GitLab

        Reviewed by Yury Semikhatsky.

        Web Inspector: add source column field to debugger call frames.
        https://bugs.webkit.org/show_bug.cgi?id=52443

        * bindings/js/JavaScriptCallFrame.cpp:
        (WebCore::JavaScriptCallFrame::JavaScriptCallFrame):
        * bindings/js/JavaScriptCallFrame.h:
        (WebCore::JavaScriptCallFrame::create):
        (WebCore::JavaScriptCallFrame::line):
        (WebCore::JavaScriptCallFrame::column):
        (WebCore::JavaScriptCallFrame::update):
        * bindings/js/ScriptDebugServer.cpp:
        (WebCore::ScriptDebugServer::createCallFrameAndPauseIfNeeded):
        (WebCore::ScriptDebugServer::updateCallFrameAndPauseIfNeeded):
        (WebCore::ScriptDebugServer::callEvent):
        (WebCore::ScriptDebugServer::atStatement):
        (WebCore::ScriptDebugServer::returnEvent):
        (WebCore::ScriptDebugServer::exception):
        (WebCore::ScriptDebugServer::willExecuteProgram):
        (WebCore::ScriptDebugServer::didExecuteProgram):
        (WebCore::ScriptDebugServer::didReachBreakpoint):
        * bindings/js/ScriptDebugServer.h:
        * bindings/v8/DebuggerScript.js:
        ():
        * bindings/v8/JavaScriptCallFrame.cpp:
        (WebCore::JavaScriptCallFrame::column):
        * bindings/v8/JavaScriptCallFrame.h:
        * inspector/InjectedScriptSource.js:
        ():
        * inspector/JavaScriptCallFrame.idl:

 2011-01-18  Antti Koivisto  <antti@apple.com>

        Reviewed by Dave Hyatt.

        https://bugs.webkit.org/show_bug.cgi?id=52370
        <rdar://problem/8856808>
        Style sharing optimization no longer works on major web sites

        The code in CSSStyleSelector::locateSharedStyle() that tries to share style information between element has been defeated
        by widespread use of certain CSS selectors (:first-child pseudo class and similar). The current implementation disables
        the sharing optimization for the whole page if one of these constructs is seen in any style sheet used by the page.

        This patch gets the style sharing back to 25-40% range on most web sites.

        * css/CSSMutableStyleDeclaration.h:
        (WebCore::CSSMutableStyleDeclaration::propertiesEqual):

            Needed to improve equality checking in NamedNodeMap::mappedMapsEquivalent.

        * css/CSSSelector.h:
        (WebCore::CSSSelector::isSiblingSelector):

            Test for selectors that are affected by the sibling elements.

        * css/CSSStyleSelector.cpp:
        (WebCore::elementCanUseSimpleDefaultStyle):
        (WebCore::collectSiblingRulesInDefaultStyle):
        (WebCore::assertNoSiblingRulesInDefaultStyle):

            Extract sibling rules from the MathML default stylesheet.
            Assert that no other default stylesheet has them.

        (WebCore::CSSStyleSelector::CSSStyleSelector):
        (WebCore::CSSStyleSelector::locateCousinList):
        (WebCore::CSSStyleSelector::matchesSiblingRules):
        (WebCore::CSSStyleSelector::canShareStyleWithElement):
        (WebCore::CSSStyleSelector::locateSharedStyle):
        (WebCore::CSSStyleSelector::styleForElement):
        (WebCore::collectIdsAndSiblingRulesFromList):
        (WebCore::CSSRuleSet::collectIdsAndSiblingRules):

            Track sibling rules and ids used in the stylesheets to allow much more fine-grained rejection of cases
            where style sharing can't be used.

        * css/CSSStyleSelector.h:
        * dom/NamedNodeMap.cpp:
        (WebCore::NamedNodeMap::mappedMapsEquivalent):

            Check also for CSSValue mutation from DOM.

        * mathml/MathMLMathElement.cpp:
        (WebCore::MathMLMathElement::insertedIntoDocument):
        * mathml/MathMLMathElement.h:

            MathML default style sheet has sibling rules which don't get noticed by the normal document
            stylesheet parsing.

        * rendering/style/RenderStyle.h:
        (WebCore::InheritedFlags::setFirstChildState):
        (WebCore::InheritedFlags::setLastChildState):
        (WebCore::InheritedFlags::setChildIndex):

            These all make style non-shareable (unique).

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@76013 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Reviewed by Dave Hyatt.

https://bugs.webkit.org/show_bug.cgi?id=52370
<rdar://problem/8856808>
Style sharing optimization no longer works on major web sites
        
The code in CSSStyleSelector::locateSharedStyle() that tries to share style information between element has been defeated 
by widespread use of certain CSS selectors (:first-child pseudo class and similar). The current implementation disables
the sharing optimization for the whole page if one of these constructs is seen in any style sheet used by the page.  
        
This patch gets the style sharing back to 25-40% range on most web sites.

* css/CSSMutableStyleDeclaration.h:
(WebCore::CSSMutableStyleDeclaration::propertiesEqual):
            
    Needed to improve equality checking in NamedNodeMap::mappedMapsEquivalent.
        
* css/CSSSelector.h:
(WebCore::CSSSelector::isSiblingSelector):
        
    Test for selectors that are affected by the sibling elements.
        
* css/CSSStyleSelector.cpp:
(WebCore::elementCanUseSimpleDefaultStyle):
(WebCore::collectSiblingRulesInDefaultStyle):
(WebCore::assertNoSiblingRulesInDefaultStyle):
            
    Extract sibling rules from the MathML default stylesheet. 
    Assert that no other default stylesheet has them.
            
