GitLab

        Silly with statement, from the future!

        * Scripts/webkitpy/common/system/executive.py:


git-svn-id: svn://svn.chromium.org/blink/trunk@83001 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Move attributedStringFromRange down to WebCore
        https://bugs.webkit.org/show_bug.cgi?id=57905



git-svn-id: svn://svn.chromium.org/blink/trunk@83000 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Tony Chang.

        Don't use Exception.message because it's deprecated
        https://bugs.webkit.org/show_bug.cgi?id=57892

        Suppress the warning for now.  When we move to Python 3, we might need
        to something more dramatic.

        * Scripts/webkitpy/common/system/executive.py:


git-svn-id: svn://svn.chromium.org/blink/trunk@82999 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Ojan Vafai.

        Add builders.js to dashboard file list
        https://bugs.webkit.org/show_bug.cgi?id=57899

        Add file added by http://crrev.com/80538 to dashboard file list.

        * TestResultServer/handlers/dashboardhandler.py:

git-svn-id: svn://svn.chromium.org/blink/trunk@82998 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Alexey Proskuryakov.

        Some plug-in content is displayed upside down in Unity plug-in
        https://bugs.webkit.org/show_bug.cgi?id=57895
        <rdar://problem/9212003>

        Create the flipped geometry layer in the web process instead of the plug-in process,
        which matches what we do in WebKit1.

        * WebProcess/Plugins/Netscape/mac/NetscapePluginMac.mm:
        (WebKit::NetscapePlugin::platformPostInitialize):
        * WebProcess/Plugins/Netscape/mac/PluginProxyMac.mm:
        (WebKit::PluginProxy::pluginLayer):


git-svn-id: svn://svn.chromium.org/blink/trunk@82997 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Dimitri Glazkov.

        [chromium] Don't have 'using namespace' declarations in header files
        https://bugs.webkit.org/show_bug.cgi?id=57882

        * src/EventListenerWrapper.cpp:
        * src/EventListenerWrapper.h:
        * src/VideoFrameChromiumImpl.h:
        * src/WebDOMEventListener.cpp:
        * src/WebDOMEventListenerPrivate.cpp:
        * src/WebDOMEventListenerPrivate.h:
        (WebKit::WebDOMEventListenerPrivate::ListenerInfo::ListenerInfo):

git-svn-id: svn://svn.chromium.org/blink/trunk@82996 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Darin Adler.

        Rename variables in border drawing code for readability
        https://bugs.webkit.org/show_bug.cgi?id=57894

        * rendering/RenderObject.cpp:
        (WebCore::RenderObject::drawLineForBoxSide):
        (WebCore::RenderObject::drawBoxSideFromPath):
        (WebCore::RenderObject::drawArcForBoxSide):

git-svn-id: svn://svn.chromium.org/blink/trunk@82995 bbb929c8-8fbe-4397-9dbb-9b2b20218538

https://bugs.webkit.org/show_bug.cgi?id=57886

Reviewed by Darin Adler.

This reduces the need to do (often expensive) platform surface invalidation.
We already do this during layout but style recalculation may trigger
invalidation as well.
    
* dom/Document.cpp:
(WebCore::Document::recalcStyle):



git-svn-id: svn://svn.chromium.org/blink/trunk@82992 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Darin Adler.

        HTML5 Player Has no Fullscreen Mode
        https://bugs.webkit.org/show_bug.cgi?id=57795

        * html/HTMLVideoElement.cpp:
        (WebCore::HTMLVideoElement::supportsFullscreen): Call the new ChromeClient API
            supportsFullScreenForElement() before the old supportsFullScreenForNode() API.

git-svn-id: svn://svn.chromium.org/blink/trunk@82991 bbb929c8-8fbe-4397-9dbb-9b2b20218538

slaves.
https://bugs.webkit.org/show_bug.cgi?id=56564

Skip fast/preloader/scan-body-from-head-script.html on Leopard to get the bots green.

* platform/mac-leopard/Skipped:



git-svn-id: svn://svn.chromium.org/blink/trunk@82990 bbb929c8-8fbe-4397-9dbb-9b2b20218538

git-svn-id: svn://svn.chromium.org/blink/trunk@82989 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Darin Adler.

