
(WebCore::PlatformMessagePortChannel::MessagePortQueue::tryGetMessage):
- fileapi/FileThread.cpp:
(WebCore::SameInstancePredicate::operator()):
- workers/WorkerRunLoop.cpp:
(WebCore::ModePredicate::operator()):
Source/WTF:
- wtf/MessageQueue.h:
- 2:01 PM Changeset in webkit [155899] by
-
- 5 edits in releases/WebKitGTK/webkit-2.2
[GTK][2.2] Enable the Wayland target by default if the GTK+ Wayland dependency is available
https://bugs.webkit.org/show_bug.cgi?id=121414
Reviewed by Martin Robinson.
- Source/autotools/FindDependencies.m4: If the Wayland target is not strictly disabled, also check that the GTK+ Wayland
package (gtk+-wayland-3.0) has a recent-enough version. If the dependency could not be satisfied, either a warning or an
error is printed out, depending if the Wayland target was required or just optional.
- Source/autotools/PrintBuildConfiguration.m4: Add a new macro, AM_APPEND_TO_DESRIPTION, that's helpful when producing
a description string that can contain multiple items that should be nicely formatted when listed in that string. This is
then used when creating the description for all the enabled targets, so for instance 'x11, wayland' string would be displayed
if both the X11 and Wayland targets are enabled.
- Source/autotools/ReadCommandLineArguments.m4: The --with-target configuration options should default to 'x11,wayland' if
not given, with the Wayland target in that case not being a hard requirement (i.e. it should be disabled if the dependencies
for the Wayland target could not be found).
- Source/autotools/Versions.m4: The GTK+ Wayland dependency should be of version 3.9.14 or greater (the 3.10 release series).
- 1:39 PM Changeset in webkit [155898] by
-
- 4 edits in trunk
WTF::Deque should work with move only types
https://bugs.webkit.org/show_bug.cgi?id=121446
Reviewed by Andreas Kling.
Source/WTF:
- wtf/Deque.h:
Use std::move and std::forward where appropriate and get rid of the PassTraits include.
Tools:
- TestWebKitAPI/Tests/WTF/Deque.cpp:
(TestWebKitAPI::TEST):
Add a test.
- 1:34 PM Changeset in webkit [155897] by
-
- 2 edits in trunk/LayoutTests
svg/animations/smil-leak-*.svg tests are flaky
https://bugs.webkit.org/show_bug.cgi?id=114280
The tests are still flaky. Ran out of ideas, so marking them as such and leaving them there.
- platform/mac/TestExpectations:
- 1:24 PM Changeset in webkit [155896] by
-
- 2 edits in trunk/Source/WebInspectorUI
Web Inspector: TypeError when updating ResourceTreeElement created in strange order
https://bugs.webkit.org/show_bug.cgi?id=121382
Patch by Joseph Pecoraro <Joseph Pecoraro> on 2013-09-16
Reviewed by Timothy Hatcher.
Patch by Brian Burg. Fix an uncaught exception that can happen on navigations.
- UserInterface/ResourceTreeElement.js:
(WebInspector.ResourceTreeElement.prototype._updateStatus):
- 1:14 PM Changeset in webkit [155895] by
-
- 3 edits in trunk/Source/WebKit2
Unreviewed, fixing the GTK WK2 build after r155888.
- UIProcess/API/gtk/PageClientImpl.cpp:
(WebKit::PageClientImpl::createDrawingAreaProxy):
- UIProcess/API/gtk/PageClientImpl.h:
- 1:09 PM Changeset in webkit [155894] by
-
- 10 edits in trunk/Source/WebCore
CTTE: FrameTree::top() should return a reference.
<https://webkit.org/b/121445>
Reviewed by Anders Carlsson.
There's always a top frame in the tree.
- 1:01 PM Changeset in webkit [155893] by
-
- 1 edit14 adds in trunk/Source/WebCore
Unreviewed, fixing GObject bindings tests after r155850 by adding the *.symbols files that are now also generated.
- bindings/scripts/test/GObject/WebKitDOMTestActiveDOMObject.symbols: Added.
- bindings/scripts/test/GObject/WebKitDOMTestCallback.symbols: Added.
- bindings/scripts/test/GObject/WebKitDOMTestCustomNamedGetter.symbols: Added.
- bindings/scripts/test/GObject/WebKitDOMTestEventConstructor.symbols: Added.
- bindings/scripts/test/GObject/WebKitDOMTestEventTarget.symbols: Added.
- bindings/scripts/test/GObject/WebKitDOMTestException.symbols: Added.
- bindings/scripts/test/GObject/WebKitDOMTestInterface.symbols: Added.
- bindings/scripts/test/GObject/WebKitDOMTestMediaQueryListListener.symbols: Added.
- bindings/scripts/test/GObject/WebKitDOMTestNamedConstructor.symbols: Added.
- bindings/scripts/test/GObject/WebKitDOMTestNode.symbols: Added.
- bindings/scripts/test/GObject/WebKitDOMTestObj.symbols: Added.
- bindings/scripts/test/GObject/WebKitDOMTestOverloadedConstructors.symbols: Added.
- bindings/scripts/test/GObject/WebKitDOMTestSerializedScriptValueInterface.symbols: Added.
- bindings/scripts/test/GObject/WebKitDOMTestTypedefs.symbols: Added.
- 12:56 PM Changeset in webkit [155892] by
-
- 3 edits in trunk/Source/WebCore
Fix the build following <http://trac.webkit.org/changeset/155591>
(https://bugs.webkit.org/show_bug.cgi?id=121200)
Fix the build for ports that enable PLUGIN_PROXY_FOR_VIDEO, such as iOS WebKit.
- html/shadow/MediaControlElements.cpp:
(WebCore::MediaControlTextTrackContainerElement::updateSizes): Substitute isWidget() for isRenderWidget()
as the latter is undefined.
- rendering/RenderLayerCompositor.cpp:
(WebCore::RenderLayerCompositor::requiresCompositingForVideo): Ditto.
- 12:48 PM Changeset in webkit [155891] by
-
- 14 edits1 add in trunk/Source/JavaScriptCore
MarkedBlocks shouldn't be put in Allocated state if they didn't produce a FreeList
https://bugs.webkit.org/show_bug.cgi?id=121236
Reviewed by Geoffrey Garen.
Right now, after a collection all MarkedBlocks are in the Marked block state. When lazy sweeping
happens, if a block returns an empty free list after being swept, we call didConsumeFreeList(),
which moves the block into the Allocated block state. This happens to both the block that was
just being allocated out of (i.e. m_currentBlock) as well as any blocks who are completely full.
We should distinguish between these two cases: m_currentBlock should transition to
Allocated (because we were just allocating out of it) and any subsequent block that returns an
empty free list should transition back to the Marked state. This will make the block state more
consistent with the actual state the block is in, and it will also allow us to speed up moving
all blocks the the Marked state during generational collection.
Added new RAII-style HeapIterationScope class that notifies the Heap when it is about to be
iterated and when iteration has finished. Any clients that need accurate liveness data when
iterating over the Heap now need to use a HeapIterationScope so that the state of Heap can
be properly restored after they are done iterating. No new GC-allocated objects can be created
until this object goes out of scope.
- JavaScriptCore.xcodeproj/project.pbxproj:
- debugger/Debugger.cpp:
(JSC::Debugger::recompileAllJSFunctions): Added HeapIterationScope for the Recompiler iteration.
- heap/Heap.cpp:
(JSC::Heap::willStartIterating): Callback used by HeapIterationScope to indicate that iteration of
the Heap is about to begin. This will cause cell liveness data to be canonicalized by calling stopAllocating.
(JSC::Heap::didFinishIterating): Same, but indicates that iteration has finished.
(JSC::Heap::globalObjectCount): Used HeapIterationScope.
(JSC::Heap::objectTypeCounts): Ditto.
(JSC::Heap::markDeadObjects): Ditto.
(JSC::Heap::zombifyDeadObjects): Ditto.
- heap/Heap.h:
- heap/HeapIterationScope.h: Added. New RAII-style object for indicating to the Heap that it's about
to be iterated or that iteration has finished.
(JSC::HeapIterationScope::HeapIterationScope):
(JSC::HeapIterationScope::~HeapIterationScope):
- heap/HeapStatistics.cpp:
(JSC::HeapStatistics::showObjectStatistics): Used new HeapIterationScope.
- heap/MarkedAllocator.cpp:
(JSC::MarkedAllocator::tryAllocateHelper): We now treat the case where we have just finished
allocating out of the current block differently from the case where we sweep a block and it
returns an empty free list. This was the primary point of this patch.
(JSC::MarkedAllocator::allocateSlowCase): ASSERT that nobody is currently iterating the Heap
when allocating.
- heap/MarkedAllocator.h:
(JSC::MarkedAllocator::reset): All allocators are reset after every collection. We need to make
sure that the m_lastActiveBlock gets cleared, which it might not always because we don't call
takeCanonicalizedBlock on blocks in the large allocators.
(JSC::MarkedAllocator::stopAllocating): We shouldn't already have a last active block,
so ASSERT as much.
(JSC::MarkedAllocator::resumeAllocating): Do the opposite of what stopAllocating
does. So, if we don't have a m_lastActiveBlock then we don't have to worry about undoing anything
done by stopAllocating. If we do, then we call resumeAllocating on the block, which returns the FreeList
as it was prior to stopping allocation. We then set the current block to the last active block and
clear the last active block.
- heap/MarkedBlock.cpp:
(JSC::MarkedBlock::resumeAllocating): Any block resuming allocation should be in
the Marked state, so ASSERT as much. We always allocate a m_newlyAllocated Bitmap if we're
FreeListed, so if we didn't allocate one then we know we were Marked when allocation was stopped,
so just return early with an empty FreeList. If we do have a non-null m_newlyAllocated Bitmap
then we need to be swept in order to rebuild our FreeList.
- heap/MarkedBlock.h:
(JSC::MarkedBlock::didConsumeEmptyFreeList): This is called if we ever sweep a block and get back
an empty free list. Instead of transitioning to the Allocated state, we now go straight back to the
Marked state. This makes sense because we weren't actually allocated out of, so we shouldn't be in
the allocated state. Also added some ASSERTs to make sure that we're in the state that we expect: all of
our mark bits should be set and we should not have a m_newlyAllocated Bitmap.
- heap/MarkedSpace.cpp:
(JSC::MarkedSpace::MarkedSpace):
(JSC::MarkedSpace::forEachAllocator): Added a new functor-style iteration method so that we can
easily iterate over each allocator for, e.g., stopping and resuming allocators without
duplicating code.
(JSC::StopAllocatingFunctor::operator()): New functors for use with forEachAllocator.
(JSC::MarkedSpace::stopAllocating): Ditto.
(JSC::ResumeAllocatingFunctor::operator()): Ditto.
(JSC::MarkedSpace::resumeAllocating): Ditto.
(JSC::MarkedSpace::willStartIterating): Callback that notifies MarkedSpace that it is being iterated.
Does some ASSERTs, sets a flag, canonicalizes cell liveness data by calling stopAllocating.
(JSC::MarkedSpace::didFinishIterating): Ditto, but to signal that iteration has completed.
- heap/MarkedSpace.h:
(JSC::MarkedSpace::iterationInProgress): Returns true if a HeapIterationScope is currently active.
(JSC::MarkedSpace::forEachLiveCell): Accepts a HeapIterationScope to enforce the rule that you have to
create one prior to iterating over the Heap.
(JSC::MarkedSpace::forEachDeadCell): Ditto.
- runtime/JSGlobalObject.cpp:
(JSC::JSGlobalObject::haveABadTime): Changed to use new HeapIterationScope.