(WebCore::CSSStyleSelector::CSSStyleSelector):
(WebCore::CSSStyleSelector::locateCousinList):
(WebCore::CSSStyleSelector::matchesSiblingRules):
(WebCore::CSSStyleSelector::canShareStyleWithElement):
(WebCore::CSSStyleSelector::locateSharedStyle):
(WebCore::CSSStyleSelector::styleForElement):
(WebCore::collectIdsAndSiblingRulesFromList):
(WebCore::CSSRuleSet::collectIdsAndSiblingRules):

    Track sibling rules and ids used in the stylesheets to allow much more fine-grained rejection of cases
    where style sharing can't be used.

* css/CSSStyleSelector.h:
* dom/NamedNodeMap.cpp:
(WebCore::NamedNodeMap::mappedMapsEquivalent):
            
    Check also for CSSValue mutation from DOM. 
            
* mathml/MathMLMathElement.cpp:
(WebCore::MathMLMathElement::insertedIntoDocument):
* mathml/MathMLMathElement.h:
        
    MathML default style sheet has sibling rules which don't get noticed by the normal document
    stylesheet parsing.
        
* rendering/style/RenderStyle.h:
(WebCore::InheritedFlags::setFirstChildState):
(WebCore::InheritedFlags::setLastChildState):
(WebCore::InheritedFlags::setChildIndex):
        
    These all make style non-shareable (unique).

LayoutTests: https://bugs.webkit.org/show_bug.cgi?id=52370
Style sharing optimization no longer works on major web sites
        
Reviewed by Dave Hyatt.

These are progressions (with no noticable visual change). 
MathML code was improperly sharing style.

* platform/mac/mathml/presentation/fractions-vertical-alignment-expected.txt:
* platform/mac/mathml/presentation/subsup-expected.txt:



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@76012 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Unreviewed crash fix on release builds.

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

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@76011 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Unreviewed, update localizedStrings.js.

        * English.lproj/localizedStrings.js:


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@76010 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Pavel Feldman.

        Web Inspector: implement script beautifier prototype.
        https://bugs.webkit.org/show_bug.cgi?id=51588

        * WebCore.gypi:
        * WebCore.vcproj/WebCore.vcproj:
        * inspector/front-end/ScriptFormatter.js: Added.
        (WebInspector.ScriptFormatter): Implements JavaScript source beautification and conversion between original and
        formatted line numbers
        (WebInspector.ScriptFormatter.findLineEndings):
        (WebInspector.ScriptFormatter.locationToPosition):
        (WebInspector.ScriptFormatter.positionToLocation):
        (WebInspector.ScriptFormatter.upperBound):
        (WebInspector.ScriptFormatter.prototype.format):
        (WebInspector.ScriptFormatter.prototype.originalLineNumberToFormattedLineNumber):
        (WebInspector.ScriptFormatter.prototype.formattedLineNumberToOriginalLineNumber):
        (WebInspector.ScriptFormatter.prototype.originalPositionToFormattedLineNumber):
        (WebInspector.ScriptFormatter.prototype.formattedLineNumberToOriginalPosition):
        (WebInspector.ScriptFormatter.prototype._convertPosition):
        * inspector/front-end/ScriptsPanel.js:
        (WebInspector.ScriptsPanel):
        (WebInspector.ScriptsPanel.prototype._clearCurrentExecutionLine):
        (WebInspector.ScriptsPanel.prototype._callFrameSelected):
        (WebInspector.ScriptsPanel.prototype._formatScript):
        * inspector/front-end/SourceFrame.js: Convert original line numbers to UI line numbers and vice versa
        * inspector/front-end/WebKit.qrc:
        * inspector/front-end/inspector.html:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@76009 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Adam Barth.

        ews: Add additional watcher to efl ews
        https://bugs.webkit.org/show_bug.cgi?id=52613

        Add gyuyoung.kim@samsung.com to watcher list of efl-ews.

        * Scripts/webkitpy/tool/commands/earlywarningsystem.py:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@76008 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Unreviewed, GTK build fix as follow-up of r76004.

        * GNUmakefile.am: Include RoundedIntRect files in the build

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@76007 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Unreviewed, rolling out r76002.
        http://trac.webkit.org/changeset/76002
        https://bugs.webkit.org/show_bug.cgi?id=49790

        breaks a bunch of fast/forms tests on mac and GTK

        * fast/forms/select-option-background-color-change.html: Removed.
        * fast/forms/select-option-color-change.html: Removed.
        * fast/forms/select-option-visibility-change.html: Removed.
        * platform/mac/fast/forms/select-option-background-color-change-expected.checksum: Removed.
        * platform/mac/fast/forms/select-option-background-color-change-expected.png: Removed.
        * platform/mac/fast/forms/select-option-background-color-change-expected.txt: Removed.
        * platform/mac/fast/forms/select-option-color-change-expected.checksum: Removed.
        * platform/mac/fast/forms/select-option-color-change-expected.png: Removed.
        * platform/mac/fast/forms/select-option-color-change-expected.txt: Removed.
        * platform/mac/fast/forms/select-option-visibility-change-expected.checksum: Removed.
        * platform/mac/fast/forms/select-option-visibility-change-expected.png: Removed.
        * platform/mac/fast/forms/select-option-visibility-change-expected.txt: Removed.
2011-01-18  Philippe Normand  <pnormand@igalia.com>

        Unreviewed, rolling out r76002.
        http://trac.webkit.org/changeset/76002
        https://bugs.webkit.org/show_bug.cgi?id=49790

        breaks a bunch of fast/forms tests on mac and GTK

        * html/HTMLOptionElement.cpp:
        (WebCore::HTMLOptionElement::setRenderStyle):

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@76006 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Unreviewed windows build fix.