        Build fix for YarrParser.h
        https://bugs.webkit.org/show_bug.cgi?id=57822

        * yarr/YarrParser.h:
        (JSC::Yarr::Parser::CharacterClassParserDelegate::CharacterClassParserDelegate):

git-svn-id: svn://svn.chromium.org/blink/trunk@82988 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Don't skip react-to-vsprops-changes.py for all production builds,
only those initiated via JavaScriptCore.make.

* JavaScriptCore.vcproj/JavaScriptCore.make:
* JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCoreGenerated.make:



git-svn-id: svn://svn.chromium.org/blink/trunk@82987 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Darin Adler.

        REGRESSION (r82849): 85,000+ JSC-related leaks seen on SnowLeopard Intel Leaks
        https://bugs.webkit.org/show_bug.cgi?id=57857

        Whoops, accidentally removed a deref().

        * bytecode/StructureStubInfo.cpp:
        (JSC::StructureStubInfo::deref):

git-svn-id: svn://svn.chromium.org/blink/trunk@82986 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Pavel Feldman.

        Web Inspector: fix data url's status and timing the network panel.
        https://bugs.webkit.org/show_bug.cgi?id=53066

        Added data url support for status of network resources.

        * English.lproj/localizedStrings.js:
        * inspector/front-end/NetworkPanel.js:
        (WebInspector.NetworkDataGridNode.prototype._refreshStatusCell):
        * inspector/front-end/Resource.js:
        (WebInspector.Resource.prototype.isDataURL):

git-svn-id: svn://svn.chromium.org/blink/trunk@82985 bbb929c8-8fbe-4397-9dbb-9b2b20218538

* JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCoreGenerated.vcproj: Add per-configuration vsprops files.
* JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCoreGeneratedCommon.vsprops: Removed inheritance from common.vsprops.
Set production environment variable before calling make.
* JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCoreGeneratedDebug.vsprops: Added.
* JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCoreGeneratedDebugAll.vsprops: Added.
* JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCoreGeneratedDebugCairoCFLite.vsprops: Added.
* JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCoreGeneratedProduction.vsprops: Added.
* JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCoreGeneratedRelease.vsprops: Added.
* JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCoreGeneratedReleaseCairoCFLite.vsprops: Added.
* JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCoreGeneratedReleasePGO.vsprops: Added.



git-svn-id: svn://svn.chromium.org/blink/trunk@82983 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Darin Adler.

        Assertion failure when navigating quickly between file URLs
        https://bugs.webkit.org/show_bug.cgi?id=57884
        <rdar://problem/9080559>

        Remove bogus assertion. The pending provisional sandbox extension can be
        non-null if the current provisional load fails because we're about to load a new
        page that also has a sandbox extension.

        * WebProcess/WebPage/WebPage.cpp:
        (WebKit::WebPage::SandboxExtensionTracker::didFailProvisionalLoad):


git-svn-id: svn://svn.chromium.org/blink/trunk@82982 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Adam Barth.

        [V8] Keep dirty primitive CSS values alive.
        https://bugs.webkit.org/show_bug.cgi?id=57810

        * bindings/v8/V8GCController.cpp:
        (WebCore::GrouperVisitor::visitDOMWrapper):

2011-04-04  Vitaly Repeshko  <vitalyr@chromium.org>

        Reviewed by Adam Barth.

        [V8] Keep dirty primitive CSS values alive.
        https://bugs.webkit.org/show_bug.cgi?id=57810

        * platform/chromium/test_expectations.txt: Mark
        fast/dom/StyleSheet/gc-inline-style-cssvalues.html as passing.


git-svn-id: svn://svn.chromium.org/blink/trunk@82981 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Adam Roben.

        Grant the web process access to ~/Library/Preferences/com.apple.universalaccess.plist
        https://bugs.webkit.org/show_bug.cgi?id=57879

        The web process needs access to the unviersal access preferences to determine if full keyboard
        access is enabled. I've filed <rdar://problem/9237619> which tracks reading this preference in the UI
        process and sending it over to the web process whenever it changes.

        * WebProcess/com.apple.WebProcess.sb:


git-svn-id: svn://svn.chromium.org/blink/trunk@82980 bbb929c8-8fbe-4397-9dbb-9b2b20218538

https://bugs.webkit.org/show_bug.cgi?id=57880
        
Update Windows expected results (with pixel tests) to greenify the bots.

