GitLab

        Reviewed by Xan Lopez.

        https://bugs.webkit.org/show_bug.cgi?id=30883
        [Gtk] Implement AtkText for HTML elements which contain text

        * accessibility/gtk/AccessibilityObjectWrapperAtk.cpp:
        (getInterfaceMaskFromObject):
        (getPangoLayoutForAtk):
        (webkit_accessible_text_get_text):
        * accessibility/gtk/AccessibilityObjectAtk.cpp:
        (AccessibilityObject::accessibilityPlatformIncludesObject):
2010-01-06  Joanmarie Diggs  <joanmarie.diggs@gmail.com>

        Reviewed by Xan Lopez.

        https://bugs.webkit.org/show_bug.cgi?id=30883
        [Gtk] Implement AtkText for HTML elements which contain text

        * tests/testatk.c
        (test_webkit_atk_get_text_at_offset):
        (test_webkit_atk_get_text_at_offset_forms):
        (test_webkit_atk_get_text_at_offset_newlines):

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52890 268f45cc-cd09-0410-ab3c-d52691b4dbfc

https://bugs.webkit.org/show_bug.cgi?id=32794

Reviewed by Sam Weinig.

WebCore: 

Test: platform/mac/fast/text/line-breaking-with-kerning.html

* rendering/RenderBlockLineLayout.cpp:
(WebCore::RenderBlock::findNextLineBreak): When kerning is enabled, include
the trailing space when measuring a word, then subtract its width. This
accounts for kerning between the last glyph of the word and the following space.

LayoutTests: 

* platform/mac/fast/text/line-breaking-with-kerning-expected.checksum: Added.
* platform/mac/fast/text/line-breaking-with-kerning-expected.png: Added.
* platform/mac/fast/text/line-breaking-with-kerning-expected.txt: Added.
* platform/mac/fast/text/line-breaking-with-kerning.html: Added.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52889 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        No review.

        REGRESSION(52854?) fast/workers/shared-worker-constructor.html failed on Leopard Build Bot
        https://bugs.webkit.org/show_bug.cgi?id=33256

        I am unable to reproduce this failure locally, so this fix is only speculative.
        Skip the test we believe is causing the log message which bleeds into fast/workers/shared-worker-constructor.html

        * platform/mac/Skipped:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52888 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Third step of:
        <rdar://problem/6398111> Integrate hardware layers with out-of-process plug-ins layer hosting mechanism

        * Plugins/WebNetscapePluginView.mm:
        (-[WebNetscapePluginView getVariable:value:]): Add case for WKNVSupportsCompositingCoreAnimationPluginsBool.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52887 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        No review, just adding Mike Belshe's new bugzilla email at his request.

        * Scripts/webkitpy/committers.py:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52886 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Second step of:
        <rdar://problem/6398111> Integrate hardware layers with out-of-process plug-ins layer hosting mechanism

        * Plugins/Hosted/WebHostedNetscapePluginView.h: Expose the CALayer of the plugin by adding -pluginLayer method.
        * Plugins/Hosted/WebHostedNetscapePluginView.mm:
        (-[WebHostedNetscapePluginView pluginLayer]): Added.
        * Plugins/WebNetscapePluginView.h: Expose the CALayer of the plugin by adding -pluginLayer method.
        * Plugins/WebNetscapePluginView.mm:
        (-[WebNetscapePluginView pluginLayer]): Added.
        Reviewed by Simon Fraser.

        Second step of:
        <rdar://problem/6398111> Integrate hardware layers with out-of-process plug-ins layer hosting mechanism

        * Plugins/Hosted/WebHostedNetscapePluginView.h: Expose the CALayer of the plugin by adding -pluginLayer method.
        * Plugins/Hosted/WebHostedNetscapePluginView.mm:
        (-[WebHostedNetscapePluginView pluginLayer]): Added.
        * Plugins/WebNetscapePluginView.h: Expose the CALayer of the plugin by adding -pluginLayer method.
        * Plugins/WebNetscapePluginView.mm:
        (-[WebNetscapePluginView pluginLayer]): Added.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52885 268f45cc-cd09-0410-ab3c-d52691b4dbfc

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52884 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        No review.