        * WebCore.vcproj/WebCore.vcproj:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@76005 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Simon Fraser.

        Refactoring: Extract RoundedIntRect class
        https://bugs.webkit.org/show_bug.cgi?id=51664

        RoundedIntRect is a pair of rect and four IntSize objects which represent
        corner radii of the rectangle. IntSize is grouped into RoundedIntRect::Radii.
        Now RenderStyle::getRoundedBorderFor() and RenderStyle::getRoundedInnerBorderWithBorderWidths()
        return RoundedIntRect and GraphicsContext::addRoundedRectClip(), GraphicsContext::fillRoundedRect()
        and other functions accept RoundedIntRect as an argument.

        No new tests. No behavioral change.

        * Android.mk:
        * CMakeLists.txt:
        * WebCore.gypi:
        * WebCore.pro:
        * WebCore.xcodeproj/project.pbxproj:
        * platform/graphics/GraphicsContext.cpp:
        (WebCore::GraphicsContext::addRoundedRectClip):
        (WebCore::GraphicsContext::clipOutRoundedRect):
        (WebCore::GraphicsContext::fillRoundedRect):
        * platform/graphics/GraphicsContext.h:
        * platform/graphics/RoundedIntRect.cpp: Added.
        (WebCore::RoundedIntRect::Radii::isZero):
        (WebCore::RoundedIntRect::Radii::scale):
        (WebCore::RoundedIntRect::Radii::expand):
        (WebCore::RoundedIntRect::Radii::includeLogicalEdges):
        (WebCore::RoundedIntRect::Radii::excludeLogicalEdges):
        (WebCore::RoundedIntRect::RoundedIntRect):
        (WebCore::RoundedIntRect::includeLogicalEdges):
        (WebCore::RoundedIntRect::excludeLogicalEdges):
        * platform/graphics/RoundedIntRect.h: Added.
        (WebCore::RoundedIntRect::Radii::Radii):
        (WebCore::RoundedIntRect::Radii::setTopLeft):
        (WebCore::RoundedIntRect::Radii::setTopRight):
        (WebCore::RoundedIntRect::Radii::setBottomLeft):
        (WebCore::RoundedIntRect::Radii::setBottomRight):
        (WebCore::RoundedIntRect::Radii::topLeft):
        (WebCore::RoundedIntRect::Radii::topRight):
        (WebCore::RoundedIntRect::Radii::bottomLeft):
        (WebCore::RoundedIntRect::Radii::bottomRight):
        (WebCore::RoundedIntRect::Radii::expand):
        (WebCore::RoundedIntRect::Radii::shrink):
        (WebCore::RoundedIntRect::location):
        (WebCore::RoundedIntRect::size):
        (WebCore::RoundedIntRect::x):
        (WebCore::RoundedIntRect::y):
        (WebCore::RoundedIntRect::width):
        (WebCore::RoundedIntRect::height):
        (WebCore::RoundedIntRect::right):
        (WebCore::RoundedIntRect::bottom):
        (WebCore::RoundedIntRect::rect):
        (WebCore::RoundedIntRect::radii):
        (WebCore::RoundedIntRect::isRounded):
        (WebCore::RoundedIntRect::isEmpty):
        (WebCore::RoundedIntRect::setRect):
        (WebCore::RoundedIntRect::setRadii):
        (WebCore::RoundedIntRect::move):
        (WebCore::RoundedIntRect::inflate):
        (WebCore::RoundedIntRect::inflateWithRadii):
        (WebCore::RoundedIntRect::expandRadii):
        (WebCore::RoundedIntRect::shrinkRadii):
        * rendering/RenderBox.cpp:
        (WebCore::RenderBox::pushContentsClip):
        * rendering/RenderBoxModelObject.cpp:
        (WebCore::RenderBoxModelObject::paintFillLayerExtended):
        (WebCore::RenderBoxModelObject::paintBorder):
        (WebCore::RenderBoxModelObject::clipBorderSidePolygon):
        (WebCore::RenderBoxModelObject::paintBoxShadow):
        * rendering/RenderBoxModelObject.h:
        * rendering/RenderObject.cpp:
        (WebCore::RenderObject::drawBoxSideFromPath):
        * rendering/RenderReplaced.cpp:
        (WebCore::RenderReplaced::paint):
        * rendering/RenderThemeChromiumWin.cpp:
        (WebCore::RenderThemeChromiumWin::paintTextFieldInternal):
        * rendering/RenderThemeMac.mm:
        (WebCore::RenderThemeMac::paintMenuListButtonGradients):
        (WebCore::RenderThemeMac::paintSliderTrack):
        * rendering/RenderThemeSafari.cpp:
        (WebCore::RenderThemeSafari::paintMenuListButtonGradients):
        (WebCore::RenderThemeSafari::paintSliderTrack):
        * rendering/RenderThemeWinCE.cpp:
        (WebCore::RenderThemeWinCE::paintSearchFieldCancelButton):
        * rendering/RenderWidget.cpp:
        (WebCore::RenderWidget::paint):
        * rendering/style/BorderData.h:
        * rendering/style/RenderStyle.cpp:
        (WebCore::calcRadiiFor):
        (WebCore::calcConstraintScaleFor):
        (WebCore::RenderStyle::getRoundedBorderFor):
        (WebCore::RenderStyle::getRoundedInnerBorderWithBorderWidths):
        * rendering/style/RenderStyle.h:
        (WebCore::InheritedFlags::setBorderRadius):

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@76004 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[Qt] Fix LoadHTMLStringItem::invoke() after r75966
https://bugs.webkit.org/show_bug.cgi?id=52614

* platform/qt/Skipped: http/tests/navigation/go-back-to-error-page.html added until fix.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@76003 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Kent Tamura.