* platform/win/fast/blockflow/Kusa-Makura-background-canvas-expected.checksum:
* platform/win/fast/blockflow/Kusa-Makura-background-canvas-expected.png:
* platform/win/fast/blockflow/Kusa-Makura-background-canvas-expected.txt:



git-svn-id: svn://svn.chromium.org/blink/trunk@82979 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Adam Barth.

        webkit-patch should print git's stderr when git svn dcommit fail
        http://webkit.org/b/57861

        * Scripts/webkitpy/common/checkout/scm.py:

git-svn-id: svn://svn.chromium.org/blink/trunk@82978 bbb929c8-8fbe-4397-9dbb-9b2b20218538

https://bugs.webkit.org/show_bug.cgi?id=57881

Update the Windows-specific results to get the bots green.

* platform/win/fast/dom/Window/window-property-descriptors-expected.txt:
onformchange and onforminput were removed in r82925.



git-svn-id: svn://svn.chromium.org/blink/trunk@82977 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by David Levin.

        Added MediaTypeFile enum value to WebContextMenuData::MediaType.
        https://bugs.webkit.org/show_bug.cgi?id=57566

        * public/WebContextMenuData.h:

git-svn-id: svn://svn.chromium.org/blink/trunk@82976 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Unreviewed, removing the chromium-win results causes this test to
        start passing on chromium win and linux.

        * platform/chromium-win/fast/images/svg-as-tiled-background-expected.checksum: Removed.
        * platform/chromium-win/fast/images/svg-as-tiled-background-expected.png: Removed.
        * platform/chromium/test_expectations.txt:

git-svn-id: svn://svn.chromium.org/blink/trunk@82975 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Simon Fraser.

        Add parentheses around && within || to avoid clang warning.

        * WebProcess/WebPage/DrawingAreaImpl.cpp:
        (WebKit::DrawingAreaImpl::sendDidUpdateBackingStoreState):

git-svn-id: svn://svn.chromium.org/blink/trunk@82974 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Adam Roben.

        Add two more entries to the sandbox profile
        https://bugs.webkit.org/show_bug.cgi?id=57875

        Allow read-access to /Library/Dictionaries since WebCore::nextBreakablePosition
        ends up calling into ICU which accesses the dictionary.

        Allow Mach access to com.apple.networkd since it's used by CFNetwork.

        * WebProcess/com.apple.WebProcess.sb:


git-svn-id: svn://svn.chromium.org/blink/trunk@82973 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Tony Chang.

        Add --baseline-search-path to NRWT
        https://bugs.webkit.org/show_bug.cgi?id=56233

        Add NRWT option to specify additional directories to look for baselines
        (will be used by hardware GPU bots which will have local per-bot
        expectations for some tests)

        * Scripts/webkitpy/common/system/filesystem_mock.py:
        * Scripts/webkitpy/layout_tests/port/base.py:
        * Scripts/webkitpy/layout_tests/port/base_unittest.py:
        * Scripts/webkitpy/layout_tests/run_webkit_tests.py:

git-svn-id: svn://svn.chromium.org/blink/trunk@82972 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Gavin Barraclough.

        Make caches window show more info about non-jsobject GC values
        https://bugs.webkit.org/show_bug.cgi?id=57874

        Add ClassInfo to the various internal JS types that currently
        don't have any, and make the text for caches window show the
        classname for non-JSObject instances.

        * runtime/Executable.cpp:
        * runtime/Executable.h:
        (JSC::ExecutableBase::createStructure):
        (JSC::NativeExecutable::createStructure):
        (JSC::NativeExecutable::NativeExecutable):
        (JSC::EvalExecutable::createStructure):
        (JSC::ProgramExecutable::createStructure):
        (JSC::FunctionExecutable::createStructure):
        * runtime/Heap.cpp:
        (JSC::TypeCounter::typeName):
        * runtime/JSGlobalData.cpp:
        (JSC::JSGlobalData::JSGlobalData):
        * runtime/JSGlobalData.h:
        * runtime/ScopeChain.cpp:
        * runtime/ScopeChain.h:
        (JSC::ScopeChainNode::createStructure):
        * runtime/StructureChain.cpp:
        * runtime/StructureChain.h:
        (JSC::StructureChain::createStructure):

git-svn-id: svn://svn.chromium.org/blink/trunk@82971 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Oliver Hunt.