        REGRESSION(52854?) fast/workers/shared-worker-constructor.html failed on Leopard Build Bot
        https://bugs.webkit.org/show_bug.cgi?id=33256

        I am unable to reproduce this failure locally, so this fix is only speculative.
        Skip the test we believe is causing the log message which bleeds into fast/workers/shared-worker-constructor.html

        * platform/mac/Skipped:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52883 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        No review, rolling out r52877.
        http://trac.webkit.org/changeset/52877
        https://bugs.webkit.org/show_bug.cgi?id=33243

        I screwedd up landing this patch and only partially applied the patch I
        was trying to land.

        * bindings/ScriptControllerBase.cpp:
        * bindings/js/ScriptController.cpp:
        (WebCore::ScriptController::executeScriptInWorld):
        * bindings/v8/ScriptController.cpp:
        (WebCore::ScriptController::ScriptController):
        (WebCore::ScriptController::~ScriptController):
        (WebCore::ScriptController::updateSecurityOrigin):
        (WebCore::ScriptController::evaluateInIsolatedWorld):
        (WebCore::ScriptController::haveInterpreter):
        (WebCore::ScriptController::mainWorldWindowShell):
        (WebCore::ScriptController::clearWindowShell):
        (WebCore::ScriptController::clearForClose):
        (WebCore::ScriptController::destroyWindowShell):
        (WebCore::ScriptController::updateDocument):
        * bindings/v8/ScriptController.h:
        (WebCore::ScriptController::proxy):
        (WebCore::ScriptController::windowShell):
        (WebCore::ScriptController::existingWindowShell):
        (WebCore::ScriptController::evaluateInWorld):
        * bindings/v8/V8DOMWindowShell.cpp:
        (WebCore::V8DOMWindowShell::create):
        (WebCore::V8DOMWindowShell::V8DOMWindowShell):
        (WebCore::V8DOMWindowShell::initContextIfNeeded):

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52882 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Eric Carlson.

        media/video-seeking.html is flaky
        https://bugs.webkit.org/show_bug.cgi?id=33273

        Removed setTimeouts from this test. It kind of looks
        like they were intentional, but I don't see how they add
        anything to this test (other than race conditions).

        * media/video-seeking-expected.txt:
        * media/video-seeking.html:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52881 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Unreviewed.  Rolling back in a change I should not have rolled out.

        The original commit http://trac.webkit.org/changeset/52862
        was rolled out again in http://trac.webkit.org/changeset/52871
        but I now believe that rollout to be in error.
        My sincerest apologies for the churn.

        Fix the bug in previous commit: we shouldn't finish
        decoding if "sizeOnly" is true.
        https://bugs.webkit.org/show_bug.cgi?id=33258

        * platform/image-decoders/png/PNGImageDecoder.cpp:
        (WebCore::PNGImageReader::decode):

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52880 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Eric Seidel.

        fast/css/transition-color-unspecified.html and fast/dom/HTMLObjectElement/object-as-frame.html are flaky
        https://bugs.webkit.org/show_bug.cgi?id=33270

        These both rely on setTimeouts where they should be using events.

        * fast/css/transition-color-unspecified.html:
        * fast/dom/HTMLObjectElement/object-as-frame.html:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52879 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Eric Seidel.

        platform/chromium/fast/dom/interval.html is flaky
        https://bugs.webkit.org/show_bug.cgi?id=33274

        We're seeing the debug bots take >100000 microseconds to stop
        and Interval.

        * platform/chromium/fast/dom/interval.html:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52878 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Eric Seidel.

        [v8] Let ScriptController have more than one windowShell
        https://bugs.webkit.org/show_bug.cgi?id=33243

        This patch lets ScriptController have more than one windowShell.  We're
        currently only using one of them (for the main world), but this patch
        lets us use other ones for isolated worlds soon.

        * bindings/ScriptControllerBase.cpp:
        (WebCore::ScriptController::executeScriptInWorld):
        * bindings/js/ScriptController.cpp:
        * bindings/v8/ScriptController.cpp:
        (WebCore::ScriptController::createWorld):
        (WebCore::ScriptController::initScript):
        (WebCore::ScriptController::ScriptController):
        (WebCore::ScriptController::~ScriptController):
        (WebCore::ScriptController::updateSecurityOrigin):
        (WebCore::ScriptController::evaluateInIsolatedWorld):
        (WebCore::ScriptController::evaluateInWorld):
        (WebCore::ScriptController::mainWorldWindowShell):
        (WebCore::ScriptController::clearWindowShell):
        (WebCore::ScriptController::clearForClose):
        (WebCore::ScriptController::destroyWindowShell):
        (WebCore::ScriptController::updateDocument):
        * bindings/v8/ScriptController.h:
        (WebCore::ScriptController::windowShell):
        (WebCore::ScriptController::existingWindowShell):
        (WebCore::ScriptController::globalObject):
        (WebCore::ScriptController::proxy):
        * bindings/v8/V8DOMWindowShell.cpp:
        (WebCore::V8DOMWindowShell::create):
        (WebCore::V8DOMWindowShell::V8DOMWindowShell):
        (WebCore::V8DOMWindowShell::initContextIfNeeded):
        * bindings/v8/V8DOMWindowShell.h:
        * bindings/v8/V8DOMWrapper.cpp:
        (WebCore::V8DOMWrapper::instantiateV8Object):
        (WebCore::V8DOMWrapper::convertNewNodeToV8Object):