        Color changes to option elements in a select multiple aren't drawn immediately
        https://bugs.webkit.org/show_bug.cgi?id=49790

        Redirect style changes on <option> element to the owner <select> element.

        Tests: fast/forms/select-option-background-color-change.html
               fast/forms/select-option-color-change.html
               fast/forms/select-option-visibility-change.html

        * html/HTMLOptionElement.cpp:
        (WebCore::HTMLOptionElement::setRenderStyle):

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@76002 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Unreviewed, rolling out r75995.
        http://trac.webkit.org/changeset/75995
        https://bugs.webkit.org/show_bug.cgi?id=51914

        Causes assertion failures.

        * dom/Element.cpp:
        (WebCore::Element::shadowRoot):
        (WebCore::Element::setShadowRoot):
        (WebCore::Element::removeShadowRoot):
        * dom/ElementRareData.h:
        (WebCore::ElementRareData::ElementRareData):

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@76001 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Unreviewed, rolling out r75999.
        http://trac.webkit.org/changeset/75999

        * dom/Element.cpp:
        (WebCore::Element::removeShadowRoot):

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@76000 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Unreviewed crash fix, which introduced at r75995

        * dom/Element.cpp:
        (WebCore::Element::removeShadowRoot):

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75999 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by David Levin.

        Rebaseline media/video-zoom-controls for chromium.
        https://bugs.webkit.org/show_bug.cgi?id=52415

        * platform/chromium-linux/media/video-zoom-controls-expected.checksum: Added.
        * platform/chromium-linux/media/video-zoom-controls-expected.png: Added.
        * platform/chromium-mac/media/video-zoom-controls-expected.checksum: Added.
        * platform/chromium-mac/media/video-zoom-controls-expected.png: Added.
        * platform/chromium-mac/media/video-zoom-controls-expected.txt: Added.
        * platform/chromium-win/media/video-zoom-controls-expected.checksum: Added.
        * platform/chromium-win/media/video-zoom-controls-expected.png: Added.
        * platform/chromium-win/media/video-zoom-controls-expected.txt: Added.
        * platform/chromium/test_expectations.txt:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75996 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Dimitri Glazkov.

        ElementRareData::m_shadowRoot should not be RefPtr.
        https://bugs.webkit.org/show_bug.cgi?id=51914

        This change makes m_shadowRoot a raw pointer, as Node::m_parent is.
        This change is safe because shadow host pointer is stored as TreeShared::m_parent
        which prevents deletion unless set to null.

        No test, no behavioral change.

        * dom/Element.cpp:
        (WebCore::Element::shadowRoot):
        (WebCore::Element::setShadowRoot):
        (WebCore::Element::removeShadowRoot):
        * dom/ElementRareData.h:
        (WebCore::ElementRareData::ElementRareData):

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75995 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Dan Bernstein.

        Some WebKit2 layout tests crash due to focus being on a frame in the page cache
        https://bugs.webkit.org/show_bug.cgi?id=52607

        No new tests. Existing tests crash in WebKit2 without this fix.
        
        I suspect there are other consequences besides the WebKit2 crash, but I
        was unable to identify any for certain.

        * history/CachedFrame.cpp:
        (WebCore::CachedFrame::CachedFrame): If a frame still has focus when its page
        goes in the page cache, reset focus to the main frame.


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75994 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Added TIMEOUT to  fast/canvas/canvas-fillPath-pattern-shadow.html

* platform/chromium-gpu/test_expectations.txt:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75993 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Csaba Osztrogonác.

        REGRESSION(r75709): Return value of fscanf() shouldn't be ignored.
        https://bugs.webkit.org/show_bug.cgi?id=52585

        gcc 4.4.4+ has warn_unused_value attribute on fscanf, so we should check
        the return value to get around the gcc warning

        * jit/ExecutableAllocatorFixedVMPool.cpp:
        (JSC::maybeModifyVMPoolSize):

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75992 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Oliver Hunt.

        [regexfuzz] Crash running regex with lookahead
        https://bugs.webkit.org/show_bug.cgi?id=52548

        Eliminated agressive chaining of backtracks.  This code was overwriting
        already valid backtrack information.

        * yarr/YarrJIT.cpp:
        (JSC::Yarr::YarrGenerator::ParenthesesTail::processBacktracks):
2011-01-17  Michael Saboff  <msaboff@apple.com>

        Reviewed by Oliver Hunt.

        [regexfuzz] Crash running regex with lookahead
        https://bugs.webkit.org/show_bug.cgi?id=52548

        New tests from regex fuzzer.

        * fast/regex/parentheses-expected.txt:
        * fast/regex/script-tests/parentheses.js:


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75991 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Same as previous patch, now for -gpu file.

* platform/chromium-gpu/test_expectations.txt:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75990 268f45cc-cd09-0410-ab3c-d52691b4dbfc

fast/canvas/canvas-fillPath-pattern-shadow.html fails on Win and Linux.