        Invalidate callbacks if the WebPageProxy is no longer valid
        https://bugs.webkit.org/show_bug.cgi?id=57873
        <rdar://problem/9059717>

        Audit all function calls that take callbacks and make sure to invalidate the callbacks
        if the web page is no longer valid.

        * UIProcess/WebPageProxy.cpp:
        (WebKit::WebPageProxy::runJavaScriptInMainFrame):
        (WebKit::WebPageProxy::getRenderTreeExternalRepresentation):
        (WebKit::WebPageProxy::getSourceForFrame):
        (WebKit::WebPageProxy::getContentsAsString):
        (WebKit::WebPageProxy::getSelectionOrContentsAsString):
        (WebKit::WebPageProxy::getMainResourceDataOfFrame):
        (WebKit::WebPageProxy::getResourceDataFromFrame):
        (WebKit::WebPageProxy::getWebArchiveOfFrame):
        (WebKit::WebPageProxy::forceRepaint):
        (WebKit::WebPageProxy::computePagesForPrinting):
        (WebKit::WebPageProxy::drawRectToPDF):
        (WebKit::WebPageProxy::drawPagesToPDF):


git-svn-id: svn://svn.chromium.org/blink/trunk@82970 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Adele Peterson.

        Add a fast path for simple color backgrounds
        https://bugs.webkit.org/show_bug.cgi?id=57773

        Add a fast path for drawing simple color backgrounds. It's
        faster to fill a path or rect than it is to set up a clip,
        and then paint the rect.

        No visible behavior change so no new tests.

        * rendering/RenderBoxModelObject.cpp:
        (WebCore::RenderBoxModelObject::paintFillLayerExtended):

git-svn-id: svn://svn.chromium.org/blink/trunk@82969 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Unreviewed EFL build fix for r82962.

        * CMakeListsEfl.txt:

git-svn-id: svn://svn.chromium.org/blink/trunk@82968 bbb929c8-8fbe-4397-9dbb-9b2b20218538

r82734 changed build.webkit.org's "view results" URLs to point straight to the results.html
files, rather than pointing to the directory that contains them. This is more convenient for
people browsing build.webkit.org, but confused Leaks Viewer.

Fixes <http://webkit.org/b/57869> REGRESSION (r82734): Links in Leaks Viewer's Recent Builds
list don't work

Reviewed by Joseph Pecoraro.

* BuildSlaveSupport/build.webkit.org-config/public_html/LeaksViewer/RecentBuildsLoader.js:
(RecentBuildsLoader.prototype.start): Strip off "/results.html" from the results URL.

git-svn-id: svn://svn.chromium.org/blink/trunk@82967 bbb929c8-8fbe-4397-9dbb-9b2b20218538

This includes machines without the necessary graphics hardware, and machines without
WebKitQuartzCoreAdditions (like the Windows 7 Release (WebKit2 Tests) bots).

Fixes <http://webkit.org/b/57870> REGRESSION (r82960): Lots of tests crashing in
DrawingAreaImpl::enterAcceleratedCompositingMode on Windows 7 Release (WebKit2 Tests)

Reviewed by Anders Carlsson.

* WebProcess/WebPage/LayerTreeHost.h:
(WebKit::LayerTreeHost::supportsAcceleratedCompositing): Added. On platforms other than
Windows, this always returns true.

* WebProcess/WebPage/WebPage.cpp:
(WebKit::WebPage::updatePreferences): Only enable accelerated compositing-related
preferences if the machine supports accelerated compositing.

* WebProcess/WebPage/ca/win/LayerTreeHostCAWin.cpp:
(WebKit::LayerTreeHostCAWin::supportsAcceleratedCompositing): Added. Creates a view, asks it
if it can draw, and returns the result.

* WebProcess/WebPage/ca/win/LayerTreeHostCAWin.h: Added supportsAcceleratedCompositing.

* WebProcess/WebPage/win/LayerTreeHostWin.cpp: Added.
(WebKit::LayerTreeHost::supportsAcceleratedCompositing): Added. Calls through to
LayerTreeHostCAWin in configurations that support that class. Otherwise just returns false.

* win/WebKit2.vcproj: Added LayerTreeHostWin.cpp. Let VS reorder some other files.

git-svn-id: svn://svn.chromium.org/blink/trunk@82966 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Jian Li.