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52877 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Unreviewed.  Rolling out r52869 and r52853 due to bot
        and local run-webkit-test failures

        REGRESSION(52854?) fast/workers/shared-worker-constructor.html failed on Leopard Build Bot
        https://bugs.webkit.org/show_bug.cgi?id=33256

        The original bug was https://bugs.webkit.org/show_bug.cgi?id=33153

        * Scripts/run-webkit-tests:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52876 268f45cc-cd09-0410-ab3c-d52691b4dbfc

* WebCore.base.exp: Export symbols WebKit would like to use.
* plugins/PluginWidget.h:
(WebCore::PluginWidget::PluginWidget): Redeclare a constructor.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52875 268f45cc-cd09-0410-ab3c-d52691b4dbfc

* WebCoreSupport/WebFrameLoaderClient.mm:



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52874 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Unreviewed "build" fix.  Just adding missing include.

        bugzilla-tool rollout --complete-rollout should make a nicer bug comment
        https://bugs.webkit.org/show_bug.cgi?id=29212

        * Scripts/webkitpy/steps/completerollout.py:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52873 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        
        First step toward:
        <rdar://problem/6398111> Integrate hardware layers with out-of-process plug-ins layer hosting mechanism
        
        Small refactoring moving the PluginWidget class from WebKit to WebCore.          
        * WebCore.xcodeproj/project.pbxproj: Added new source files and a "mac" group; exposes PluginWidget.h as a "Private" role
        * platform/Widget.h: Added isPluginWidget(). 
        (WebCore::Widget::isPluginWidget):
        * plugins/PluginWidget.h: Added.
        (WebCore::PluginWidget::isPluginWidget): Added as an interim solution until Mac plug-ins inheirt from PluginView.
        * plugins/mac/PluginWidgetMac.mm: Added.
        (WebCore::PluginWidget::invalidateRect): Added. Extracted from the old WebKit/FrameLoaderClientMac version of PluginWidget

        Reviewed by Eric Carlson.

        First step toward:
        <rdar://problem/6398111> Integrate hardware layers with out-of-process plug-ins layer hosting mechanism
        
        Small refactoring moving the PluginWidget class from WebKit to WebCore.

        * WebCoreSupport/WebFrameLoaderClient.mm: Use the WebCore version of PluginWidget.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52872 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        No review, rolling out r52862.
        http://trac.webkit.org/changeset/52862
        https://bugs.webkit.org/show_bug.cgi?id=33258

        Caused svg/W3C-SVG-1.1/linking-a-01-b.svg to crash on the
        Leopard Release builder.

        * platform/image-decoders/png/PNGImageDecoder.cpp:
        (WebCore::PNGImageReader::decode):

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52871 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Adam Barth.

        bugzilla-tool rollout --complete-rollout should make a nicer bug comment
        https://bugs.webkit.org/show_bug.cgi?id=29212

        * Scripts/webkitpy/bugzilla.py:
         - Make reopen_bug robust against the bug already being open.
        * Scripts/webkitpy/commands/download.py:
         - Remove dead code from Rollout.
        * Scripts/webkitpy/steps/completerollout.py:
         - Improve the rollout comment.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52870 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Darin Adler.

        Set isHttpdOpen to 0 if pidfile does not exist for some reason.
        Speculative fix for https://bugs.webkit.org/show_bug.cgi?id=33256.

        * Scripts/run-webkit-tests:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52869 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Not reviewed. Revert the part of my last commit touching Base.xcconfig, was not meant to be commited.

        * Configurations/Base.xcconfig:


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52868 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Dimitri Glazkov.