* platform/chromium/test_expectations.txt:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75989 268f45cc-cd09-0410-ab3c-d52691b4dbfc

createRenderer/destroyRenderer are now hidden behind setHostWindow.
WKCACFLayerRendererClient::animationsStarted has been removed, as it
was never called. (The work it was supposed to do was already being
accomplished by WKCACFLayerRenderer::render telling each layer that
animations are starting.)

Fixes <http://webkit.org/b/52587> WKCACFLayerRenderer is hard to use

Reviewed by Chris Marrin.

Source/WebCore:

* platform/graphics/win/MediaPlayerPrivateFullscreenWindow.cpp:
(WebCore::MediaPlayerPrivateFullscreenWindow::createWindow): Moved the
call to setHostWindow here from the WM_CREATE handler. The WM_CREATE
handler was causing the Direct3D device to be created, but MSDN says
you shouldn't create a device while handling WM_CREATE. Removed
no-longer-needed call to createRenderer (setHostWindow does this now)
and the never-needed call to setNeedsDisplay (we never draw into the
root layer; this was just creating an unnecessary backing store the
size of the screen!).
(WebCore::MediaPlayerPrivateFullscreenWindow::wndProc): Moved WM_CREATE
code, as described above. Removed call to destroyRenderer when handling
WM_DESTROY; setHostWindow does this now. Fixed up our WM_PAINT handler
to do a synchronous paint and to clear our dirty region, while I was in
here.

* platform/graphics/win/WKCACFLayerRenderer.cpp:
(WebCore::WKCACFLayerRenderer::~WKCACFLayerRenderer): Changed to call
setHostWindow instead of destroyRenderer; the former calls the latter
if needed.
(WebCore::WKCACFLayerRenderer::setHostWindow): Moved here from the
header file. Destroys our old renderer (i.e., IDirect3DDevice9) if
we're losing our window, or creates a renderer if we're gaining a
window.
(WebCore::WKCACFLayerRenderer::createRenderer): Updated for WKSI function rename.
(WebCore::WKCACFLayerRenderer::destroyRenderer): Changed to clear the
D3D device from our context before releasing the device.

* platform/graphics/win/WKCACFLayerRenderer.h: Removed
WKCACFLayerRendererClient::animationsStarted. Removed setNeedsDisplay.
Make createRenderer, destroyRenderer, and renderSoon private.

WebKit/win:

Update for WKCACFLayerRenderer changes

* WebView.cpp:
(WebView::WebViewWndProc): Removed call to layerRendererBecameVisible
when handling WM_SHOWWINDOW. All this did was try to create the
renderer, but it had already been created in setAcceleratedCompositing,
so wasn't needed.
(WebView::setAcceleratedCompositing): Removed call to createRenderer;
setHostWindow does this for us now.

* WebView.h: Removed animationsStarted and layerRendererBecameVisible.

WebKitLibraries:

Make it possible to both set and clear a wkCACFContext's D3D device

* win/include/WebKitSystemInterface/WebKitSystemInterface.h:
* win/lib/WebKitSystemInterface.lib:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75988 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Before this patch, we were painting into our normal backing store
HBITMAP, wrapping it in a CGImage, and handing it off to Core
Animation. This had at least two disadvantages:
   1) The bitmap could be modified while Core Animation was using it.
   2) It introduced extra complexity.

When entering accelerated compositing mode, WebView now creates a
GraphicsLayer to draw into. This GraphicsLayer sits between the root
layer (owned by WKCACFLayerRenderer) and the RenderView's layer. When
WebView invalidates, it just calls setNeedsDisplayInRect on its
GraphicsLayer. When WebView paints, it just tells its
WKCACFLayerRenderer to paint, which will call back to WebView to draw
into the GraphicsLayer if it has a dirty region. This is very similar
to the current implementation of LayerBackedDrawingArea in WebKit2.

Fixes <http://webkit.org/b/52582> WebView should paint directly into a
GraphicsLayer when in accelerated compositing mode

Reviewed by Simon Fraser and Chris Marrin.

WebCore:

Remove special-cased support for providing content for the root layer

Clients will just have to provide content through the normal
GraphicsLayer channels now!

* platform/graphics/win/WKCACFLayerRenderer.cpp:
(WebCore::WKCACFLayerRenderer::WKCACFLayerRenderer): Removed
initialization of m_backingStoreDirty, which has itself been removed.
(WebCore::WKCACFLayerRenderer::setNeedsDisplay): Moved code to schedule
a sync from here to syncCompositingStateSoon. We only need to call
renderSoon if we don't call syncCompositingStateSoon; the latter
function calls the former.
(WebCore::WKCACFLayerRenderer::paint): Removed code to handle
m_backingStoreDirty. We don't want to know anything about clients'
backing stores.
(WebCore::WKCACFLayerRenderer::syncCompositingStateSoon): Added. Code
came from setNeedsDisplay.

* platform/graphics/win/WKCACFLayerRenderer.h:
Removed setRootContents[AndDisplay], setBackingStoreDirty, and
m_backingStoreDirty. Made paint() public so that clients can force a
synchronous render (e.g., when handling WM_PAINT).

WebKit/win:

* WebCoreSupport/WebChromeClient.cpp:
(WebChromeClient::attachRootGraphicsLayer):
(WebChromeClient::scheduleCompositingLayerSync):
Updated for WebView changes.