- runtime/VM.cpp:
(JSC::VM::releaseExecutableMemory): Ditto.
- 12:47 PM WebKitGTK/2.2.x edited by
- (diff)
- 12:36 PM Changeset in webkit [155890] by
-
- 5 edits in trunk/LayoutTests
svg/animations/smil-leak-*.svg tests are flaky
https://bugs.webkit.org/show_bug.cgi?id=114280
The tests were still flaky. Changed retry timeout from 0 to 100 ms as previously
suggested by Geoff.
- svg/animations/smil-leak-dynamically-added-element-instances.svg:
- svg/animations/smil-leak-element-instances-noBaseValRef.svg:
- svg/animations/smil-leak-element-instances.svg:
- svg/animations/smil-leak-elements.svg:
- 12:06 PM WebKitGTK/WebKit2Roadmap edited by
- (diff)
- 11:58 AM Changeset in webkit [155889] by
-
- 3 edits in trunk/Source/JavaScriptCore
Inlining should work in debug mode (i.e. Executable::newCodeBlock() should call recordParse())
https://bugs.webkit.org/show_bug.cgi?id=121444
Reviewed by Mark Hahnenberg.
- dfg/DFGArgumentPosition.h: Fix a bug discovered by reenabling inlining. ArgumentPosition may point to the non-canonical VariableAccessData but users of someVariable() want the canonical one.
(JSC::DFG::ArgumentPosition::someVariable):
- runtime/Executable.cpp: Call recordParse() so that the Executable knows things about itself (like if it has captured variables). Otherwise those fields are uninitialized.
(JSC::ScriptExecutable::newCodeBlockFor):
- 11:57 AM Changeset in webkit [155888] by
-
- 17 edits in trunk/Source/WebKit2
Remove some uses of PassOwnPtr in WebKit2
https://bugs.webkit.org/show_bug.cgi?id=121443
Reviewed by Andreas Kling.
- Scripts/webkit2/messages.py:
(message_to_struct_declaration):
(generate_message_handler):
- Scripts/webkit2/messages_unittest.py:
- UIProcess/API/mac/FindIndicatorWindow.h:
- UIProcess/API/mac/FindIndicatorWindow.mm:
- UIProcess/API/mac/PageClientImpl.h:
- UIProcess/API/mac/PageClientImpl.mm:
(WebKit::PageClientImpl::createDrawingAreaProxy):
- UIProcess/API/mac/WKView.mm:
(-[WKView WebKit::]):
(-[WKView _setFindIndicator:fadeOut:animate:]):
- UIProcess/API/mac/WKViewInternal.h:
- UIProcess/DrawingAreaProxyImpl.cpp:
- UIProcess/DrawingAreaProxyImpl.h:
- UIProcess/PageClient.h:
- UIProcess/mac/TiledCoreAnimationDrawingAreaProxy.h:
- UIProcess/mac/TiledCoreAnimationDrawingAreaProxy.mm:
- WebProcess/InjectedBundle/API/c/WKBundlePageOverlay.cpp:
(PageOverlayClientImpl::PageOverlayClientImpl):
(WKBundlePageOverlayCreate):
- WebProcess/Plugins/PluginView.cpp:
(WebKit::PluginView::createWebEvent):
- WebProcess/Plugins/PluginView.h:
- 11:26 AM Changeset in webkit [155887] by
-
- 2 edits in trunk/Source/WebCore
Build fix following <https://trac.webkit.org/r154358>
(https://bugs.webkit.org/show_bug.cgi?id=120078)
Fix the build for ports that enable PLUGIN_PROXY_FOR_VIDEO, such as iOS WebKit.
- html/HTMLMediaElement.cpp:
(WebCore::HTMLMediaElement::rendererIsNeeded): Substitute "style" for "context".
- 11:21 AM Changeset in webkit [155886] by
-
- 9 edits in trunk/Source
Make InspectorTypeBuilder generate better enums in C++.
https://bugs.webkit.org/show_bug.cgi?id=121440
Reviewed by Joseph Pecoraro.
Source/WebCore:
- inspector/CodeGeneratorInspector.py:
(fix_camel_case): Add more abbrivations.
(TypeBindings.create_type_declaration_.EnumBinding.get_code_generator.CodeGenerator.generate_type_builder): Use fix_camel_case
- inspector/ConsoleMessage.cpp:
(WebCore::messageSourceValue):
(WebCore::messageTypeValue):
- inspector/InspectorCSSAgent.cpp:
(WebCore::InspectorCSSAgent::detectOrigin):
- inspector/InspectorRuntimeAgent.cpp:
(WebCore::InspectorRuntimeAgent::parse):
- inspector/InspectorStyleSheet.cpp:
(WebCore::InspectorStyleSheet::resourceStyleSheetText):
- inspector/InspectorStyleSheet.h:
(WebCore::InspectorStyleSheet::canBind):
Source/WebInspectorUI:
- UserInterface/InspectorBackendCommands.js: Updated to make Css be CSS.
- 11:09 AM Changeset in webkit [155885] by
-
- 2 edits in trunk/LayoutTests
Flaky Test: svg/batik/filters/feTile.svg
https://bugs.webkit.org/show_bug.cgi?id=114375
- platform/mac/TestExpectations: Marking as flaky. We already have a lot of these
batik tests marked as flaky, with similar symptoms.
- 10:56 AM Changeset in webkit [155884] by
-
- 2 edits in trunk/Source/JavaScriptCore
Aligned argument signatures of setupArgumentsWithExecState are missing on MIPS.
https://bugs.webkit.org/show_bug.cgi?id=121439
Patch by Balazs Kilvady <kilvadyb@homejinni.com> on 2013-09-16
Reviewed by Geoffrey Garen.
Missing implementations of setupArgumentsWithExecState added.
- dfg/DFGCCallHelpers.h:
(JSC::DFG::CCallHelpers::setupArgumentsWithExecState):
- 10:50 AM Changeset in webkit [155883] by
-
- 8 edits in trunk/Source/WTF
Change a couple of COMPILE_ASSERTs to static_assert
https://bugs.webkit.org/show_bug.cgi?id=121441
Reviewed by Andreas Kling.
- wtf/BloomFilter.h:
- wtf/PackedIntVector.h:
(WTF::PackedIntVector::PackedIntVector):
- wtf/StdLibExtras.h:
(WTF::bitwise_cast):
(WTF::safeCast):
(WTF::roundUpToMultipleOf):
- wtf/StringHasher.h:
(WTF::StringHasher::hashMemory):
- wtf/Vector.h:
- wtf/text/AtomicString.cpp:
- wtf/unicode/Unicode.h:
- 10:40 AM Changeset in webkit [155882] by
-
- 5 edits in trunk/LayoutTests
svg/animations/smil-leak-*.svg tests are flaky
https://bugs.webkit.org/show_bug.cgi?id=114280
Reviewed by Geoffrey Garen.
Try garbage collection multiple times, who knows which queues the deleted nodes
could still be in. Also, removed last week's experimental change to make 10x more
nodes to see how that affects the number of leaks (it didn't).
- svg/animations/smil-leak-dynamically-added-element-instances.svg:
- svg/animations/smil-leak-element-instances-noBaseValRef.svg:
- svg/animations/smil-leak-element-instances.svg:
- svg/animations/smil-leak-elements.svg:
- 10:21 AM Changeset in webkit [155881] by
-
- 3 edits in trunk/Source/WebCore
MediaStream API: Storing the constraints in MediaStreamSource
https://bugs.webkit.org/show_bug.cgi?id=120882
based on: https://chromium.googlesource.com/chromium/blink/+/04ac7655b54ae98f55774afde3f8e92b6c6302e6
Patch by Thiago de Barros Lacerda <thiago.lacerda@openbossa.org> on 2013-09-16
Reviewed by Eric Carlson.
No new tests needed.
- Modules/mediastream/UserMediaRequest.cpp:
(WebCore::UserMediaRequest::succeed):
- platform/mediastream/MediaStreamSource.h:
(WebCore::MediaStreamSource::setConstraints):
(WebCore::MediaStreamSource::constraints):
- 10:21 AM Changeset in webkit [155880] by
-
- 2 edits in releases/WebKitGTK/webkit-2.2/Source/WebCore/platform/gtk/po
Merge 155867 - Add new Galician translations
https://bugs.webkit.org/show_bug.cgi?id=105079
Patch by Fran Dieguez <frandieguez@gnome.org> on 2013-09-16
Reviewed by Gustavo Noronha Silva.
- gl.po: updated
- 10:21 AM Changeset in webkit [155879] by
-
- 2 edits in releases/WebKitGTK/webkit-2.2/Source/WebCore/platform/gtk/po
Merge 155866 - L10n - en_GB PO file for WebKitGTK+
https://bugs.webkit.org/show_bug.cgi?id=100255
Patch by Chris Leonard <cjlhomeaddress@gmail.com> on 2013-09-16
Reviewed by Gustavo Noronha Silva.
- en_GB.po: updated.
- 10:20 AM Changeset in webkit [155878] by
-
- 2 edits in releases/WebKitGTK/webkit-2.2/Source/WebCore/platform/gtk/po
Merge 155869 - Updated spanish Translation
https://bugs.webkit.org/show_bug.cgi?id=100677
Patch by Daniel Mustieles <daniel.mustieles@gmail.com> on 2013-09-16
Reviewed by Gustavo Noronha Silva.
- es.po: updated.
- 10:16 AM Changeset in webkit [155877] by
-
- 2 edits in trunk/LayoutTests
http/tests/misc/submit-post-keygen.html is extremely slow on bots, often timing out
https://bugs.webkit.org/show_bug.cgi?id=121331
Marking as possibly failing too, because when DRT detects a timeout, run-webkit-tests
thinks that it's a failure.
- platform/mac/TestExpectations:
- 10:13 AM Changeset in webkit [155876] by
-
- 2 edits in trunk/Source/JavaScriptCore
[sh4] Fix typo in subp implementation in LLINT.
https://bugs.webkit.org/show_bug.cgi?id=121438
Patch by Julien Brianceau <jbriance@cisco.com> on 2013-09-16
Reviewed by Andreas Kling.
- offlineasm/sh4.rb:
- 10:12 AM Changeset in webkit [155875] by
-
- 2 edits in trunk/Source/WebCore
Rebaseline bindings tests after Sam's Document& changes.
- 10:07 AM Changeset in webkit [155874] by
-
- 5 edits in trunk/Source/WebCore
Destroying a Document's render tree shouldn't make it impossible to recreate.
<https://webkit.org/b/121437>
Reviewed by Antti Koivisto.
Rename Document::detach() to destroyRenderTree() and stop automatically
disconnecting Document from its Frame after the deforestation.
Added Document::disconnectFromFrame() and do that after every call to
destroyRenderTree() that we currently have.
This change doesn't alter any behavior on its own, but is a step towards
being able to destroy and rebuild the render tree.
- 9:35 AM Changeset in webkit [155873] by
-
- 2 edits in releases/WebKitGTK/webkit-2.2/Source/WebKit2
Merge r155861 - Web Inspector: Do not try to parse incomplete HTTP requests
https://bugs.webkit.org/show_bug.cgi?id=121123
Patch by Andre Moreira Magalhaes <Andre Moreira Magalhaes> on 2013-09-16
Reviewed by Carlos Garcia Campos.
Update to incorporate additional review suggestions.
- UIProcess/API/gtk/tests/TestInspectorServer.cpp:
(sendIncompleteRequest):
Fix memory leaks, change test timeout from 2 seconds to 1 second,
use "0" instead of "NULL" and use g_assert_no_error when checking
for GError.