        Generate header declarations for V8 binding security checks
        and delete V8CustomBinding.cpp

        * Android.v8bindings.mk: Removed V8CustomBinding.cpp.
        * WebCore.gypi: Removed V8CustomBinding.cpp.
        * bindings/scripts/CodeGeneratorV8.pm: Generate security check declarations.
        * bindings/v8/V8DOMWindowShell.cpp:
        (WebCore::getTargetFrame): Moved from V8CustomBinding.cpp.
        (WebCore::reportUnsafeJavaScriptAccess):
        (WebCore::V8DOMWindowShell::createNewContext):
        (WebCore::V8DOMWindowShell::setLocation): Moved from V8CustomBinding.cpp.
        * bindings/v8/V8DOMWindowShell.h:
        * bindings/v8/V8DOMWrapper.cpp:
        (WebCore::downcastSVGPathSeg): Moved from V8CustomBinding.cpp.
        (WebCore::V8DOMWrapper::convertSVGObjectWithContextToV8Object):
        (WebCore::V8DOMWrapper::getTemplate):
        * bindings/v8/custom/V8CustomBinding.cpp: Removed.
        * bindings/v8/custom/V8CustomBinding.h:
        * bindings/v8/custom/V8DOMWindowCustom.cpp:
        (WebCore::V8DOMWindow::locationAccessorSetter):
        (WebCore::V8DOMWindow::namedSecurityCheck):
        (WebCore::V8DOMWindow::indexedSecurityCheck):
        * bindings/v8/custom/V8DocumentLocationCustom.cpp:
        (WebCore::V8Document::locationAccessorSetter):
        * bindings/v8/custom/V8HistoryCustom.cpp:
        (WebCore::V8History::indexedSecurityCheck):
        (WebCore::V8History::namedSecurityCheck):
        * bindings/v8/custom/V8LocationCustom.cpp:
        (WebCore::V8Location::indexedSecurityCheck):
        (WebCore::V8Location::namedSecurityCheck):


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52867 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Dirk Schulze.

        Marker code is buggy: referencePoint translation is off
        https://bugs.webkit.org/show_bug.cgi?id=33012

        Rewrite marker code, as it had many issues:
        - Correct refX/refY handling (covered by new marker-referencePoint.svg test)
        - Calculate marker bounds on layout time, instead of paint time, to deliver correct repaint rects
        - Cleanup RenderPath code, which was cluttered by marker stuff (outlay in SVGMarkerLayoutInfo)
        - Fix SVGResource assuming that there is only one resource per resource type per element. Markers
          can have three different resources of the same type (three markers, start/mid/end markers)

        Filed bug 33115 (PathElementAddQuadCurveToPoint not handled for <marker>), which was just a TODO in the code before.

        Fixes all svg/custom/marker*.svg tests, that use dynamic updates (propagation failed before).
        Test: svg/custom/marker-referencePoint.svg
              svg/custom/marker-opacity.svg
              svg/custom/marker-strokeWidth-changes.svg