        [chromium] send-sliced-dragged-file.html started to crash on chromium bots
        https://bugs.webkit.org/show_bug.cgi?id=57539

        * platform/chromium/test_expectations.txt: Removed passing test
          as it has been fixed in the chromium network stack as of chromium r80114
          and it was introduced in r79905 so no need to change the chromium deps
          since it is before the regression.

git-svn-id: svn://svn.chromium.org/blink/trunk@82965 bbb929c8-8fbe-4397-9dbb-9b2b20218538

       Unreviewed, mark unexpectedly passing Linux Chromium tests as such.

        * platform/chromium/test_expectations.txt:

git-svn-id: svn://svn.chromium.org/blink/trunk@82964 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Kenneth Russell.

        Add support for offline audio rendering to AudioContext API
        https://bugs.webkit.org/show_bug.cgi?id=57676

        No new tests since audio API is not yet implemented.

        * DerivedSources.make:
        * WebCore.gypi:
        * WebCore.xcodeproj/project.pbxproj:
        * bindings/js/JSAudioContextCustom.cpp:
        (WebCore::JSAudioContextConstructor::constructJSAudioContext):
        * bindings/js/JSEventCustom.cpp:
        (WebCore::toJS):
        * bindings/js/JSEventTarget.cpp:
        (WebCore::toJS):
        * bindings/v8/V8DOMWrapper.cpp:
        (WebCore::V8DOMWrapper::convertEventTargetToV8Object):
        * bindings/v8/custom/V8AudioContextCustom.cpp:
        (WebCore::V8AudioContext::constructorCallback):
        * bindings/v8/custom/V8EventCustom.cpp:
        (WebCore::toV8):
        * dom/Event.cpp:
        (WebCore::Event::isOfflineAudioCompletionEvent):
        * dom/Event.h:
        * dom/EventTarget.cpp:
        (WebCore::EventTarget::toAudioContext):
        * dom/EventTarget.h:
        * platform/audio/HRTFDatabaseLoader.cpp:
        (WebCore::HRTFDatabaseLoader::waitForLoaderThreadCompletion):
        * platform/audio/HRTFDatabaseLoader.h:
        (WebCore::HRTFDatabaseLoader::loader):
        * webaudio/AudioContext.cpp:
        (WebCore::AudioContext::createOfflineContext):
        (WebCore::AudioContext::AudioContext):
        (WebCore::AudioContext::constructCommon):
        (WebCore::AudioContext::document):
        (WebCore::AudioContext::scriptExecutionContext):
        (WebCore::AudioContext::toAudioContext):
        (WebCore::AudioContext::startRendering):
        (WebCore::AudioContext::fireCompletionEvent):
        * webaudio/AudioContext.h:
        (WebCore::AudioContext::isOfflineContext):
        (WebCore::AudioContext::eventTargetData):
        (WebCore::AudioContext::ensureEventTargetData):
        (WebCore::AudioContext::refEventTarget):
        (WebCore::AudioContext::derefEventTarget):
        * webaudio/AudioContext.idl:
        * webaudio/AudioDestinationNode.cpp:
        (WebCore::AudioDestinationNode::AudioDestinationNode):
        * webaudio/AudioDestinationNode.h:
        (WebCore::AudioDestinationNode::reset):
        (WebCore::AudioDestinationNode::numberOfChannels):
        * webaudio/ConvolverNode.cpp:
        (WebCore::ConvolverNode::setBuffer):
        * webaudio/DefaultAudioDestinationNode.cpp: Added.
        (WebCore::DefaultAudioDestinationNode::DefaultAudioDestinationNode):
        (WebCore::DefaultAudioDestinationNode::~DefaultAudioDestinationNode):
        (WebCore::DefaultAudioDestinationNode::initialize):
        (WebCore::DefaultAudioDestinationNode::uninitialize):
        (WebCore::DefaultAudioDestinationNode::startRendering):
        * webaudio/DefaultAudioDestinationNode.h: Added.
        (WebCore::DefaultAudioDestinationNode::create):
        (WebCore::DefaultAudioDestinationNode::sampleRate):
        * webaudio/OfflineAudioCompletionEvent.cpp: Added.
        (WebCore::OfflineAudioCompletionEvent::create):
        (WebCore::OfflineAudioCompletionEvent::OfflineAudioCompletionEvent):
        (WebCore::OfflineAudioCompletionEvent::~OfflineAudioCompletionEvent):
        (WebCore::OfflineAudioCompletionEvent::isOfflineAudioCompletionEvent):
        * webaudio/OfflineAudioCompletionEvent.h: Added.
        (WebCore::OfflineAudioCompletionEvent::renderedBuffer):
        * webaudio/OfflineAudioCompletionEvent.idl: Added.
        * webaudio/OfflineAudioDestinationNode.cpp: Added.
        (WebCore::OfflineAudioDestinationNode::OfflineAudioDestinationNode):
        (WebCore::OfflineAudioDestinationNode::~OfflineAudioDestinationNode):
        (WebCore::OfflineAudioDestinationNode::initialize):
        (WebCore::OfflineAudioDestinationNode::uninitialize):
        (WebCore::OfflineAudioDestinationNode::startRendering):
        (WebCore::OfflineAudioDestinationNode::renderEntry):
        (WebCore::OfflineAudioDestinationNode::render):
        (WebCore::OfflineAudioDestinationNode::notifyCompleteDispatch):
        (WebCore::OfflineAudioDestinationNode::notifyComplete):
        * webaudio/OfflineAudioDestinationNode.h: Added.
        (WebCore::OfflineAudioDestinationNode::create):
        (WebCore::OfflineAudioDestinationNode::sampleRate):