* WebView.cpp:
(WebView::repaint): When using accelerated compositing, just invalidate
our backing layer. This matches the behavior of LayerBackedDrawingArea.
(WebView::deleteBackingStore): Removed accelerated compositing code.
The WKCACFLayerRenderer doesn't need to know about our backing store
anymore; we don't use it at all when in accelerated compositing mode.
(WebView::addToDirtyRegion): When using accelerated compositing, just
invalidate our backing layer.
(WebView::scrollBackingStore): Added an assertion that this isn't
called in accelerated compositing mode.
(WebView::sizeChanged): Update our backing layer's size, too, and
invalidate it.
(WebView::updateBackingStore): Added an assertion that this isn't
called in accelerated compositing mode.
(WebView::paint): If we're in accelerated compositing mode, sync our
compositing state. If we're *still* in accelerated compositing mode,
just tell our WKCACFLayerRenderer to paint and clear our dirty region.
(The later changes in this function are just un-indenting code that
used to be inside an if.)

(WebView::paintIntoBackingStore):
(WebView::paintIntoWindow):
Added assertions that these aren't called in accelerated compositing
mode.

(WebView::WebViewWndProc): Updated WM_XP_THEMECHANGED handling for
removal of setRootLayerNeedsDisplay.
(WebView::setRootChildLayer): Changed to take a GraphicsLayer. We now
set the layer as a child of our own backing layer.
(WebView::scheduleCompositingLayerSync): Just call through to
WKCACFLayerRenderer.
(WebView::setAcceleratedCompositing): Create our backing layer and set
it as the child of WKCACFLayerRenderer's root layer.
(WebView::notifyAnimationStarted): Added. We never expect this
GraphicsLayerClient override to be called, as we don't use animations
on our backing layer.
(WebView::notifySyncRequired): Added. Just schedule a sync.
(WebView::paintContents): Added. Just clip and paint!

(WebView::showDebugBorders):
(WebView::showRepaintCounter):
Added. These just call through to Settings.

(WebView::syncCompositingState): Changed to first update layout, then
sync state for our backing layer, then sync WebCore's state. This
matches LayerBackedDrawingArea.

* WebView.h: WebView now implements the GraphicsLayerClient interface.
Removed setRootLayerNeedsDisplay; it's been replaced by calling
setNeedsDisplay on our backing layer and calling
syncCompositingStateSoon on WKCACFLayerRenderer as needed. Removed
updateRootLayerContents; that function was used to pass our backing
store to Core Animation, which we no longer do. Added m_backingLayer.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75987 268f45cc-cd09-0410-ab3c-d52691b4dbfc

We aren't really using WKCACFContextFlusher for anything useful at the
moment, but that will probably change in the near future.

I couldn't come up with a way to test this because it isn't possible to
resize a window in DumpRenderTree.

Fixes <http://webkit.org/b/52573> REGRESSION (r75262): Crash beneath
WKCACFContextFlusher::flushAllContexts when resizing window on page
that uses accelerated compositing

Reviewed by Simon Fraser.

* platform/graphics/win/WKCACFLayerRenderer.cpp:
(WebCore::WKCACFLayerRenderer::~WKCACFLayerRenderer): Remove our
context from WKCACFContextFlusher. (This code was erroneously removed
from WKCACFLayerRenderer::destroyRenderer in r75262. This is a more
sensible place for it.)

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75986 268f45cc-cd09-0410-ab3c-d52691b4dbfc

There might be a secret extra sublayer (the tile parent layer) that
PlatformCALayer doesn't know about. When PlatformCALayer would
encounter this, it would try to use the tile parent layer's
PlatformCALayer wrapper, which was null, and then would crash. We now
ask PlatformCALayerWinInternal for the sublayer list, since that class
knows about the tile parent layer and can exclude it from the sublayer
list.

Covered by compositing/tiling/huge-layer-resize.html.

Fixes <http://webkit.org/b/52597> Crash beneath
PlatformCALayer::adoptSublayers when switching out of tiling mode
(null-dereference of a PlatformCALayer)

Reviewed by Darin Adler and Chris Marrin.

LayoutTests:

Make compositing/tiling/huge-layer-resize.html faster, more reliable,
and more crashy (when there's a WebKit bug)

This test was trying to cause a layout/paint to happen by returning to
the event loop for a certain amount of time via setTimeout. But this
didn't always result in a layout/paint (at least on Windows). We now
force the layout/paint explicitly, which also lets us speed up the test
by removing the setTimeouts.

* compositing/tiling/huge-layer-resize.html:
(testOnLoad): Changed to use recordLayerTree, which forces a
layout/paint, instead of hoping that setTimeout will do the trick.
(recordLayerTree): Forces a layout/paint, then dumps the layer tree.

Source/WebCore:

* platform/graphics/ca/win/PlatformCALayerWin.cpp:
(PlatformCALayer::adoptSublayers):
(printLayer):
Changed to use PlatformCALayerWinInternal::getSublayers.

* platform/graphics/ca/win/PlatformCALayerWinInternal.cpp:
(PlatformCALayerWinInternal::getSublayers): Added. Retrieves the list
of PlatformCALayers that represent our sublayers. Significantly, this
code knows about the tile parent layer and can thus exclude it.

* platform/graphics/ca/win/PlatformCALayerWinInternal.h: Added
getSublayers.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75985 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Explicitly set mode during file upload to ensure web server can read the files.


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75984 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Unreviewed.

        Roll chromium DEPS forward to pick up the skia roll to skia r692.
        https://bugs.webkit.org/show_bug.cgi?id=52600

        * DEPS:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75983 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Kent Tamura.

        [Chromium] Fix popup menu re-positioning when the menu is opened upward, above the corresponding form field.
        https://bugs.webkit.org/show_bug.cgi?id=51382
        http://crbug.com/60427

        Calculate correct location of popup window whenever the items in the window change.