        * Android.mk: Add SVGMarkerData.h / SVGMarkerLayoutInfo.(cpp,h) to build.
        * GNUmakefile.am: Ditto.
        * WebCore.gypi: Ditto.
        * WebCore.pro: Ditto.
        * WebCore.vcproj/WebCore.vcproj: Ditto.
        * WebCore.xcodeproj/project.pbxproj: Ditto.
        * page/FrameView.cpp: Notify RenderSVGRoot children about viewport size changes, just like it's handled for <body> in HTML.
        (WebCore::FrameView::layout): This is important, as marker bounding boxes may depend on the current viewport size through relative coordinates used in <svg>.
        * rendering/RenderPath.cpp:
        (WebCore::RenderPath::markerBoundingBox): Calculates and caches marker boundaries in the layout() phase.
        (WebCore::RenderPath::repaintRectInLocalCoordinates): Use the cached marker bounding box during paint() time.
        (WebCore::RenderPath::setPath): Clear marker boundaries upon path changes.
        (WebCore::RenderPath::paint): Use SVGMarkerLayoutInfo::drawMarkers() instead of the local drawMarkersIfNeeded() function, which is gone now.
        (WebCore::RenderPath::calculateMarkerBoundsIfNeeded): Utilized by markerBoundingBox() to request SVGResourceMarker objects and calculate their boundaries.
        * rendering/RenderPath.h: Store SVGMarkerLayoutInfo objects, and rename m_markerBounds to m_cachedLocalMarkerBBox for consistency.
        (WebCore::RenderPath::path): Inline this accessor.
        * rendering/RenderSVGContainer.h: Make paint() public (to be used from SVGResourceMarker), all other methods protected (to be used from RenderSVGViewportContainer).
        * rendering/RenderSVGViewportContainer.cpp: Remove useless paint() override, as the comment in the code indicated.
        (WebCore::RenderSVGViewportContainer::markerBoundaries): Maps the marker contents repaintRectInLocalCoordinates() into the final coordinate system (see code).
        (WebCore::RenderSVGViewportContainer::markerContentTransformation): Add marker-specific helper method, allowing to make viewportTransform() private again.
        * rendering/RenderSVGViewportContainer.h: Marking viewportTransform() private, it's never meant to be used outside this renderer.
        * rendering/SVGMarkerData.h: Added. Refactored from existing RenderPath code, preserving original copyrights.
        (WebCore::SVGMarkerData::): Helper struct for SVGMarkerLayoutInfo.
        (WebCore::SVGMarkerData::origin): Holds the final computed destination of a marker.
        (WebCore::SVGMarkerData::marker): Holds a pointer to the associated SVGResourceMarker object, which we're aiming to layout.
        (WebCore::SVGMarkerData::currentAngle): Calculates the current angle for a certain marker type (start/mid/end) based upon the computed inslope/outslope values.
        (WebCore::SVGMarkerData::updateTypeAndMarker): Helper method.
        (WebCore::SVGMarkerData::updateOutslope): Ditto.
        (WebCore::SVGMarkerData::updateMarkerDataForPathElement): Does the actual inslope/outslope calculation.
        (WebCore::SVGMarkerData::updateInslope): Helper method.
        * rendering/SVGMarkerLayoutInfo.cpp: Added. New approach to build marker layout info, this time correct during layout() time instead of paint() time.
        (WebCore::SVGMarkerLayoutInfo::SVGMarkerLayoutInfo):
        (WebCore::SVGMarkerLayoutInfo::~SVGMarkerLayoutInfo):
        (WebCore::processStartAndMidMarkers): Static helper method used to visit each path element in order to calculate inslopes/outslopes (-> orientation angles)
        (WebCore::SVGMarkerLayoutInfo::calculateBoundaries): Build list of MarkerLayout objects, used to query boundaries, and to paint markers during drawMarkers().
        (WebCore::SVGMarkerLayoutInfo::drawMarkers): Cheap method walking existing MarkerLayout vector, drawing the individual RenderSVGViewportContainer of the markers.
        (WebCore::SVGMarkerLayoutInfo::addLayoutedMarker): Helper method.
        * rendering/SVGMarkerLayoutInfo.h: Added.
        (WebCore::MarkerLayout::MarkerLayout): Helper struct storing a pair of SVGResourceMarker* and a TranformationMatrix.
        (WebCore::SVGMarkerLayoutInfo::markerData): Public helper methods returning references to private member variables, used from processStartAndMidMarkers.
        (WebCore::SVGMarkerLayoutInfo::midMarker): Ditto.
        (WebCore::SVGMarkerLayoutInfo::elementIndex): Ditto.
        * svg/SVGMarkerElement.cpp: Adopt to name change: setRef() -> setReferencePoint(). setMarker() -> setRenderer()
        (WebCore::SVGMarkerElement::~SVGMarkerElement): Remove obsolete detach() hack, fixed by SVGUseElement changes.
        (WebCore::SVGMarkerElement::canvasResource): Add ASSERT(renderer()) to clarify that canvasResource() is only meant to used after renderer creation.
        * svg/SVGSVGElement.cpp:
        (WebCore::SVGSVGElement::~SVGSVGElement): Remove obsolete detach() hack, fixed by SVGUseElement changes.
        * svg/SVGUseElement.cpp: Rework buildPendingResource() to remove SVGSVGElement/SVGMarkerElement detach() hacks (called manually from destructors), see below.
        (WebCore::SVGUseElement::buildPendingResource): Always detach() the shadow tree root element, before destructing it - this caused the regression with this patch before.
        * svg/graphics/SVGResource.cpp: Instead of maintaining a global hashmap between SVGStyledElements and an array of possible SVGResources, use a HashSet of SVGResources
                                        and directly scan the already existing list of clients for a certain SVGResource. This wrong approach has been introduced over two years
                                        ago, making the assumption that there's only one resource type per element. Though markers can provide three resources of the same type
                                        per element (all SVGResourceMarker objects, but in different rules: start/mid/end marker). That information is only available while painting.
        (WebCore::resourceSet): Add global static HashSet<SVGResource*>.
        (WebCore::SVGResource::SVGResource): Add us to the resourceSet.
        (WebCore::SVGResource::~SVGResource): Remove us from resourceSet.
        (WebCore::SVGResource::removeClient): Traverse all SVGResource clients and remove the passed SVGStyledElement from the client list (invoked by SVGStyledElement destructor).
        (WebCore::SVGResource::addClient): Remove the offending part of the old approach, making the wrong assumption regarding resource types.
        * svg/graphics/SVGResourceMarker.cpp:
        (WebCore::SVGResourceMarker::SVGResourceMarker):
        (WebCore::SVGResourceMarker::markerTransformation): Create TransformationMatrix in the local marker content (RenderSVGViewportContainer) coordinate system, to position a marker.
        (WebCore::SVGResourceMarker::draw): Simplify marker drawing a lot: used the passing TransformationMatrix (cached in SVGMarkerLayoutInfo) to draw the marker content renderer.
        (WebCore::SVGResourceMarker::externalRepresentation): Adopt to refX/refY -> referencePoint changes.
        * svg/graphics/SVGResourceMarker.h: Store a FloatPoint m_referecePoint instead of two refX/refY float values.
        (WebCore::SVGResourceMarker::setRenderer): Renamed from setMarker().
        (WebCore::SVGResourceMarker::setReferencePoint): Renamed from setRef().
        (WebCore::SVGResourceMarker::referencePoint): New accessor replacing refX() / refY().