- 9:32 AM Changeset in webkit [155872] by
-
- 3 edits in releases/WebKitGTK/webkit-2.2/Source/WebKit2
Merge r155642 - Web Inspector: Do not try to parse incomplete HTTP requests
https://bugs.webkit.org/show_bug.cgi?id=121123
Patch by Andre Moreira Magalhaes <Andre Moreira Magalhaes> on 2013-09-12
Reviewed by Darin Adler.
When working on a patch for bug #121121 I found an issue with the InspectorServer where it would try
to parse an HTTP message before receiving the full message and thus fail connecting with the
chromedevtools plugin.
What happens is that the WebSocketServerConnection receives buffers on
WebSocketServerConnection::didReceiveSocketStreamData and calls
WebSocketServerConnection::readHTTPMessage which then checks if we have a valid request by calling
HTTPRequest::parseHTTPRequestFromBuffer. If the request is valid it tries to parse the message and
clears the buffer, otherwise it continues adding data to the internal buffer until we have a valid
request.
The problem is that currently HTTPRequest::parseHTTPRequestFromBuffer considers the request as valid
before receiving the full message. To solve this we should make the method check if the request
headers end with a blank line otherwise we consider the request as invalid (see also
http://www.w3.org/Protocols/rfc2616/rfc2616-sec4.html).
- UIProcess/API/gtk/tests/TestInspectorServer.cpp:
(sendIncompleteRequest):
(beforeAll):
Add GTK specific test to check if the inspector server replies to incomplete requests.
- UIProcess/InspectorServer/HTTPRequest.cpp:
(WebKit::HTTPRequest::parseHeaders):
Do not consider request valid if headers didn't end with a blank line.
- 9:29 AM Changeset in webkit [155871] by
-
- 2 edits in trunk/Source/WebCore
Unreviewed warning correction. Avoid uninitialized renderer.
- css/CSSComputedStyleDeclaration.cpp:
(WebCore::ComputedStyleExtractor::propertyValue): Renderer can be uninitialized.
- 9:24 AM Changeset in webkit [155870] by
-
- 2 edits in trunk/Source/WebCore
Unreviewed warning correction.
- css/CSSParser.cpp:
(WebCore::CSSParser::parseFilter): Don't use 'value' as the input argument
to the method, and the internal loop variable for the parser.
- 9:21 AM Changeset in webkit [155869] by
-
- 2 edits in trunk/Source/WebCore/platform/gtk/po
Updated spanish Translation
https://bugs.webkit.org/show_bug.cgi?id=100677
Patch by Daniel Mustieles <daniel.mustieles@gmail.com> on 2013-09-16
Reviewed by Gustavo Noronha Silva.
- es.po: updated.
- 9:19 AM Changeset in webkit [155868] by
-
- 2 edits in trunk/Source/WebCore
[Win] Compile errors in WebCore derived sources.
https://bugs.webkit.org/show_bug.cgi?id=121420
Patch by peavo@outlook.com <peavo@outlook.com> on 2013-09-16
Reviewed by Brent Fulgham.
- dom/make_names.pl:
(printNamesCppFile): Added const modifier.
(printDefinitions): Use C style cast.
- 9:18 AM Changeset in webkit [155867] by
-
- 2 edits in trunk/Source/WebCore/platform/gtk/po
Add new Galician translations
https://bugs.webkit.org/show_bug.cgi?id=105079
Patch by Fran Dieguez <frandieguez@gnome.org> on 2013-09-16
Reviewed by Gustavo Noronha Silva.
- gl.po: updated
- 9:16 AM Changeset in webkit [155866] by
-
- 2 edits in trunk/Source/WebCore/platform/gtk/po
L10n - en_GB PO file for WebKitGTK+
https://bugs.webkit.org/show_bug.cgi?id=100255
Patch by Chris Leonard <cjlhomeaddress@gmail.com> on 2013-09-16
Reviewed by Gustavo Noronha Silva.
- en_GB.po: updated.
- 9:16 AM Changeset in webkit [155865] by
-
- 5 edits8 adds in trunk
[CSS Shapes] Winding rule polygon issues
https://bugs.webkit.org/show_bug.cgi?id=120236
Reviewed by Dirk Schulze.
Source/WebCore:
This patch corrects some of the problems with shape-inside and
self-intersecting polygons and eliminates the layout flashing behavior
that prompted the original bug report. The FloatPolygon::contains()
method now respoects the polygon's fillRule and the PolygonShape code
used to find shape-inside intervals for the top and bottom of a line
now also handles fillRule correctly.
Tests: fast/shapes/shape-inside/shape-inside-complex-polygon-001.html
fast/shapes/shape-inside/shape-inside-complex-polygon-002.html
fast/shapes/shape-inside/shape-inside-complex-polygon-003.html
fast/shapes/shape-inside/shape-inside-complex-polygon-004.html
- platform/graphics/FloatPolygon.cpp:
(WebCore::FloatPolygon::containsEvenOdd):
(WebCore::FloatPolygon::containsNonZero):
(WebCore::FloatPolygon::contains):
- platform/graphics/FloatPolygon.h:
- rendering/shapes/PolygonShape.cpp:
(WebCore::computeXIntersections):
LayoutTests:
Verify that self-intersecting rectilinear shape-inside polygons work
correctly for the evenodd and nonzero fill rules.
The shape for tests 001 and 002 contains a loop (the vertices are
numbered in the figure):
0----4----1----5
| | | |
| 3----2 |
7--------------6
When the fillRule is nonzero the entire rectangle is filled because
edges 7-0 and 3-4 wind in same direction. When fillRule is evenodd the
1-2-3-4 rectangle is not filled, leaving a U shape.
The shape for tests 003 and 004 is similar, but the adjacent vertical edges
wind in opposite directions. As a result both nonzero and evenodd fills
produce the same U shaped result.
0----2---1,5---6
| | | |
| 3----4 |
8--------------7
- fast/shapes/shape-inside/shape-inside-complex-polygon-001-expected.html: Added.
- fast/shapes/shape-inside/shape-inside-complex-polygon-001.html: Added.
- fast/shapes/shape-inside/shape-inside-complex-polygon-002-expected.html: Added.
- fast/shapes/shape-inside/shape-inside-complex-polygon-002.html: Added.
- fast/shapes/shape-inside/shape-inside-complex-polygon-003-expected.html: Added.
- fast/shapes/shape-inside/shape-inside-complex-polygon-003.html: Added.
- fast/shapes/shape-inside/shape-inside-complex-polygon-004-expected.html: Added.
- fast/shapes/shape-inside/shape-inside-complex-polygon-004.html: Added.
- 9:01 AM Changeset in webkit [155864] by
-
- 16 edits1 delete in releases/WebKitGTK/webkit-2.2
Merge r155534 - [GTK] Remove Gail dependency from build system for GTK3
https://bugs.webkit.org/show_bug.cgi?id=119673
Reviewed by Gustavo Noronha Silva.
.:
- Source/autotools/FindDependencies.m4: Don't look for GAIL at all.
- Source/autotools/Versions.m4: Removed any reference to GAIL.
- Source/cmake/FindGAIL3.cmake: Removed.
- Source/cmake/OptionsGTK.cmake: Don't look for the GAIL package.
Source/WebCore:
- GNUmakefile.am: Removed GAIL_CFLAGS.
- PlatformGTK.cmake: Removed GAIL3_INCLUDE_DIRS and GAIL3_LIBRARIES.
Source/WebKit/gtk:
- GNUmakefile.am: Removed GAIL_CFLAGS and GAIL_LIBS.
Source/WebKit2:
- GNUmakefile.am: Removed GAIL_LIBS.
Tools:
- EWSTools/ubuntu-ews-packages: Removed libgail-dev package.
- Scripts/webkitpy/port/gtk.py:
(GtkPort.setup_environ_for_server): Do not load the 'gail' module anymore.
- TestWebKitAPI/GNUmakefile.am: Removed GAIL_LIBS.
- 9:00 AM WebKitGTK/2.2.x edited by
- (diff)
- 8:56 AM Changeset in webkit [155863] by
-
- 2 edits in releases/WebKitGTK/webkit-2.2/Source/WebCore
Merge r155520 - [GTK] Get rid of Pango/Gail dependencies in accessibility for ATK
https://bugs.webkit.org/show_bug.cgi?id=114867
Reviewed by Martin Robinson.
Removed all trace of Gail and Pango specific code from the AtkText
implementation, now everything has been reimplemented.
- accessibility/atk/WebKitAccessibleInterfaceText.cpp:
(webkitAccessibleTextGetTextForOffset): Removed fallback code
relying in Gail/Pango, now all the related code has been
removed. Also, replaced the collection of if statements with a
switch, for better readability of the code.
- 8:55 AM WebKitGTK/2.2.x edited by
- (diff)
- 8:53 AM Changeset in webkit [155862] by
-
- 4 edits in releases/WebKitGTK/webkit-2.2/Source
Merge r155516 - [GTK] Reimplement atk_text_get_text_*_offset for LINE boundaries
https://bugs.webkit.org/show_bug.cgi?id=114872
Reviewed by Gustavo Noronha Silva.
Source/WebCore:
Re-implement these functions without using GailTextUtil nor Pango.
- accessibility/atk/WebKitAccessibleInterfaceText.cpp:
(lineAtPositionForAtkBoundary): New helper function to find the
line at a given position considering values of AtkTextBoundary.
(webkitAccessibleTextLineForBoundary): New function,
implementing atk_text_get_text_*_offset for LINE.
(webkitAccessibleTextGetTextForOffset): Replace usage of Gail for
LINE boundaries with webkitAccessibleTextLineForBoundary().
Source/WebKit/gtk:
Fixed wrong unit test.
- tests/testatk.c:
(testWebkitAtkGetTextAtOffsetWithPreformattedText): This test was
reporting a trailing '\n' for some reason for a <pre> block, which
is plainly wrong since, in order to return that, there should be
at least a trailing empty space after that and before the </pre>
closing tag. This is fixed now.
(testWebkitAtkGetTextAtOffsetWithWrappedLines): Uncommented tests
that were previously not passing due to a bug in GailTextUtil.
- 8:52 AM WebKitGTK/2.2.x edited by
- (diff)
- 8:49 AM Changeset in webkit [155861] by
-
- 2 edits in trunk/Source/WebKit2
Web Inspector: Do not try to parse incomplete HTTP requests
https://bugs.webkit.org/show_bug.cgi?id=121123
Patch by Andre Moreira Magalhaes <Andre Moreira Magalhaes> on 2013-09-16
Reviewed by Carlos Garcia Campos.
Update to incorporate additional review suggestions.
- UIProcess/API/gtk/tests/TestInspectorServer.cpp:
(sendIncompleteRequest):
Fix memory leaks, change test timeout from 2 seconds to 1 second,
use "0" instead of "NULL" and use g_assert_no_error when checking
for GError.
- 8:49 AM Changeset in webkit [155860] by
-
- 2 edits in releases/WebKitGTK/webkit-2.2/Source/WebCore
Merge r155031 - [GTK] Reimplement atk_text_get_text_*_offset for SENTENCE boundaries
https://bugs.webkit.org/show_bug.cgi?id=114873
Reviewed by Chris Fleizach.
Re-implement these functions without using GailTextUtil nor Pango.
- accessibility/atk/WebKitAccessibleInterfaceText.cpp:
(webkitAccessibleTextWordForBoundary): Renamed from webkitAccessibleTextGetWordForBoundary,
to keep it consistent with names for new functions.
(isSentenceBoundary): Helper function to know when we are either
at the beginning or the end of a sentence.
(isWhiteSpaceBetweenSentences): It returns true if we are in the
middle of a white space between sentences. Useful for implementing
the SENTENCE_END boundary type.