        No new tests, because this fix is for Chromium project and hard to test only in WebKit project

        * platform/chromium/PopupMenuChromium.cpp:
        (WebCore::PopupContainer::layoutAndCalculateWidgetRect): New Function to layout and calculate popup widget rect.
        (WebCore::PopupContainer::showPopup): Move widgetRect calculation logic to calculateWidgetRect().
        (WebCore::PopupContainer::refresh): Add parameter focusRect to take the location and the size of focus text input field to calculate correct popup window location.
        * platform/chromium/PopupMenuChromium.h: Append new input parameter for refresh().
2011-01-17  Naoki Takano  <takano.naoki@gmail.com>

        Reviewed by Kent Tamura.

        [Chromium] Fix popup menu re-positioning when the menu is opened upward, above the corresponding form field.
        https://bugs.webkit.org/show_bug.cgi?id=51382
        http://crbug.com/60427

        Calculate correct location of popup window whenever the items in the window change.

        No new tests, because this fix is for Chromium project and hard to test only in WebKit project

       * WebKit/chromium/src/WebViewImpl.cpp:
       (WebKit::WebViewImpl::refreshAutoFillPopup): Change the logic in refreshAutoFilPopup() to check both the location and the size of popup window.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75982 268f45cc-cd09-0410-ab3c-d52691b4dbfc

* Shared/PrintInfo.cpp:
Use #include, not #import.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75981 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        * UIProcess/API/mac/WKView.mm:
        (-[WKView _recursiveDisplayRectIfNeededIgnoringOpacity:isVisibleRect:rectIsVisibleRectForView:topView:]):



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75980 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        https://bugs.webkit.org/show_bug.cgi?id=52595
        Make basic printing work in WebKit2

        * Shared/WebPreferencesStore.h:
        * UIProcess/API/C/WKPreferences.cpp:
        (WKPreferencesSetShouldPrintBackgrounds):
        (WKPreferencesGetShouldPrintBackgrounds):
        * UIProcess/API/C/WKPreferences.h:
        Implement a preference for printing backgrounds.

        * Shared/PrintInfo.cpp: Added.
        (WebKit::PrintInfo::encode):
        (WebKit::PrintInfo::decode):
        * Shared/PrintInfo.h: Added.
        * Shared/mac/PrintInfoMac.mm: Added.
        (WebKit::PrintInfo::PrintInfo):
        A new class for passing page setup information down to the web process.

        * WebKit2.xcodeproj/project.pbxproj:
        * GNUmakefile.am:
        * WebKit2.pro:
        * win/WebKit2.vcproj:
        Added PrintInfo.

        * Scripts/webkit2/messages.py: Added PrintInfo.h to a list of structs.

        * UIProcess/API/mac/WKView.mm:
        (WKViewData): Cache page count and geometry.
        (WebFrameWrapper): Added a wrapper class for adding a WebFrameProxy to an NSDictionary.
        (-[WKView drawRect:]): Added logging.
        (setFrameBeingPrinted): Store a frame into NSPrintOperation's NSPrintInfo, so that all 
        methods invoked for this operation would know what's being printed.
        (frameBeingPrinted): Retrieve this data.
        (-[WKView printOperationWithPrintInfo:forFrame:]): Create a print operation that's appropriate
        for the frame.
        (-[WKView knowsPageRange:]): Ask the web process.
        (-[WKView _recursiveDisplayRectIfNeededIgnoringOpacity:isVisibleRect:rectIsVisibleRectForView:topView:]):
        Take over printing, because I couldn't make drawRect: work correctly with WebKit custom fit
        to page implementation.
        (-[WKView _provideTotalScaleFactorForPrintOperation:]): Help AppKit clip correctly.
        (-[WKView rectForPage:]): Return a pre-calculated rect.

        * UIProcess/WebPageProxy.h: Added m_isInPrintingMode to track whether the web page is in
        printing mode. The UI process controls that.

        * UIProcess/WebPageProxy.cpp:
        (WebKit::WebPageProxy::WebPageProxy): Initialize m_isInPrintingMode.
        (WebKit::WebPageProxy::beginPrinting): Change printing mode in web process if the value changes.
        (WebKit::WebPageProxy::endPrinting): Ditto.
        (WebKit::WebPageProxy::computePagesForPrinting): Ask the web process about output page geometry.
        (WebKit::WebPageProxy::drawRectToPDF): Ask the web process about printed page data.

        * WebProcess/WebPage/WebPage.cpp:
        (WebKit::WebPage::close): Delete m_printContext while there is still a Frame pointer.
        (WebKit::WebPage::updatePreferences): Pass through ShouldPrintBackgrounds.
        (WebKit::WebPage::beginPrinting): Switch into printing mode (or update parameters if we're
        already in printing mode).
        (WebKit::WebPage::endPrinting): Delete m_printContext, implicitly calling end().
        (WebKit::WebPage::computePagesForPrinting): Make sure that we're in printing mode, and
        copmute page rects.
        (WebKit::WebPage::drawRectToPDF): Added a CG specific function to create a PDF document out
        of a page rect. It should really work for onscreen content, but we only need it for printing.

        * WebProcess/WebPage/WebPage.h:

        * WebProcess/WebPage/WebPage.messages.in: Added BeginPrinting, EndPrinting,
        ComputePagesForPrinting and DrawRectToPDF.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75979 268f45cc-cd09-0410-ab3c-d52691b4dbfc

* WebCore.exp.in:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75976 268f45cc-cd09-0410-ab3c-d52691b4dbfc

https://bugs.webkit.org/show_bug.cgi?id=52598

Reviewed by Darin Adler.