git-svn-id: svn://svn.chromium.org/blink/trunk@82963 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Eric Seidel.

        [Cairo] Better separate the concerns of GraphicsContextCairo
        https://bugs.webkit.org/show_bug.cgi?id=55150

        Add a PlatformContextCairo which right now stores the cairo_t* for a
        GraphicsContextCairo. Later patches will move logic for tracking ContextShadow
        and image masking layers into this PlatformContextCairo class.

        No new tests. This patch is only a code cleanup.

        * GNUmakefile.am:
        * platform/graphics/GraphicsContext.h: The platform context is no longer a
        cairo_t, but our new class the PlatformContextCairo.
        * platform/graphics/cairo/ContextShadowCairo.cpp: Updated to reflect new class.j
        * platform/graphics/cairo/FontCairo.cpp: Ditto.
        * platform/graphics/cairo/GradientCairo.cpp: Ditto.
        * platform/graphics/cairo/GraphicsContextCairo.cpp: Mostly mechanical
        changes which now reference platformContext()->cr() to get the cairo_t.
        * platform/graphics/cairo/GraphicsContextPlatformPrivateCairo.h: Now hold the
        PlatformContextCairo instead of the cairo_t.
        * platform/graphics/cairo/ImageBufferCairo.cpp: Update to reflect new class.
        * platform/graphics/cairo/ImageCairo.cpp: Ditto.
        * platform/graphics/cairo/PathCairo.cpp: Ditto.
        * platform/graphics/cairo/PlatformContextCairo.cpp: Added.
        * platform/graphics/cairo/PlatformContextCairo.h: Added.
        * platform/graphics/gtk/FontGtk.cpp: Update to reflect new class.
        * platform/graphics/gtk/IconGtk.cpp: Ditto.
        * platform/graphics/win/GraphicsContextCairoWin.cpp: Now fill out
        m_data with a private section containing the platform context instead of
        just a cairo_t.
        * platform/gtk/RenderThemeGtk.cpp: Update to reflect new class.
        * platform/gtk/WidgetRenderingContext.cpp: Ditto.
        (WebCore::WidgetRenderingContext::~WidgetRenderingContext): Ditto.
        * plugins/gtk/PluginViewGtk.cpp: Ditto.
2011-04-05  Martin Robinson  <mrobinson@igalia.com>

        Reviewed by Eric Seidel.

        [Cairo] Better separate the concerns of GraphicsContextCairo
        https://bugs.webkit.org/show_bug.cgi?id=55150

        Add a PlatformContextCairo which right now stores the cairo_t* for a
        GraphicsContextCairo. Later patches will move logic for tracking ContextShadow
        and image masking layers into this PlatformContextCairo class.

        * webkit/webkitwebframe.cpp:
        (draw_page_callback):
        * webkit/webkitwebview.cpp:
        (webkit_web_view_expose_event):
        (webkit_web_view_draw):
2011-04-05  Martin Robinson  <mrobinson@igalia.com>

        Reviewed by Eric Seidel.