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52866 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Nikolas Zimmermann.

        SVG feImage support
        https://bugs.webkit.org/show_bug.cgi?id=31905

        This implements the SVG Filter effect feImage with support of fragment urls.
        It also includes a bug fix for feComposite. feComposite didn't cover values
        bigger than 255 correctly on composite oeprator arithmetic.

        Tests: There are already many feImage tests in trunk. They just needed updated
               pixel test results.
               The feComposite bug is covered by svg/W3C-SVG-1.1/filters-composite-02-b.svg
               and depends on feImage.

        * platform/graphics/filters/FEComposite.cpp:
        (WebCore::arithmetic):
        * rendering/RenderSVGImage.cpp:
        (WebCore::RenderSVGImage::paint):
        * rendering/RenderSVGImage.h:
        * svg/SVGFEImageElement.cpp:
        (WebCore::SVGFEImageElement::requestImageResource):
        (WebCore::SVGFEImageElement::parseMappedAttribute):
        (WebCore::SVGFEImageElement::notifyFinished):
        (WebCore::SVGFEImageElement::build):
        * svg/SVGFEImageElement.h:
        * svg/SVGFEImageElement.idl:
        * svg/SVGPreserveAspectRatio.cpp:
        (WebCore::SVGPreserveAspectRatio::transformRect):
        * svg/SVGPreserveAspectRatio.h:
        * svg/graphics/filters/SVGFEImage.cpp:
        (WebCore::FEImage::FEImage):
        (WebCore::FEImage::create):
        (WebCore::FEImage::apply):
        * svg/graphics/filters/SVGFEImage.h:

2010-01-06  Dirk Schulze  <krit@webkit.org>

        Reviewed by Nikolas Zimmermann.

        SVG feImage support
        https://bugs.webkit.org/show_bug.cgi?id=31905

        The implementation of feImage requires pixel test result updates of the already
        available tests.

        There is also an update for svg/filters/feComposite.svg, since the patch also
        fixed a bug in feComposite.

        * platform/mac/svg/W3C-SVG-1.1/filters-composite-02-b-expected.checksum:
        * platform/mac/svg/W3C-SVG-1.1/filters-composite-02-b-expected.png:
        * platform/mac/svg/W3C-SVG-1.1/filters-displace-01-f-expected.checksum:
        * platform/mac/svg/W3C-SVG-1.1/filters-displace-01-f-expected.png:
        * platform/mac/svg/W3C-SVG-1.1/filters-image-01-b-expected.checksum:
        * platform/mac/svg/W3C-SVG-1.1/filters-image-01-b-expected.png:
        * platform/mac/svg/custom/feDisplacementMap-01-expected.checksum:
        * platform/mac/svg/custom/feDisplacementMap-01-expected.png:
        * platform/mac/svg/filters/feComposite-expected.checksum:
        * platform/mac/svg/filters/feComposite-expected.png:
        * platform/mac/svg/webarchive/svg-feimage-subresources-expected.checksum:
        * platform/mac/svg/webarchive/svg-feimage-subresources-expected.png:

2010-01-06  Dirk Schulze  <krit@webkit.org>

        Reviewed by Nikolas Zimmermann.