Source/WebCore: 

Test: fast/css/font-face-zero-hash-key.html

* css/CSSFontFaceSource.cpp:
(WebCore::CSSFontFaceSource::getFontData): Add 1 to the font size to avoid a 0 hash key.
* css/CSSSegmentedFontFace.cpp:
(WebCore::CSSSegmentedFontFace::getFontData): Ditto.

LayoutTests: 

* fast/css/font-face-zero-hash-key-expected.txt: Added.
* fast/css/font-face-zero-hash-key.html: Added.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75975 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Reviewed by Dan Bernstein.

WebCore:

This fixes two issues:

Add missing DOMDocumentFragmentPrivate.h and
DOMDocumentPrivate.h files to the project.  These files were
never added to the the project although their *Internal.h
counterparts were added in r16548 and r17390.

Add missing DOMDocumentFragmentInternal.h to the Headers
section.  It was moved from Headers to "Copy Generated Headers"
in r31045, but it should have an entry in both sections.

* WebCore.xcodeproj/project.pbxproj:

WebKit/mac:

* MigrateHeaders.make: Copy DOMDocumentFragmentPrivate.h and
DOMDocumentPrivate.h to the PrivateHeaders directory.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75974 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Andreas Kling.

        [Qt] Incorrect shadow alpha with pattern fillStyle
        https://bugs.webkit.org/show_bug.cgi?id=52559

        The shadow color opacity needs to be set on the shadow painter.
        This is related to bug 52556.

        * fast/canvas/canvas-fillPath-pattern-shadow-expected.txt: Added.
        * fast/canvas/canvas-fillPath-pattern-shadow.html: Added.
        * fast/canvas/script-tests/canvas-fillPath-pattern-shadow.js: Added.
2011-01-17  Helder Correia  <helder@sencha.com>

        Reviewed by Andreas Kling.

        [Qt] Incorrect shadow alpha with pattern fillStyle
        https://bugs.webkit.org/show_bug.cgi?id=52559

        The shadow color opacity needs to be set on the shadow painter.
        This is related to bug 52556.

        Test: fast/canvas/canvas-fillPath-pattern-shadow.html

        * platform/graphics/qt/GraphicsContextQt.cpp:
        (WebCore::GraphicsContext::fillPath):

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75973 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Andreas Kling.

        [GTK] Port a crash fix from libsoup upstream to the libsoup cache code
        https://bugs.webkit.org/show_bug.cgi?id=52586

        No new tests. This is just a merge from upstream.

        * platform/network/soup/cache/soup-http-input-stream.c:
        (webkit_soup_http_input_stream_got_chunk): Properly handle unsigned numbers.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75972 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Alexey Proskuryakov.

        Fix some headers with missing or misspelled #ifndef guards
        https://bugs.webkit.org/show_bug.cgi?id=52545

        * wtf/RefPtrHashMap.h:
2011-01-17  Tony Gentilcore  <tonyg@chromium.org>

        Reviewed by Alexey Proskuryakov.

        Fix some headers with missing or misspelled #ifndef guards
        https://bugs.webkit.org/show_bug.cgi?id=52545

        No new tests because no new functionality.

        * ForwardingHeaders/runtime/InitializeThreading.h:
        * editing/SmartReplace.h:
        * loader/CrossOriginAccessControl.h:
        * loader/NetscapePlugInStreamLoader.h:
        * platform/chromium/ClipboardUtilitiesChromium.h:
        * platform/graphics/cairo/DrawErrorUnderline.h:
        * platform/graphics/cairo/GraphicsContextPlatformPrivateCairo.h:
        * platform/graphics/cg/GraphicsContextPlatformPrivateCG.h:
        * platform/graphics/cg/PDFDocumentImage.h:
        * platform/win/WebCoreTextRenderer.h:
2011-01-17  Tony Gentilcore  <tonyg@chromium.org>

        Reviewed by Alexey Proskuryakov.

        Fix some headers with missing or misspelled #ifndef guards
        https://bugs.webkit.org/show_bug.cgi?id=52545

        * WebKitStatistics.h:
2011-01-17  Tony Gentilcore  <tonyg@chromium.org>

        Reviewed by Alexey Proskuryakov.

        Fix some headers with missing or misspelled #ifndef guards
        https://bugs.webkit.org/show_bug.cgi?id=52545

        * Shared/UserMessageCoders.h:
        * WebProcess/InjectedBundle/DOM/InjectedBundleNodeHandle.h:
        * WebProcess/InjectedBundle/InjectedBundleUserMessageCoders.h:
        * WebProcess/WebProcessMain.h:
2011-01-17  Tony Gentilcore  <tonyg@chromium.org>

        Reviewed by Alexey Proskuryakov.

        Fix some headers with missing or misspelled #ifndef guards
        https://bugs.webkit.org/show_bug.cgi?id=52545

        * DumpRenderTree/chromium/TestEventPrinter.h:
        * DumpRenderTree/chromium/WebPreferences.h:

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75971 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        <http://webkit.org/b/52524> fast/dom/Range/range-clone-contents.html fails only on Windows

        Reviewed by Alexey Proskuryakov.

        * platform/win/Skipped: Removed
        fast/dom/Range/range-clone-contents.html.  It appears the
        Windows buildbots need a clean build.

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@75970 268f45cc-cd09-0410-ab3c-d52691b4dbfc