        [Cairo] Better separate the concerns of GraphicsContextCairo
        https://bugs.webkit.org/show_bug.cgi?id=55150

        * WebFrame.cpp:
        (hdcFromContext): Modify this method to take PlatformContextCairo
        instead of a cairo_t.
        (WebFrame::spoolPage): Update to reflect new platform context.
        (WebFrame::spoolPages): Ditto.
        * WebFrame.h: Ditto.

git-svn-id: svn://svn.chromium.org/blink/trunk@82962 bbb929c8-8fbe-4397-9dbb-9b2b20218538

       Unreviewed, mark some unexpectedly passing Chromium tests as such.

        * platform/chromium/test_expectations.txt:

git-svn-id: svn://svn.chromium.org/blink/trunk@82961 bbb929c8-8fbe-4397-9dbb-9b2b20218538

LayerTreeHostCAWin uses WKCACFView to render each frame to an image, then has
DrawingAreaImpl send that image over to the UI process (just like it does for
non-accelerated rendering). It's unfortunate that this requires reading every frame back
from the GPU into system memory. More efficient solutions can be explored in the future.

Fixes <http://webkit.org/b/45567>.

Reviewed by Anders Carlsson.

* DerivedSources.make: Added $(WebKit2) to the VPATH so that DerivedSources.make can be
found when used as a target dependency. Added rules to generate a HeaderDetection.h file on
Windows. If we can find WebKitQuartzCoreAdditions headers, we define HAVE_WKQCA in
HeaderDetection.h.

* WebProcess/WebPage/DrawingAreaImpl.cpp:
(WebKit::DrawingAreaImpl::setLayerHostNeedsDisplay):
* WebProcess/WebPage/DrawingAreaImpl.h:
Added setLayerHostNeedsDisplay. Lets the layer host tell the DrawingAreaImpl that it has a
new frame to render. This should only be called by layer hosts that participate in
DrawingAreaImpl's display mechanism.

* WebProcess/WebPage/LayerTreeHost.cpp:
(WebKit::LayerTreeHost::create): Only try to instantiate LayerTreeHostCAWin if we have
WebKitQuartzCoreAdditions, since it depends on that library.

* WebProcess/WebPage/WebPage.cpp:
(WebKit::WebPage::updatePreferences): Removed code that forced accelerated compositing to be
disabled on Windows.

* WebProcess/WebPage/ca/win/LayerTreeHostCAWin.cpp:
(WebKit::registerDummyWindowClass): Registers the window class we use for the dummy window.
(WebKit::createDummyWindow): Creates the dummy window we pass to WKCACFView so that D3D can
determine the display mode, etc. (The window is never shown on screen.)
(WebKit::LayerTreeHostCAWin::LayerTreeHostCAWin): Initialize new members.
(WebKit::LayerTreeHostCAWin::platformInitialize): Create our dummy window and view and
associate the two. We set ourselves as the view's context's user data so that
PlatformCALayer can get our AbstractCACFLayerTreeHost pointer as needed.
(WebKit::LayerTreeHostCAWin::invalidate): Cancel any pending flushes, tear down our view,
and destroy the dummy window if no other layer host is using it.
(WebKit::LayerTreeHostCAWin::scheduleLayerFlush): Ask LayerChangesFlusher to call us back
soon to perform the flush.
(WebKit::LayerTreeHostCAWin::participatesInDisplay): Added. Returns true, since we render
each frame to an image.
(WebKit::LayerTreeHostCAWin::needsDisplay): Added. Returns true if it's now time to
displayReturns true if it's now time to display.
(WebKit::LayerTreeHostCAWin::timeUntilNextDisplay): Added. Returns how many seconds remain
before we need to display again.
(WebKit::size): Added. Helper function to get the size of a WKCACFImage.
(WebKit::toShareableBitmap): Added. Helper function to convert a WKCACFImage to a
ShareableBitmap.
(WebKit::LayerTreeHostCAWin::display): Added. Renders the next frame to an image and stuffs
the image into the UpdateInfo struct.
(WebKit::LayerTreeHostCAWin::sizeDidChange): Added. Tells the view about the new size.
(WebKit::LayerTreeHostCAWin::forceRepaint): Added. Flushes any pending changes to the view.
(WebKit::LayerTreeHostCAWin::contextDidChangeCallback): Added. WKCACFView calls this
whenever any changes made to the view or its layer tree have been flushed. Just calls
through to contextDidChange.
(WebKit::LayerTreeHostCAWin::contextDidChange): Added. Tells layers that they've started
animating, and tells the DrawingAreaImpl that we need to display again.
(WebKit::LayerTreeHostCAWin::rootLayer): Added. Gets the root layer's PlatformCALayer.
(WebKit::LayerTreeHostCAWin::addPendingAnimatedLayer): Added. Stores the layer so that we
can tell it animations have started the next time we get a contextDidChange callback.
(WebKit::LayerTreeHostCAWin::layerTreeDidChange): Added. Schedules a flush, unless we're
already in the process of flushing. This code came from WebCore::WKCACFViewLayerTreeHost.
(WebKit::LayerTreeHostCAWin::flushPendingLayerChangesNow): Added. Performs the flush. This
code was adapted from WebCore::WKCACFViewLayerTreeHost.