(sentenceAtPositionForAtkBoundary): New helper function to find the
sentence at a given position considering values of AtkTextBoundary.
(webkitAccessibleTextSentenceForBoundary): New function,
implementing atk_text_get_text_*_offset for SENTENCE.
(webkitAccessibleTextGetTextForOffset): Replace usage of Gail for
SENTENCE boundaries with webkitAccessibleTextSentenceForBoundary().
- 8:48 AM WebKitGTK/2.2.x edited by
- (diff)
- 8:17 AM Changeset in webkit [155859] by
-
- 5 edits in trunk/Source/WebCore
AXObjectCache::m_document should be a reference.
<https://webkit.org/b/121425>
Reviewed by Antti Koivisto.
AXObjectCache is always owned by a Document so remove the uncertainty
by turning its m_document into a Document&.
- 8:17 AM Changeset in webkit [155858] by
-
- 2 edits in releases/WebKitGTK/webkit-2.2/Source/WebCore/platform/gtk/po
Merge 155855 - [l10n] [pt_BR] Updated Brazilian Portuguese translation of WebKitGTK+ - 12/09/2013
https://bugs.webkit.org/show_bug.cgi?id=121255
Patch by Enrico Nicoletto <liverig@gmail.com> on 2013-09-16
Reviewed by Gustavo Noronha.
- pt_BR.po: updated.
- 8:16 AM Changeset in webkit [155857] by
-
- 2 edits in releases/WebKitGTK/webkit-2.2/Source/WebCore/platform/gtk/po
Merge 155831 - [l10n] Updated Polish translation of WebKitGTK+
https://bugs.webkit.org/show_bug.cgi?id=121392
Patch by Piotr Drąg <piotrdrag@gmail.com> on 2013-09-15
Reviewed by Gustavo Noronha Silva.
- pl.po: updated.
- 8:16 AM Changeset in webkit [155856] by
-
- 2 edits in releases/WebKitGTK/webkit-2.2/Source/WebCore/platform/gtk/po
Merge 154677 - [l10n] Updated Polish translation of WebKitGTK+
https://bugs.webkit.org/show_bug.cgi?id=119986
Patch by Piotr Drąg <piotrdrag@gmail.com> on 2013-08-27
Reviewed by Gustavo Noronha Silva.
- pl.po: updated.
- 8:16 AM WebKitGTK/2.2.x edited by
- (diff)
- 8:03 AM Changeset in webkit [155855] by
-
- 2 edits in trunk/Source/WebCore/platform/gtk/po
[l10n] [pt_BR] Updated Brazilian Portuguese translation of WebKitGTK+ - 12/09/2013
https://bugs.webkit.org/show_bug.cgi?id=121255
Patch by Enrico Nicoletto <liverig@gmail.com> on 2013-09-16
Reviewed by Gustavo Noronha.
- pt_BR.po: updated.
- 7:00 AM Changeset in webkit [155854] by
-
- 4 edits4 moves in trunk
Unreviewed, rolling out r155851.
http://trac.webkit.org/changeset/155851
https://bugs.webkit.org/show_bug.cgi?id=121429
checkbox states should be exposed through AtkState, not
AtkValue (Requested by msanchez on #webkit).
Source/WebCore:
- accessibility/atk/WebKitAccessibleInterfaceValue.cpp:
(webkitAccessibleValueGetCurrentValue):
- accessibility/atk/WebKitAccessibleWrapperAtk.cpp:
(getInterfaceMaskFromObject):
LayoutTests:
- platform/mac/accessibility/mixed-checkbox-expected.txt: Renamed from LayoutTests/accessibility/mixed-checkbox-expected.txt.
- platform/mac/accessibility/mixed-checkbox.html: Renamed from LayoutTests/accessibility/mixed-checkbox.html.
- platform/mac/accessibility/native-vs-nonnative-checkboxes-expected.txt: Renamed from LayoutTests/accessibility/native-vs-nonnative-checkboxes-expected.txt.
- platform/mac/accessibility/native-vs-nonnative-checkboxes.html: Renamed from LayoutTests/accessibility/native-vs-nonnative-checkboxes.html.
- 6:54 AM Changeset in webkit [155853] by
-
- 4 edits4 adds in trunk/LayoutTests
[Qt] Unreviewed gardening. Skipping and rebase failing tests on Qt bots.
Patch by Gabor Abraham <abrhm@inf.u-szeged.hu> on 2013-09-16
- platform/qt-wk1/TestExpectations:
- platform/qt-wk1/fast/events/before-unload-returnValue-expected.txt: Added r155367.
- platform/qt-wk1/inspector/console/command-line-api-expected.txt:
- platform/qt/TestExpectations:
- platform/qt/compositing/overflow/clipping-behaviour-change-is-not-propagated-to-descendants-expected.txt: Added r155607.
- platform/qt/compositing/overflow/clipping-behaviour-change-is-not-propagated-to-descendants2-expected.txt: Added r155607.
- 6:38 AM WebKitGTK/2.2.x edited by
- (diff)
- 6:25 AM WebKitGTK/WebKit2Roadmap edited by
- (diff)
- 5:51 AM Changeset in webkit [155852] by
-
- 5 edits2 adds in trunk/Source/WebKit2
[EFL][WK2] Implement pan and flick gesture.
https://bugs.webkit.org/show_bug.cgi?id=107101
Patch by Eunmi Lee <eunmi15.lee@samsung.com> on 2013-09-16
Reviewed by Gyuyoung Kim.
Implement pan gesture to scroll page by movement of point and stop
scrolling smoothly using ease-in-out-quad algorithm after touched
point is removed from the screen.
Use Ecore_Animator in order to process movement once per each framing
time slot instead of processing all movement.
- PlatformEfl.cmake:
- UIProcess/API/efl/EwkView.cpp:
(EwkView::scrollBy):
- UIProcess/API/efl/EwkView.h:
- UIProcess/API/efl/GestureRecognizer.cpp:
(WebKit::GestureHandler::view):
(WebKit::GestureHandler::GestureHandler):
(WebKit::GestureHandler::~GestureHandler):
(WebKit::GestureHandler::reset):
(WebKit::GestureHandler::panAnimatorCallback):
(WebKit::GestureHandler::handlePanStarted):
(WebKit::GestureHandler::handlePan):
(WebKit::GestureHandler::handlePanFinished):
(WebKit::GestureHandler::flickAnimatorCallback):
(WebKit::GestureHandler::handleFlick):
(WebKit::GestureRecognizer::noGesture):
(WebKit::GestureRecognizer::reset):
- UIProcess/efl/EasingCurves.cpp: Added.
(WebKit::easeInOutQuad):
- UIProcess/efl/EasingCurves.h: Added.
- 5:38 AM Changeset in webkit [155851] by
-
- 4 edits4 moves in trunk
[ATK] Extends atk value interface to return proper checkbox states
https://bugs.webkit.org/show_bug.cgi?id=121413
Patch by Krzysztof Czech <k.czech@samsung.com> on 2013-09-16
Reviewed by Mario Sanchez Prada.
Source/WebCore:
Tests: accessibility/mixed-checkbox.html
accessibility/native-vs-nonnative-checkboxes.html
- accessibility/atk/WebKitAccessibleInterfaceValue.cpp:
(webkitAccessibleValueValueForAccessibilityObject):
(webkitAccessibleValueGetCurrentValue):
- accessibility/atk/WebKitAccessibleWrapperAtk.cpp:
(getInterfaceMaskFromObject):
LayoutTests:
Sharing mac tests with other ports (GTK/EFL).
- accessibility/mixed-checkbox-expected.txt: Renamed from LayoutTests/platform/mac/accessibility/mixed-checkbox-expected.txt.
- accessibility/mixed-checkbox.html: Renamed from LayoutTests/platform/mac/accessibility/mixed-checkbox.html.
- accessibility/native-vs-nonnative-checkboxes-expected.txt: Renamed from LayoutTests/platform/mac/accessibility/native-vs-nonnative-checkboxes-expected.txt.
- accessibility/native-vs-nonnative-checkboxes.html: Renamed from LayoutTests/platform/mac/accessibility/native-vs-nonnative-checkboxes.html.
- 5:28 AM Changeset in webkit [155850] by
-
- 4 edits5 adds in trunk/Source/WebCore
[GTK] Check DOM bindings API compatibility while building
https://bugs.webkit.org/show_bug.cgi?id=101224
Reviewed by Gustavo Noronha Silva.
A .symbols file is now generated for every DOM binding public API
object. It contains the signature of all public methods. We keep a
single file with the signature of all public methods in the source
tree. When DOM bindings are generated, a new symbols file is
created concatenating all .symbols files generated and it's
compared with the symbols file in the source tree using a script
that checks if the changes are API compatible or not. In case of
API break the build finishes showing the differences found in the
symbol files. If API compatible changes are found, the diff is
shown in stdout, suggesting to run the gobject-run-api-break-test
with the --reset-results option to update the symbols file, and
the build continues. If there aren't API changes the build
continues silently.
- GNUmakefile.am:
- bindings/gobject/GNUmakefile.am:
- bindings/gobject/WebKitDOMCustom.symbols: Added.
- bindings/gobject/WebKitDOMEventTarget.symbols: Added.
- bindings/gobject/WebKitDOMObject.symbols: Added.
- bindings/gobject/webkitdom.symbols: Added.
- bindings/scripts/CodeGeneratorGObject.pm:
(GenerateFunction):
(WriteData):
- bindings/scripts/gobject-run-api-break-test: Added.
(check_api):
- 4:44 AM Changeset in webkit [155849] by
-
- 5 edits in trunk/Source/WebCore
Devirtualize Document::detach().
<https://webkit.org/b/121418>
Reviewed by Antti Koivisto.
The only override was in PluginDocument which needs to decouple itself
from the plugin element at the start of detach.
Added PluginDocument::detachFromPluginElement() and call it from detach().
- 3:49 AM Changeset in webkit [155848] by
-
- 3 edits in trunk/LayoutTests
Unreviewed gardening. Removed fail expectations for test in
the GTK & EFL ports, consistently passing after r155374.
- platform/efl/TestExpectations: Removed test.
- platform/gtk/TestExpectations: Ditto.
- 2:27 AM Changeset in webkit [155847] by
-
- 8 edits in trunk
Source/WebKit2: [WKTR] WebKitTestRunner's eventSender.contextClick() returns objects without implemented click() method.
https://bugs.webkit.org/show_bug.cgi?id=98410
Patch by Wojciech Bielawski <w.bielawski@samsung.com> on 2013-09-16
Reviewed by Darin Adler.
Add API required to implement click() method for context menu entry, used by WebkitTestRunner::EventSender.
- WebProcess/InjectedBundle/API/c/WKBundlePage.cpp:
(WKBundlePageClickMenuItem):
(WKBundlePageCopyContextMenuItemTitle):
(WKBundlePageGetContextMenuItems):
- WebProcess/InjectedBundle/API/c/WKBundlePagePrivate.h:
Tools: [WKTR] WebKitTestRunner's eventSender.contextClick() returns objects without implemented click() method.
https://bugs.webkit.org/show_bug.cgi?id=98410
Patch by Wojciech Bielawski <w.bielawski@samsung.com> on 2013-09-16
Reviewed by Darin Adler.
Add implementation of click() method of objects returned by eventSender.contextClick().