        SVG feImage support
        https://bugs.webkit.org/show_bug.cgi?id=31905

        Update SVG status page. We support feDisplacementMap and feImage now.

        * projects/svg/status.xml:


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52865 268f45cc-cd09-0410-ab3c-d52691b4dbfc

https://bugs.webkit.org/show_bug.cgi?id=32295

* platform/mac/fast/repaint/block-no-inflow-children-expected.png: Added.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52864 268f45cc-cd09-0410-ab3c-d52691b4dbfc

https://bugs.webkit.org/show_bug.cgi?id=33239

Reviewed by Eric Seidel.

* fast/dom/StyleSheet/get-stylesheet-byname.html: Updated to use
a standard test wrapper instead of a custom one.
* fast/dom/StyleSheet/script-tests/get-stylesheet-byname.js:
Changed test to not depend on a non-standard wrapper.
* fast/dom/StyleSheet/get-stylesheet-byname-expected.txt: Updated
results since I changed the test around.

* platform/mac/editing/selection/script-tests/TEMPLATE.html: Fix
incorrect path in template that broke the test when we called
make-script-test-wrappers.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52863 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Darin Adler.

        Fix the bug in previous commit: we shouldn't finish
        decoding if "sizeOnly" is true.
        https://bugs.webkit.org/show_bug.cgi?id=33258

        * platform/image-decoders/png/PNGImageDecoder.cpp:
        (WebCore::PNGImageReader::decode):

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52862 268f45cc-cd09-0410-ab3c-d52691b4dbfc

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52861 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        No review, rolling out r52855.
        http://trac.webkit.org/changeset/52855
        https://bugs.webkit.org/show_bug.cgi?id=33222

        Caused 6 layout tests to fail on Mac

        * loader/DocumentLoader.cpp:
        (WebCore::DocumentLoader::substituteResourceDeliveryTimerFired):
        * loader/MainResourceLoader.cpp:
        (WebCore::MainResourceLoader::continueAfterContentPolicy):
        * loader/ResourceLoader.cpp:
        * loader/ResourceLoader.h:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52860 268f45cc-cd09-0410-ab3c-d52691b4dbfc

* runtime/Identifier.cpp:
(JSC::createIdentifierTableSpecific):



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52859 268f45cc-cd09-0410-ab3c-d52691b4dbfc

* JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.def:



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52858 268f45cc-cd09-0410-ab3c-d52691b4dbfc

* runtime/Identifier.cpp:
(JSC::createIdentifierTableSpecificCallback):



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52857 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Remove m_identifierTable pointer from UString

Reviewed by Sam Weinig.

JavaScriptCore: 

Currently every string holds a pointer so that during destruction,
if a string has been used as an identifier, it can remove itself
from the table.  By instead accessing the identifierTable via a
thread specific tracking the table associated with the current
globaldata, we can save the memory cost of this pointer.

* API/APIShims.h:
(JSC::APIEntryShimWithoutLock::APIEntryShimWithoutLock):
(JSC::APIEntryShimWithoutLock::~APIEntryShimWithoutLock):
(JSC::APICallbackShim::APICallbackShim):
(JSC::APICallbackShim::~APICallbackShim):

    - change the API shims to track the identifierTable of the current JSGlobalData.

* API/JSContextRef.cpp:
(JSContextGroupCreate):

    - update creation of JSGlobalData for API usage to use new create method.
    - fix shim instanciation bug in JSGlobalContextCreateInGroup.

* JavaScriptCore.exp:
* runtime/Completion.cpp:
(JSC::checkSyntax):
(JSC::evaluate):

    - add asserts to check the identifierTable is being tracked correctly.

* runtime/Identifier.cpp:
(JSC::IdentifierTable::~IdentifierTable):
(JSC::IdentifierTable::add):
(JSC::Identifier::remove):
(JSC::Identifier::checkSameIdentifierTable):
(JSC::createIdentifierTableSpecificCallback):
(JSC::createIdentifierTableSpecific):
(JSC::createDefaultDataSpecific):

    - Use currentIdentifierTable() instead of UStringImpl::m_identifierTable.
    - Define methods to access the thread specific identifier tables.