* WebProcess/WebPage/ca/win/LayerTreeHostCAWin.h: Wrapped this whole header in HAVE(WKQCA),
since this class only works when WebKitQuartzCoreAdditions is available. Added a bunch of
new members.

git-svn-id: svn://svn.chromium.org/blink/trunk@82960 bbb929c8-8fbe-4397-9dbb-9b2b20218538

When LayerTreeHost participates in display, it renders each frame into a bitmap.
DrawingAreaImpl sends the bitmap to the UI process in an Update message, just like in
non-accelerated compositing mode. The UI process never knows that accelerated compositing is
going on. (When LayerTreeHost does not participate in display, as on Mac, it is responsible
for getting bits from the web process to the UI process.)

No LayerTreeHost uses this mechanism (yet). This patch should cause no change in behavior.

Fixes <http://webkit.org/b/57792> LayerTreeHost needs a way to render frames to a bitmap

Reviewed by Anders Carlsson.

* WebProcess/WebPage/DrawingAreaImpl.cpp:
(WebKit::DrawingAreaImpl::forceRepaint): If the layer tree host participates in display,
perform a display (like in the non-accelerated case).
(WebKit::DrawingAreaImpl::layerHostDidFlushLayers): Don't send an
EnterAcceleratedCompositing message to the UI process if the layer tree host participates in
display. In that case, the UI process doesn't even need to know we're using accelerated
compositing in the web process.
(WebKit::DrawingAreaImpl::sendDidUpdateBackingStoreState): Go through the normal display
mechanism if we have a layer tree host that participates in display.
(WebKit::DrawingAreaImpl::didUpdate): Don't ignore DidUpdate messages when we have a layer
tree host that participates in display.
(WebKit::DrawingAreaImpl::enterAcceleratedCompositingMode): If the new layer tree host
participates in display, let the display timer continue to run and continue to expect
DidUpdate messages so that we don't interrupt the normal display mechanism. We still clear
out m_dirtyRegion, m_scrollRect, and m_scrollOffset, though, because the layer tree host is
still responsible for keeping track of those things even when it participates in display.
(WebKit::DrawingAreaImpl::exitAcceleratedCompositingMode): If the outgoing layer tree host
was participating in display, just send another Update message instead of an
ExitAcceleratedCompositing message, since the UI process never even knew we were using
accelerated compositing.
(WebKit::DrawingAreaImpl::displayTimerFired): If we have a layer tree host that's
participating in display, ask it when it next needs to display when scheduling our timer.

(WebKit::DrawingAreaImpl::scheduleDisplay):
(WebKit::DrawingAreaImpl::display):
If we have a layer tree host that's participating in display, it will keep track of its own
dirty region, so ask it if it needs display rather than checking our own dirty region.

(WebKit::DrawingAreaImpl::display): If we have a layer tree host that's participating in
display, don't bail out. Instead, tell it to display rather than asking the page to paint.
Moved the setting of updateInfo.viewSize earlier so that it will be set even when the layer
tree host is displaying. Other changes are just due to indentation.

* WebProcess/WebPage/LayerTreeHost.h:
(WebKit::LayerTreeHost::participatesInDisplay):
(WebKit::LayerTreeHost::needsDisplay):
(WebKit::LayerTreeHost::timeUntilNextDisplay):
(WebKit::LayerTreeHost::display):
Stubbed out these functions. Derived classes can override them to participate in display.

git-svn-id: svn://svn.chromium.org/blink/trunk@82959 bbb929c8-8fbe-4397-9dbb-9b2b20218538