- WebKitTestRunner/InjectedBundle/EventSendingController.cpp:
(MenuItemPrivateData):
(WTR::menuItemClickCallback):
(WTR::getMenuItemTitleCallback):
(WTR::staticMenuItemFinalize):
(WTR::getMenuItemClass):
(WTR::EventSendingController::contextClick):
LayoutTests: [WKTR] Add implementation of click() method of objects returned by eventSender.contextClick().
https://bugs.webkit.org/show_bug.cgi?id=98410
Patch by Wojciech Bielawski <w.bielawski@samsung.com> on 2013-09-16
Reviewed by Darin Adler.
Unskipping tests related with click() method.
- platform/efl-wk2/TestExpectations: Corrected bug number for tests: editing/pasteboard/can-read-in-copy-and-cut-events.html editing/pasteboard/can-read-in-dragstart-event.html
- platform/gtk-wk2/TestExpectations:
- 2:05 AM Changeset in webkit [155846] by
-
- 2 edits in trunk/Source/JavaScriptCore
[sh4] Handle subp opcode with 3 operands and bpbeq opcode in LLINT.
https://bugs.webkit.org/show_bug.cgi?id=121412
Patch by Julien Brianceau <jbriance@cisco.com> on 2013-09-16
Reviewed by Andreas Kling.
- offlineasm/sh4.rb:
- 1:42 AM Changeset in webkit [155845] by
-
- 4 edits in trunk/Source/WebCore
CTTE: RenderCombineText always has a Text node.
<https://webkit.org/b/121411>
Reviewed by Antti Koivisto.
This renderer is never anonymous and always has a corresponding Text node.
Replaced node() with a textNode() reference getter.
- 1:00 AM Changeset in webkit [155844] by
-
- 2 edits in releases/WebKitGTK/webkit-2.2/Source/WebCore
Merge r155778 - Missing allow-none introspection annotation for DomDocument.evaluate method
https://bugs.webkit.org/show_bug.cgi?id=118310
Reviewed by Martin Robinson.
Add (allow-none) introspection annotation to parameters that can
be NULL.
- bindings/scripts/CodeGeneratorGObject.pm:
(GenerateFunction):
- 12:59 AM WebKitGTK/2.2.x edited by
- (diff)
- 12:48 AM Changeset in webkit [155843] by
-
- 10 edits in releases/WebKitGTK/webkit-2.2
Merge r155768 - [GTK] WebKitGTK+ is linking against libxslt in too many places
https://bugs.webkit.org/show_bug.cgi?id=121356
Reviewed by Martin Robinson.
Source/WebCore:
Don't use LIBXSLT_CFLAGS when compiling libWebCorePlatform and
libWebCoreGtk, no source files use libxslt there.
- GNUmakefile.am:
Source/WebKit/gtk:
Don't use LIBXSLT_CFLAGS when compiling libwebkitgtk, no source
files use libxslt there.
The webkit tests don't use libxslt, so we shouldn't link against
it.
- GNUmakefile.am:
Source/WebKit2:
Don't use LIBXSLT_CFLAGS when compiling libWebCoreGtk2, no source
files use libxslt there.
- GNUmakefile.am:
Tools:
TestWebCore, WebKitTestRunner and MiniBrowser don't use libxslt,
so we don't need to link against it.
- MiniBrowser/gtk/GNUmakefile.am:
- TestWebKitAPI/GNUmakefile.am:
- WebKitTestRunner/GNUmakefile.am:
- 12:14 AM Changeset in webkit [155842] by
-
- 2 edits in releases/WebKitGTK/webkit-2.2/Tools
Merge r155758 - Uses gtk-doc tools even when --disable-gtk-doc passed
https://bugs.webkit.org/show_bug.cgi?id=91239
Patch by Ross Burton <ross.burton@intel.com> on 2013-09-14
Reviewed by Martin Robinson.
Don't run generate-gtkdoc on install if gtk-doc is disabled.
- GNUmakefile.am:
- 12:13 AM WebKitGTK/2.2.x edited by
- (diff)
- 12:09 AM Changeset in webkit [155841] by
-
- 2 edits in releases/WebKitGTK/webkit-2.2/Tools
Merge r155772 - [GTK] Fails to compile if the build directory is in a different partition
https://bugs.webkit.org/show_bug.cgi?id=121296
Reviewed by Martin Robinson.
Instead of using os.symlink() to install the gtk-doc templates,
use os.link() first and fall back to os.symlink() if it fails.
- gtk/gtkdoc.py:
(GTKDoc._copy_doc_files_to_output_dir.copy_file_replacing_existing):
- 12:08 AM WebKitGTK/2.2.x edited by
- (diff)
- 12:07 AM Changeset in webkit [155840] by
-
- 2 edits in releases/WebKitGTK/webkit-2.2/Tools
Merge r155757 - [GTK] Fails to compile if the build directory is in a different partition
https://bugs.webkit.org/show_bug.cgi?id=121296
Reviewed by Carlos Garcia Campos.
Use symlinks instead of hard links to support building on a
different file system.
- gtk/gtkdoc.py:
(GTKDoc._copy_doc_files_to_output_dir.copy_file_replacing_existing):
- 12:05 AM Changeset in webkit [155839] by
-
- 2 edits in releases/WebKitGTK/webkit-2.2/Source/WebKit/gtk
Merge r155677 - download-requested arg should be WEBKIT_TYPE_DOWNLOAD not G_TYPE_OBJECT
https://bugs.webkit.org/show_bug.cgi?id=57634
Reviewed by Darin Adler.
- webkit/webkitwebview.cpp:
(webkit_web_view_class_init):
- 12:04 AM WebKitGTK/2.2.x edited by
- (diff)
- 12:04 AM Changeset in webkit [155838] by
-
- 2 edits in releases/WebKitGTK/webkit-2.2/Source/WebKit2
Merge r155533 - [WK2] [GTK] LayerTreeHostGtk: don't set m_isValid to false if glContext() returns null
https://bugs.webkit.org/show_bug.cgi?id=120892
Reviewed by Martin Robinson.
If the GL context cannot be created then m_isValid is set to false
during the initialization of LayerTreeHostGtk.
This is not really necessary since the rest of the code already
deals with that situation, so it doesn't have any effect other
than breaking the assertion in invalidate().
- WebProcess/WebPage/gtk/LayerTreeHostGtk.cpp:
(WebKit::LayerTreeHostGtk::initialize):
- 12:03 AM WebKitGTK/2.2.x edited by
- (diff)
Sep 15, 2013:
- 11:59 PM Changeset in webkit [155837] by
-
- 5 edits in trunk/Source/WebCore
CTTE: RenderSVGInlineText always has a Text node.
<https://webkit.org/b/121409>
Reviewed by Antti Koivisto.
This renderer is never anonymous and always has a corresponding Text node.
Replaced node() with a textNode() reference getter.
Only one null check was harmed in the making of this patch.
- 11:59 PM Changeset in webkit [155836] by
-
- 2 edits in releases/WebKitGTK/webkit-2.2
Merge r155517 - autogen.sh: fix removal of autom4te.cache
https://bugs.webkit.org/show_bug.cgi?id=121150
Reviewed by Carlos Garcia Campos.
The removal of autom4te.cache is wrong, it uses 'rm -f' instead of
'rm -rf' and it relies on an undefined variable.
In addition to that, it should be done after running autoreconf,
which is when it's no longer needed.
- autogen.sh:
- 11:58 PM WebKitGTK/2.2.x edited by
- (diff)
- 11:58 PM Changeset in webkit [155835] by
-
- 2 edits in releases/WebKitGTK/webkit-2.2
Merge r155515 - Unquoted $ORIGDIR in autogen.sh
https://bugs.webkit.org/show_bug.cgi?id=19512
Reviewed by Carlos Garcia Campos.
Quote all directory names. This doesn't mean that all possible
directory names are safe for building webkit, but the configure
script already runs a sanity check.
- autogen.sh:
- 11:57 PM WebKitGTK/2.2.x edited by
- (diff)
- 10:53 PM Changeset in webkit [155834] by
-
- 2 edits in trunk/Source/WebCore
Try to fix the Windows build.
- dom/make_names.pl: The static_cast here was running into const-correctness issues, so for now change it back to a C-style cast.
- 10:40 PM Changeset in webkit [155833] by
-
- 29 edits in trunk/Source/WebCore
Adding "explicit" keyword in rendering related constructors
https://bugs.webkit.org/show_bug.cgi?id=121398
Patch by Santosh Mahto <santosh.ma@samsung.com> on 2013-09-15
Reviewed by Andreas Kling.
Adding explicit in below constructor.
All files belong to Source/WebCore/rendering/ path.
- rendering/AutoTableLayout.h:
- rendering/ClipPathOperation.h:
(WebCore::ClipPathOperation::ClipPathOperation):
(WebCore::ShapeClipPathOperation::ShapeClipPathOperation):
- rendering/FixedTableLayout.h:
- rendering/FloatingObjects.h:
(WebCore::FloatingObject::FloatingObject):
- rendering/FlowThreadController.h:
- rendering/HitTestResult.h:
- rendering/ImageQualityController.h:
- rendering/InlineBox.h:
(WebCore::InlineBox::InlineBox):
(WebCore::InlineBox::InlineBoxBitfields::InlineBoxBitfields):
- rendering/InlineFlowBox.h:
(WebCore::InlineFlowBox::InlineFlowBox):
- rendering/InlineTextBox.h:
(WebCore::InlineTextBox::InlineTextBox):
- rendering/LayoutState.h:
- rendering/LogicalSelectionOffsetCaches.h:
(WebCore::LogicalSelectionOffsetCaches::LogicalSelectionOffsetCaches):
- rendering/RenderBoxModelObject.h:
- rendering/RenderFlexibleBox.h:
- rendering/RenderFrameSet.h:
(WebCore::FrameEdgeInfo::FrameEdgeInfo):
- rendering/RenderGeometryMap.h:
- rendering/RenderImage.h:
- rendering/RenderImageResourceStyleImage.h:
- rendering/RenderLayerFilterInfo.h:
- rendering/RenderMultiColumnSet.h:
- rendering/RenderNamedFlowThread.h:
- rendering/RenderReplaced.h:
- rendering/RenderSearchField.h:
- rendering/RenderSelectionInfo.h:
(WebCore::RenderSelectionInfoBase::RenderSelectionInfoBase):
(WebCore::RenderBlockSelectionInfo::RenderBlockSelectionInfo):
- rendering/RenderTableSection.h:
- rendering/RenderTextControl.h:
- rendering/RenderTextControlSingleLine.h:
- rendering/TrailingFloatsRootInlineBox.h:
- 6:29 PM Changeset in webkit [155832] by
-
- 10 edits in trunk
Unreviewed make distcheck fix.
Source/JavaScriptCore:
- GNUmakefile.list.am:
Source/WebCore:
- GNUmakefile.am:
- GNUmakefile.list.am:
Source/WebKit2:
- GNUmakefile.list.am:
Source/WTF:
- GNUmakefile.list.am:
- 6:19 PM Changeset in webkit [155831] by
-
- 2 edits in trunk/Source/WebCore/platform/gtk/po
[l10n] Updated Polish translation of WebKitGTK+
https://bugs.webkit.org/show_bug.cgi?id=121392
Patch by Piotr Drąg <piotrdrag@gmail.com> on 2013-09-15
Reviewed by Gustavo Noronha Silva.
- pl.po: updated.
- 6:01 PM Changeset in webkit [155830] by
-
- 5 edits in trunk/Source/WebKit/win
Remove dependency on CoreFoundation from WebIconDatabase
https://bugs.webkit.org/show_bug.cgi?id=119387
Reviewed by Brent Fulgham.