* runtime/Identifier.h:
(JSC::ThreadIdentifierTableData::ThreadIdentifierTableData):
(JSC::defaultIdentifierTable):
(JSC::setDefaultIdentifierTable):
(JSC::currentIdentifierTable):
(JSC::setCurrentIdentifierTable):
(JSC::resetCurrentIdentifierTable):

    - Declare methods to access the thread specific identifier tables.

* runtime/JSGlobalData.cpp:
(JSC::JSGlobalData::createNonDefault):
(JSC::JSGlobalData::create):
(JSC::JSGlobalData::sharedInstance):

    - creation of JSGlobalData objects, other than for API usage, associate themselves with the current thread.

* runtime/JSGlobalData.h:
* runtime/UStringImpl.cpp:
(JSC::UStringImpl::destroy):

    - destroy() method should be using isIdentifier().

* runtime/UStringImpl.h:
(JSC::UStringImpl::isIdentifier):
(JSC::UStringImpl::setIsIdentifier):
(JSC::UStringImpl::checkConsistency):
(JSC::UStringImpl::UStringImpl):

    - replace m_identifierTable with a single m_isIdentifier bit.

* wtf/StringHashFunctions.h:
(WTF::stringHash):

    - change string hash result from 32-bit to 31-bit, to free a bit in UStringImpl for m_isIdentifier.

JavaScriptGlue: 

Add API shims similar to those used in the JSC API to track the current identifierTable.

* JSBase.cpp:
(JSBase::Release):
* JSUtils.cpp:
(JSObjectKJSValue):
(KJSValueToCFTypeInternal):
(unprotectGlobalObject):
(JSGlueAPIEntry::JSGlueAPIEntry):
(JSGlueAPIEntry::~JSGlueAPIEntry):
(JSGlueAPICallback::JSGlueAPICallback):
(JSGlueAPICallback::~JSGlueAPICallback):
* JSUtils.h:
* JSValueWrapper.cpp:
(JSValueWrapper::JSObjectCopyPropertyNames):
(JSValueWrapper::JSObjectCopyProperty):
(JSValueWrapper::JSObjectSetProperty):
(JSValueWrapper::JSObjectCallFunction):
(JSValueWrapper::JSObjectCopyCFValue):
* JavaScriptGlue.cpp:
(JSRunCreate):
(JSRunEvaluate):
(JSRunCheckSyntax):
(JSCollect):
* JavaScriptGlue.xcodeproj/project.pbxproj:
* UserObjectImp.cpp:
(UserObjectImp::callAsFunction):



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52856 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Darin Adler.

        Let ReourceLoader deliver segmented data
        when loading resource from a SharedBuffer object.
        https://bugs.webkit.org/show_bug.cgi?id=33222

        * loader/DocumentLoader.cpp:
        (WebCore::DocumentLoader::substituteResourceDeliveryTimerFired):
        * loader/MainResourceLoader.cpp:
        (WebCore::MainResourceLoader::continueAfterContentPolicy):
        * loader/ResourceLoader.cpp:
        (WebCore::ResourceLoader::didReceiveData):
        * loader/ResourceLoader.h:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52855 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Prefix the phonon includes with phonon/ to avoid conflicts with the S60
audio routing API ( http://wiki.forum.nokia.com/index.php/Audio_Routing_API ).

* platform/graphics/qt/MediaPlayerPrivatePhonon.cpp:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52854 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Darin Adler.

        Change hardcoded /tmp in run-webkit-tests to File::Spec->tmpdir(),
        and remove the httpd's pidfile directory if httpd terminated.

        * Scripts/run-webkit-tests:


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52853 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Fix release build bustage.

        * src/WebKit.cpp: Need to include Logging.h



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52852 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Adam Barth.

        [Chromium] Expose more APIs to eliminate WebCore usage in Chromium.
        https://bugs.webkit.org/show_bug.cgi?id=33249

        * WebKit.gyp:
        * public/WebFrame.h:
        * public/WebGlyphCache.h: Added.
        * public/WebKit.h:
        * src/WebFrameImpl.cpp:
        (WebKit::WebFrame::instanceCount):
        (WebKit::WebFrameImpl::WebFrameImpl):
        (WebKit::WebFrameImpl::~WebFrameImpl):
        * src/WebFrameImpl.h:
        * src/WebGlyphCache.cpp: Added.
        (WebKit::WebGlyphCache::pageCount):
        * src/WebKit.cpp:
        (WebKit::enableLogChannel):



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@52851 268f45cc-cd09-0410-ab3c-d52691b4dbfc