- COMVariantSetter.h:
(COMVariant::operator&):
(COMVariant::operator->):
- WebIconDatabase.cpp:
(WebIconDatabase::iconDatabaseNotificationUserInfoURLKey):
(postDidAddIconNotification):
- WebIconDatabase.h:
- WebView.cpp:
(WebView::notifyDidAddIcon):
- 5:45 PM Changeset in webkit [155829] by
-
- 2 edits in trunk/Source/WTF
Fix build with Visual Studio 2012 after r153764.
https://bugs.webkit.org/show_bug.cgi?id=121406
Reviewed by Brent Fulgham.
- wtf/MathExtras.h:
- 5:02 PM Changeset in webkit [155828] by
-
- 2 edits in trunk/Source/WebCore
[WIN] Fix build without precompiled header after r154146.
- platform/win/WebCoreBundleWin.h:
- 2:07 PM Changeset in webkit [155827] by
-
- 4 edits in trunk/Source/WebCore
CTTE: EllipsisBox owner renderer is always a RenderBlock.
<https://webkit.org/b/121402>
Reviewed by Antti Koivisto.
Let the constructor reflect this. Also out-of-line it because reasons.
- 1:51 PM Changeset in webkit [155826] by
-
- 2 edits in trunk/Source/WTF
Unreviewed GTK gardening.
- GNUmakefile.list.am: remove FastAllocBase.h, this file disappeared
in r155251.
- 12:56 PM Changeset in webkit [155825] by
-
- 44 edits in trunk/Source/WebCore
CTTE: RenderObject's createAnonymous() and setDocumentForAnonymous() should take Document references
https://bugs.webkit.org/show_bug.cgi?id=121400
Reviewed by Andreas Kling.
The Document used for anonymous renderers is never null.
- dom/Document.cpp:
- page/FrameView.cpp:
- rendering/FlowThreadController.cpp:
- rendering/RenderBlock.cpp:
- rendering/RenderBlock.h:
- rendering/RenderFlexibleBox.cpp:
- rendering/RenderFlexibleBox.h:
- rendering/RenderFullScreen.cpp:
- rendering/RenderFullScreen.h:
- rendering/RenderImage.cpp:
- rendering/RenderInline.cpp:
- rendering/RenderInline.h:
- rendering/RenderLayer.cpp:
- rendering/RenderListMarker.cpp:
- rendering/RenderMultiColumnBlock.cpp:
- rendering/RenderMultiColumnFlowThread.cpp:
- rendering/RenderMultiColumnFlowThread.h:
- rendering/RenderMultiColumnSet.cpp:
- rendering/RenderMultiColumnSet.h:
- rendering/RenderNamedFlowThread.cpp:
- rendering/RenderNamedFlowThread.h:
- rendering/RenderObject.h:
- rendering/RenderReplica.cpp:
- rendering/RenderReplica.h:
- rendering/RenderRuby.cpp:
- rendering/RenderRubyBase.cpp:
- rendering/RenderRubyBase.h:
- rendering/RenderRubyRun.cpp:
- rendering/RenderScrollbar.cpp:
- rendering/RenderScrollbarPart.cpp:
- rendering/RenderScrollbarPart.h:
- rendering/RenderTable.cpp:
- rendering/RenderTableCell.cpp:
- rendering/RenderTableCell.h:
- rendering/RenderTableRow.cpp:
- rendering/RenderTableRow.h:
- rendering/RenderTableSection.cpp:
- rendering/RenderText.cpp:
- rendering/RenderView.cpp:
- rendering/RenderView.h:
- rendering/mathml/RenderMathMLBlock.cpp:
- rendering/mathml/RenderMathMLRow.cpp:
- rendering/mathml/RenderMathMLScripts.cpp:
- 12:48 PM Changeset in webkit [155824] by
-
- 2 edits in trunk/Tools
Make the GDB webkit.py script compatible with Python v3
https://bugs.webkit.org/show_bug.cgi?id=121395
Reviewed by Ryosuke Niwa.
GDB fails to load the printing helpers from the webkit.py script if Python v3 is used. To avoid that,
uses of xrange() in the script are replaced with range(), and the print statements are changed into
the print() function invocations.
- gdb/webkit.py:
(guess_string_length):
(ustring_to_string):
(lstring_to_string):
(WTFCStringPrinter.to_string):
(PrintPathToRootCommand.invoke):
- 12:19 PM Changeset in webkit [155823] by
-
- 6 edits in trunk/Source/WebCore
CTTE: ContentData::createRenderer() should take references
https://bugs.webkit.org/show_bug.cgi?id=121399
Reviewed by Andreas Kling.
- Convert createRenderer(Document*, RenderStyle*) to createRenderer(Document&, RenderStyle&)
- Do a drive by final-ing of ContentData derived classes.
- dom/PseudoElement.cpp:
(WebCore::PseudoElement::didAttachRenderers):
- rendering/RenderImage.cpp:
(WebCore::RenderImage::createAnonymous):
- rendering/RenderImage.h:
- rendering/style/ContentData.cpp:
(WebCore::ImageContentData::createRenderer):
(WebCore::TextContentData::createRenderer):
(WebCore::CounterContentData::createRenderer):
(WebCore::QuoteContentData::createRenderer):
- rendering/style/ContentData.h:
- 12:10 PM Changeset in webkit [155822] by
-
- 2 edits in trunk/LayoutTests
Attempt to fix a local flakiness on Lion.
Rubber stamped by Andreas Kling.
- fast/history/history-back-while-pdf-in-pagecache.html:
- 12:00 PM Changeset in webkit [155821] by
-
- 4 edits in trunk/Source/WebCore
CTTE: RenderFrameSet is never anonymous.
<https://webkit.org/b/121396>
Reviewed by Antti Koivisto.
This renderer is never anonymous, and always has a corresponding HTMLFrameSetElement.
Replaced element() with a frameSetElement().
- 11:53 AM Changeset in webkit [155820] by
-
- 4 edits in trunk/Source/JavaScriptCore
Deoptimize deoptimization: make DFGOSRExitCompiler64.cpp more hackable
https://bugs.webkit.org/show_bug.cgi?id=121374
Reviewed by Geoffrey Garen.
This reduces the size of DFGOSRExitCompiler64.cpp by almost 50%, and makes it
super easy to add new recovery kinds. For recoveries that involve reboxing, it
allows you to keep most of the code common between the on-stack and in-reg
cases: they all get funneled through the "load from scratch buffer, convert,
and then store to stack" logic.
This opens up a bunch of possibilities. It'll make adding Int48 much easier,
and it probably will come in handy as we do various DFG stack layout changes in
support of the FTL.
- bytecode/ValueRecovery.h:
(JSC::ValueRecovery::dumpInContext):
(JSC::ValueRecovery::dump):
- dfg/DFGOSRExitCompiler.cpp:
(JSC::DFG::shortOperandsDump):
- dfg/DFGOSRExitCompiler64.cpp:
(JSC::DFG::OSRExitCompiler::compileExit):
- 11:47 AM Changeset in webkit [155819] by
-
- 9 edits in trunk/Source/WebCore
CTTE: Node subclasses should take a Document by reference in their constructor (Part 10)
https://bugs.webkit.org/show_bug.cgi?id=121397
Reviewed by Andreas Kling.
Convert the bindings code to pass a Document reference for named constructors.
- bindings/js/JSImageConstructor.cpp:
(WebCore::constructImage):
- bindings/scripts/CodeGeneratorJS.pm:
(GenerateConstructorDefinition):
- html/HTMLAudioElement.cpp:
(WebCore::HTMLAudioElement::createForJSConstructor):
- html/HTMLAudioElement.h:
- html/HTMLImageElement.cpp:
(WebCore::HTMLImageElement::createForJSConstructor):
- html/HTMLImageElement.h:
- html/HTMLOptionElement.cpp:
(WebCore::HTMLOptionElement::createForJSConstructor):
- html/HTMLOptionElement.h:
- 11:25 AM Changeset in webkit [155818] by
-
- 5 edits in trunk/Source/WebCore
CTTE: Node subclasses should take a Document by reference in their constructor (Part 9)
https://bugs.webkit.org/show_bug.cgi?id=121394
Reviewed by Andreas Kling.
Remove the 'constructorTakesDocumentReference' property. It is the default now.
- dom/make_names.pl:
Always use a Document reference.
- html/HTMLTagNames.in:
- mathml/mathtags.in:
- svg/svgtags.in:
Remove the 'constructorTakesDocumentReference' property.
- 11:12 AM Changeset in webkit [155817] by
-
- 4 edits in trunk/Source/WebCore
RenderWidget doesn't need to cache a FrameView pointer.
<https://webkit.org/b/121158>
Reviewed by Antti Koivisto.
The FrameView can be found through RenderObject::view().frameView().
This way also gets you a reference, so no need for null-checking.
- 11:08 AM Changeset in webkit [155816] by
-
- 2 edits in releases/WebKitGTK/webkit-2.2/Source/WebKit2
Merge r155814 - [GTK] The web inspector is not rendered when docked in WebKit2 when AC is used
https://bugs.webkit.org/show_bug.cgi?id=121381
Reviewed by Martin Robinson.
When accelerated compositing results are rendered we return early
from webkitWebViewBaseDraw and other WebView children are not
drawn. We need to chain up the draw signal handler, so that the
GtkContainer propagates the draw to all its children.
- UIProcess/API/gtk/WebKitWebViewBase.cpp:
(webkitWebViewBaseDraw):
- 11:06 AM Changeset in webkit [155815] by
-
- 185 edits in trunk/Source/WebCore
CTTE: Node subclasses should take a Document by reference in their constructor (Part 8)
https://bugs.webkit.org/show_bug.cgi?id=121393
Reviewed by Andreas Kling.
Converts the following to take a Document reference:
- All of SVG
- 10:41 AM Changeset in webkit [155814] by
-
- 2 edits in trunk/Source/WebKit2
[GTK] The web inspector is not rendered when docked in WebKit2 when AC is used
https://bugs.webkit.org/show_bug.cgi?id=121381
Reviewed by Martin Robinson.
When accelerated compositing results are rendered we return early
from webkitWebViewBaseDraw and other WebView children are not
drawn. We need to chain up the draw signal handler, so that the
GtkContainer propagates the draw to all its children.
- UIProcess/API/gtk/WebKitWebViewBase.cpp:
(webkitWebViewBaseDraw):
- 10:10 AM Changeset in webkit [155813] by
-
- 16 edits in trunk/Source/WebCore
[CTTE] Node subclasses should take a Document by reference in their constructor (Part 7)
https://bugs.webkit.org/show_bug.cgi?id=121389
Reviewed by Andreas Kling.
Converts the following to take a Document reference:
- Element
- MathMLElement
- MathMLInlineContainerElement
- MathMLMathElement
- MathMLTextElement
- 7:58 AM Changeset in webkit [155812] by
-
- 2 edits in trunk/Source/WTF
[WIN] Fix build without precompiled header after r155454.
- wtf/win/GDIObject.h: Added missing windows.h include.
- 7:13 AM Changeset in webkit [155811] by
-
- 9 edits in trunk/Source/WebCore
CTTE: RenderMedia and RenderVideo are never anonymous.
<https://webkit.org/b/121388>
Reviewed by Sam Weinig.
Codify the following:
- RenderMedia always has an HTMLMediaElement.
- RenderVideo always has an HTMLVideoElement.
None of these renderers are ever anonymous, so delete element() and provide
strongly typed reference getters instead.
- 7:12 AM Changeset in webkit [155810] by
-
- 9 edits5 adds in trunk/Source/WebCore
Unreviewed, rolling out r155809.
http://trac.webkit.org/changeset/155809
https://bugs.webkit.org/show_bug.cgi?id=121388
Accidentally removed files, let's redo this.
- 7:06 AM Changeset in webkit [155809] by
-
- 9 edits5 deletes in trunk/Source/WebCore
CTTE: RenderMedia and RenderVideo are never anonymous.
<https://webkit.org/b/121388>
Reviewed by Sam Weinig.
Codify the following:
- RenderMedia always has an HTMLMediaElement.
- RenderVideo always has an HTMLVideoElement.
None of these renderers are ever anonymous, so delete element() and provide
strongly typed reference getters instead.
- 6:31 AM Changeset in webkit [155808] by
-
- 47 edits in trunk/Source/WebCore
[CTTE] Node subclasses should take a Document by reference in their constructor (Part 6)
https://bugs.webkit.org/show_bug.cgi?id=121387
Reviewed by Andreas Kling.
Converts the following to take a Document reference:
- Attr
- CDATASection
- CharacterData
- Comment
- DOMImplementation
- DocumentFragment
- DocumentType
- EntityReference
- Notation
- ProcessingInstruction
- ShadowRoot
- TemplateContentDocumentFragment
- Text
- 5:59 AM Changeset in webkit [155807] by
-
- 2 edits in trunk/Source/WebCore
Remove RenderObject::clearNode().
<https://webkit.org/b/121386>
Reviewed by Anders Carlsson.
This was used by ref-counted RenderWidgets after getting removed from their
parent renderer but kept alive by an external ref. We have no need for this
awkward state anymore, so remove clearNode().
- 5:52 AM Changeset in webkit [155806] by
-
- 21 edits in trunk/Source/WebCore
CTTE: RenderWidgets are never anonymous.
<https://webkit.org/b/121385>
Reviewed by Anders Carlsson.
Codify the following:
- RenderWidget always has an HTMLFrameOwnerElement.
- RenderEmbeddedObject always has an HTMLFrameOwnerElement.
- RenderSnapshottedPlugIn always has an HTMLPlugInImageElement.
- RenderFrame always has an HTMLFrameElement.
- RenderIFrame always has an HTMLIFrameElement.
None of these renderers are ever anonymous, so delete element() and provide
strongly typed reference getters instead.
- 5:51 AM Changeset in webkit [155805] by
-
- 2 edits in trunk/Source/WTF
Remove Ref(const T&) constructor.
<https://webkit.org/b/121384>
Reviewed by Anders Carlsson.
This constructor was added as a stopgap measure until we got better support
for move semantics in Vector.
- wtf/Ref.h:
- 5:07 AM WebKitGTK/2.2.x edited by
- (diff)
- 4:42 AM WebKitGTK/2.0.x edited by
- (diff)
- 3:35 AM Changeset in webkit [155804] by
-
- 2 edits in trunk/Source/WTF
Unreviewed, fix ridiculous bug in WeakPtr.
I made operator!() null check the pointer to the WeakReference. :|
Caught by plugins/destroy-on-setwindow.html on bots.
- wtf/WeakPtr.h:
(WTF::WeakPtr::operator!):
- 2:29 AM Changeset in webkit [155803] by
-
- 4 edits in trunk/Tools
[webkitpy] perf profiler attaches to WKTR after the WebProcess was launched
https://bugs.webkit.org/show_bug.cgi?id=119631
Reviewed by Ryosuke Niwa.
The perf profiler is attached to the test runner after the test runner has been launched,
but in the case of WebKitTestRunner, this is done after the WebProcess was spawned by the
UIProcess, meaning that WebProcess operations will not be profiled.
To avoid that, the perf profiler is changed to provide a wrapper command for the test runner.
This is done in the scope of already present support for wrapper scripts (such as valgrind)
that are settable through the --wrapper option. This causes the perf profiler to correctly
continue profiling the WebProcess after it was spawned.
When the profiler is launched, a wait process is launched with the profiler's pid as the
argument. This ensures we can check that the profiler exited normally. If it did not, its
output can be corrupted and can cause problems when that output is examined. The after-exit
report is avoided in that case.
- Scripts/webkitpy/common/system/profiler.py:
(Profiler.wrapper_arguments): The method returns an empty list by default.
(Perf.init): Remove unnecessary member variables.
(Perf.attach_to_pid): Launch the wait process with the profiler's pid as the argument.
(Perf.wrapper_arguments): Used instead of attaching the profiler to the profiled process' pid,
the method returns a list of arguments that should be used as a wrapper for the test runner.
The arguments specify the 'perf record' command, with the output file and the instruction to
construct the call graph given as well.
(Perf._first_ten_lines_of_profile): Prettify the output by avoiding the unnecessary trailing
whitespace in the profile lines.
(Perf.profile_after_exit): Kill the profiler process through its pid if the wait process is still
running. If it is running even after that, or if the profiler process produced a non-zero exit
code, the method reports the profiling failure and returns. Otherwise it continues on reporting
the top 10 lines of the profile and how to examine the profile further.
- Scripts/webkitpy/port/driver.py:
(Driver._command_wrapper): The method is now a member of the Driver class, returning the wrapper
as specified by the --wrapper option if given, or the profiler wrapper if the profiler is present,
or an empty list otherwise.
(Driver.cmd_line):
- Scripts/webkitpy/port/driver_unittest.py:
(DriverTest.make_port): Adjusted so mock host or options objects can be specified
for the testing-specific Port object.
(DriverTest._assert_wrapper): Pass a MockOptions object with the wrapper value to the make_port method,
as the Driver._command_wrapper now operates on the option's value directly rather than accepting it through
a parameter.
(DriverTest.test_profiler_as_wrapper): Provide a test case to check the correct profiler wrapper is used
when profiling on the Linux platform with the perf profiler.
- 1:51 AM Changeset in webkit [155802] by
-
- 16 edits in trunk/Source/WebCore
Use FINAL instead of virtualChildren trick in render tree classes
https://bugs.webkit.org/show_bug.cgi?id=121373
Reviewed by Andreas Kling.
- rendering/RenderBlock.cpp:
(WebCore::RenderBlock::updateFirstLetterStyle): Use children instead of
virtualChildren.
- rendering/RenderBox.cpp:
(WebCore::RenderBox::splitAnonymousBoxesAroundChild): Ditto.
- rendering/RenderBoxModelObject.cpp:
(WebCore::RenderBoxModelObject::moveChildTo): Ditto.
- rendering/RenderObject.cpp:
(WebCore::RenderObject::addChild): Ditto.
(WebCore::RenderObject::removeChild): Ditto.
(WebCore::RenderObject::handleDynamicFloatPositionChange): Ditto.
(WebCore::RenderObject::willBeDestroyed): Ditto.
(WebCore::RenderObject::removeFromRenderFlowThreadRecursive): Ditto.
- rendering/RenderBlock.h: Changed firstChild and lastChild to use
m_children directly instead of a function. Made the children function
a final virtual override. Deleted the virtualChildren override. Added
override keywords to other virtual functions.
- rendering/RenderEmbeddedObject.h: Ditto.
- rendering/RenderFrameSet.h: Ditto.
- rendering/RenderInline.h: Ditto.
- rendering/RenderMedia.h: Ditto.
- rendering/RenderTableCol.h: Ditto.
- rendering/RenderTableRow.h: Ditto.
- rendering/RenderTableSection.h: Ditto.
- rendering/svg/RenderSVGContainer.h: Ditto.
- rendering/svg/RenderSVGRoot.h: Ditto.
- rendering/RenderObject.h:
(WebCore::RenderObject::firstChild): Use children instead of virtualChildren.
(WebCore::RenderObject::lastChild): Ditto.
(WebCore::RenderObject::children): Renamed from virtualChildren.
(WebCore::RenderObject::canHaveChildren): Use children instead of virtualChildren.
- 1:50 AM Changeset in webkit [155801] by
-
- 15 edits in trunk/Source/WebCore
Shrink factory functions
https://bugs.webkit.org/show_bug.cgi?id=121378
Reviewed by Andreas Kling.
This makes the functions that builds four different types of per-element tables
all have smaller code size.
- DerivedSources.make: Tweaked how HTML and SVG wrapper factories are made a bit.
It was OK, but a little sloppy.
- dom/Document.cpp:
(WebCore::Document::createElement): Changed to call the createElement functions
in the factories. Also did some style improvement.
- dom/make_names.pl:
(printConstructorSignature): Make constructors take a Document& instead of
a Document*.
(printConstructorInterior): Updated for Document& instead of Document*.
(printFunctionTable): Renamed from printFunctionInits, because this now
creates table entries. The actual code to add each function is now done
in a loop instead.
(printNamesHeaderFile): Made the table const.
(printNamesCppFile): Made the table const.
(printDefinitions): Generate a table, table entries, and a loop that calls
createQualifiedName, rather than an "unrolled loop".
(printFactoryCppFile): Tweaked indentation. Updated to take Document& instead
of Document*. Generate a table, table entries, and a loop that builds a map,
rather than an "unrolled" loop.
(printFactoryHeaderFile): Eliminated the unused createElement member function
that was part of a dream of factory objects in a namespace registry, giving
the name createElement to the static member function so it would be less
redundant. Also made the function takes a Document& instead of Document*.
Also made the default "not created by parser", since the parser is the simpler
use case, outnumbered by the non-parser uses, at least at the moment. Also,
use nullptr instead of 0.
(printWrapperFactoryCppFile): Generate a table, table entries, and a loop
that builds a map rather than an "unrolled" loop.
- editing/htmlediting.cpp:
(WebCore::createHTMLElement): Put the null check here that used to be in
HTMLElementFactory::createElement, and call with a reference instead of a
pointer. Also don't need to pass "0, false" since "not created by the parser"
is now the default.
- html/HTMLDocument.cpp:
(WebCore::HTMLDocument::createElement): Ditto.
- html/HTMLElement.cpp:
(WebCore::populateEventNameForAttributeLocalNameMap): Changed to use an
AtomicStringImpl* as the key instead of AtomicString, since all the strings
come from the local names of tags, and so don't need to be ref'd. This saves
a bit of reference count churn when building the map and is the same pattern
used in some maps in the make_names.pl script above.
(WebCore::HTMLElement::parseAttribute): Updated for change above.
- html/HTMLObjectElement.cpp:
(WebCore::isRecognizedTagName): More const.
- html/parser/HTMLConstructionSite.cpp:
(WebCore::HTMLConstructionSite::createHTMLElement): Changed to call the
HTMLElementFactory::createElement function under its new name with a reference
rather than a pointer.
- html/parser/HTMLIdentifier.cpp:
(WebCore::nameForIndex): More const.
(WebCore::HTMLIdentifier::addNames): More const.
- html/parser/HTMLTreeBuilder.cpp:
(WebCore::mapLoweredLocalNameToName): More const.
(WebCore::adjustSVGTagNameCase): Eliminated local variable so we would not have
to utter type with more const.
(WebCore::adjustAttributes): Ditto.
(WebCore::addNamesWithPrefix): More const. Also tweaked to use a bit more reference,
and a bit less pointer. Also used array instead of pointer for argument since they
are synonyms and this use is more array-like and const array is slightly easier to
understand than the double const pointer.
(WebCore::adjustForeignAttributes): Eliminated local variable so we would not have
to utter type with more const.
- html/track/TextTrackCue.cpp:
(WebCore::TextTrackCue::copyWebVTTNodeToDOMTree): Pass a reference to
WebVTTElement::createEquivalentHTMLElement.
- html/track/WebVTTElement.cpp:
(WebCore::WebVTTElement::WebVTTElement): Take a reference instead of a pointer.
(WebCore::WebVTTElement::create): Ditto.
(WebCore::WebVTTElement::cloneElementWithoutAttributesAndChildren): Pass a
reference instead of a pointer to create.
(WebCore::WebVTTElement::createEquivalentHTMLElement): Take a reference instead
of a pointer, and update uses of HTMLElementFactory::createElement. Also made
this return null when a bad node type is passed in, which should never happen.
- html/track/WebVTTElement.h: Use Document& instead of Document*. Removed unused
and undefined create/constructor overloads.
- html/track/WebVTTParser.cpp:
(WebCore::WebVTTParser::constructTreeFromToken): Pass document reference rather than
pointer to WebVTTElement::create.
- 1:21 AM Changeset in webkit [155800] by
-
- 4 edits in trunk/Source/WebCore
Get rid of ref-counting on RenderArena.
<https://webkit.org/b/121379>
Reviewed by Darin Adler.
This was used to guard the arena during ref-counted RenderWidget teardown.
- 1:18 AM Changeset in webkit [155799] by
-
- 2 edits in trunk/Source/WebCore
Fix Qt build.
- css/CSSSelector.cpp: (WebCore::populatePseudoTypeByNameMap): Removed stray line that was breaking the build for any platform with !ENABLE(VIDEO_TRACK).
- 1:15 AM Changeset in webkit [155798] by
-
- 3 edits in trunk/Source/WebCore
Restore two-pass mechanism for FrameView::updateEmbeddedObjects().
<https://webkit.org/b/121380>
Reviewed by Darin Adler.
Restore the code to make a second pass over the embedded objects needing
an update in case more objects were added to the set during the first pass.
- 1:08 AM Changeset in webkit [155797] by
-
- 15 edits20 adds2 deletes in trunk
Implement the mmultiscripts tag
https://bugs.webkit.org/show_bug.cgi?id=99618
Patch by Frédéric Wang <fred.wang@free.fr> on 2013-09-15
Reviewed by Chris Fleizach.
Source/WebCore:
Tests: mathml/invalid-scripts-crash.html
mathml/presentation/multiscripts-equivalence.html
mathml/presentation/multiscripts-noscripts.html
mathml/presentation/multiscripts-positions.html
mathml/presentation/scripts-base-alignment.html
mathml/presentation/scripts-horizontal-alignment.html
mathml/presentation/scripts-vertical-alignment.html
mathml/scripts-addChild.html
mathml/scripts-removeChild.html
This relies on the existing msub/msup/msubsup code to implement the mmultiscripts tag. This also improves dynamic addition/removal of children for these MathML elements and adds a specific style for invalid children, so that they render like an merror tag. Finally, this fixes a bad memory access in the Accessibility render object of msubsup.
- CMakeLists.txt:
- GNUmakefile.list.am:
- Target.pri:
- WebCore.vcxproj/WebCore.vcxproj:
- WebCore.vcxproj/WebCore.vcxproj.filters:
- WebCore.xcodeproj/project.pbxproj:
- accessibility/AccessibilityRenderObject.cpp:
(WebCore::AccessibilityRenderObject::isMathSubscriptSuperscript):
(WebCore::AccessibilityRenderObject::mathSuperscriptObject):
- css/mathml.css:
(mo, mrow, mfenced, mfrac, msub, msup, msubsup, mmultiscripts, mprescripts, none, munder, mover, munderover, msqrt, mroot):
(mover > :last-child, munderover > :last-child):
(msub > * + *, msup > * + *, msubsup > * + *, mmultiscripts > * + *, munder > * + *, mover > * + *, munderover > * + *):
(merror, msub > * + * + *, msup > * + * + *, msubsup > * + * + * + *, msub > mprescripts, msup > mprescripts, msubsup > mprescripts, msub > none, msup > none, msubsup > none, mmultiscripts > mprescripts ~ mprescripts, mmultiscripts > mprescripts ~ mprescripts ~ *):
- mathml/MathMLInlineContainerElement.cpp:
(WebCore::MathMLInlineContainerElement::createRenderer):
- mathml/mathtags.in:
- rendering/mathml/RenderMathMLBlock.h:
(WebCore::RenderMathMLBlock::isRenderMathMLScripts):
(WebCore::RenderMathMLBlock::isRenderMathMLScriptsWrapper):
- rendering/mathml/RenderMathMLScripts.cpp: Added.
(WebCore::isMPrescripts):
(WebCore::RenderMathMLScripts::RenderMathMLScripts):
(WebCore::RenderMathMLScripts::base):
(WebCore::RenderMathMLScripts::fixAnonymousStyleForSubSupPair):
(WebCore::RenderMathMLScripts::fixAnonymousStyles):
(WebCore::RenderMathMLScripts::addChildInternal):
(WebCore::RenderMathMLScripts::removeChildInternal):
(WebCore::RenderMathMLScripts::addChild):
(WebCore::RenderMathMLScripts::removeChild):
(WebCore::RenderMathMLScripts::styleDidChange):
(WebCore::RenderMathMLScripts::unembellishedOperator):
(WebCore::RenderMathMLScripts::layout):
(WebCore::RenderMathMLScripts::firstLineBoxBaseline):
(WebCore::RenderMathMLScriptsWrapper::createAnonymousWrapper):
(WebCore::RenderMathMLScriptsWrapper::addChildInternal):
(WebCore::RenderMathMLScriptsWrapper::addChild):
(WebCore::RenderMathMLScriptsWrapper::removeChildInternal):
(WebCore::RenderMathMLScriptsWrapper::removeChild):
- rendering/mathml/RenderMathMLScripts.h: Added.
(WebCore::RenderMathMLScriptsWrapper::RenderMathMLScriptsWrapper):
(WebCore::RenderMathMLScriptsWrapper::renderName):
(WebCore::RenderMathMLScriptsWrapper::isRenderMathMLScriptsWrapper):
(WebCore::toRenderMathMLScriptsWrapper):
(WebCore::RenderMathMLScripts::isRenderMathMLScripts):
(WebCore::RenderMathMLScripts::renderName):
(WebCore::toRenderMathMLScripts):
- rendering/mathml/RenderMathMLSubSup.cpp: Removed.
- rendering/mathml/RenderMathMLSubSup.h: Removed.
LayoutTests:
Add many tests for script msub/msup/msubsup/mmultiscripts:
- invalid markup
- equivalence between mmultiscripts without scripts and mrow
- equivalence between mmultiscripts and msub/msup/msubsup
- position of scripts in mmultiscripts
- baseline alignment of msub/msup/msubsup elements
- horizontal/vertical alignment of scripts
- adding/removing children with the DOM
- TestExpectations: Some remaining MathML pixel tests are broken by this patch.
- mathml/invalid-scripts-crash-expected.txt: Added.
- mathml/invalid-scripts-crash.html: Added.
- mathml/presentation/multiscripts-equivalence-expected.html: Added.
- mathml/presentation/multiscripts-equivalence.html: Added.
- mathml/presentation/multiscripts-noscripts-expected.html: Added.
- mathml/presentation/multiscripts-noscripts.html: Added.
- mathml/presentation/multiscripts-positions-expected.html: Added.
- mathml/presentation/multiscripts-positions.html: Added.
- mathml/presentation/scripts-base-alignment-expected.html: Added.
- mathml/presentation/scripts-base-alignment.html: Added.
- mathml/presentation/scripts-horizontal-alignment-expected.html: Added.
- mathml/presentation/scripts-horizontal-alignment.html: Added.
- mathml/presentation/scripts-vertical-alignment-expected.html: Added.
- mathml/presentation/scripts-vertical-alignment.html: Added.
- mathml/scripts-addChild-expected.html: Added.
- mathml/scripts-addChild.html: Added.
- mathml/scripts-removeChild-expected.html: Added.
- mathml/scripts-removeChild.html: Added.
- platform/mac/accessibility/mathml-multiscript-expected.txt: Reference updated.
- 12:27 AM Changeset in webkit [155796] by
-
- 13 edits1 delete in trunk/Source
Get rid of ref-counting on RenderWidget.
<https://webkit.org/b/121357>
Reviewed by Darin Adler.
Source/WebCore:
Instead of RenderView tracking RenderWidgets, have FrameView track Widgets
that are currently in the render tree.
To protect ourselves during NPAPI tomfoolery, we now let RenderWidget hand
out weak pointers through a createWeakPtr() method so call sites can monitor
the renderer for deletion without having to take shared ownership.
This works out quite nicely since instead of keeping a limping object alive
for a little longer just so we can call methods on it (to accomplish nothing),
we're forced to check right away if it's gone, and take immediate action.
De-virtualized RenderObject::destroy() since it's no longer needed for
RenderWidget to defer destruction.
- page/FrameView.cpp:
(WebCore::FrameView::layout):
(WebCore::FrameView::repaintFixedElementsAfterScrolling):
Call updateWidgetPositions() on FrameView instead of RenderView.
(WebCore::FrameView::updateEmbeddedObject):
Turn null checking of embedded object's element backpointer into an
assertion. This will eventually go away completely once that renderer
can return a HTMLFrameOwnerElement&.
Use WeakPtr to check for renderer destruction following the call out
to updateWidget().
(WebCore::FrameView::updateEmbeddedObjects):
Slap a WidgetHierarchyUpdatesSuspensionScope guard on this function
to defer Widget updates until all the updateEmbeddedObject calls are
done. This avoids RenderWidget::setWidget() having to handle 'this'
disappearing from underneath.
Also use a ListHashSet with a null sentinel to avoid looping forever.
(WebCore::FrameView::performPostLayoutTasks):
Only call updateEmbeddedObjects() once since that function no longer
operates in chunks.
(WebCore::FrameView::notifyWidgetsInAllFrames):
Call notifyWidgets() on FrameView instead of RenderView.
(WebCore::FrameView::didAddWidgetToRenderTree):
(WebCore::FrameView::willRemoveWidgetFromRenderTree):
Added. These are called by RenderWidget when a Widget is being
added or removed from a RenderWidget.
(WebCore::collectWidgets):
Helper to collect raw Widget pointers into a Vector and ref them.
(WebCore::FrameView::updateWidgetPositions):
Moved here from RenderView. This function holds a ref on all the
attached Widgets and calls RenderWidget::updateWidgetPosition() on
their corresponding renderers.
(WebCore::FrameView::notifyWidgets):
Moved here from RenderView. Holds a ref on all the widgets while
calling Widget::notifyWidget() on each one.
- rendering/RenderLayer.cpp:
(WebCore::RenderLayer::scrollTo):
Call updateWidgetPositions() on FrameView instead of RenderView.
- rendering/RenderObject.h:
De-virtualized destroy().
- rendering/RenderView.cpp:
- rendering/RenderView.h:
Moved a bunch of things to FrameView. Made protected section private
since nothing inherits from RenderView.
- rendering/RenderWidget.h:
(WebCore::RenderWidget::createWeakPtr):
Added a WeakPtr factory for clients that want to monitor this object
for destruction.
- rendering/RenderWidget.cpp:
(WebCore::RenderWidget::RenderWidget):
(WebCore::RenderWidget::willBeDestroyed):
(WebCore::RenderWidget::~RenderWidget):
Removed ref counting.
Removed registration with RenderView in ctor/willBeDestroyed.
(WebCore::RenderWidget::setWidgetGeometry):
Monitor the RenderWidget itself through a WeakPtr and check on it
after each step that we're still alive. In that case just bail out.
(WebCore::RenderWidget::setWidget):
Register the incoming/outgoing Widget with the FrameView.
Use a WeakPtr to check on 'this' after updateWidgetGeometry().
(WebCore::RenderWidget::updateWidgetPosition):
Use a WeakPtr to check on 'this' after updateWidgetGeometry().
- GNUmakefile.list.am:
- rendering/RenderWidgetProtector.h:
Removed.
Source/WTF:
- wtf/WeakPtr.h:
(WTF::WeakPtr::operator!):
Add operator! to WeakPtr.
