Fix a crash on exit in WinLauncher

        Reviewed by Sam Weinig.

        * WinLauncher/WinLauncher.cpp:
        (_tWinMain): Call shutDownWebKit before exiting.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38954 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Add a Debug_Internal configuration to WinLauncher

        This matches our other projects.

        Reviewed by Sam Weinig.

        * WinLauncher/WinLauncher.vcproj: Added a Debug_Internal
        configuration, which is identical to Debug except that it also
        references debug_internal.vsprops. Also removed some settings that we
        should be picking up from the .vsprops files.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38953 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-03  Dirk Schulze  <krit@webkit.org>

        Reviewed by Nikolas Zimmermann.

        WebCore:

        Make use of the gradient code in GraphicsContext and get rid of most of the
        platform dependent code.

        SVG should use the new Gradient support on GraphicsContext
        https://bugs.webkit.org/show_bug.cgi?id=20543

        * GNUmakefile.am:
        * WebCore.pro:
        * WebCore.vcproj/WebCore.vcproj:
        * WebCore.xcodeproj/project.pbxproj:
        * platform/graphics/FloatSize.h:
        (WebCore::FloatSize::shrunkTo):
        * svg/SVGLinearGradientElement.cpp:
        (WebCore::SVGLinearGradientElement::buildGradient):
        * svg/SVGRadialGradientElement.cpp:
        (WebCore::SVGRadialGradientElement::buildGradient):
        * svg/graphics/SVGPaintServer.h:
        * svg/graphics/SVGPaintServerGradient.cpp:
        (WebCore::SVGPaintServerGradient::SVGPaintServerGradient):
        (WebCore::SVGPaintServerGradient::~SVGPaintServerGradient):
        (WebCore::SVGPaintServerGradient::gradient):
        (WebCore::SVGPaintServerGradient::setGradient):
        (WebCore::findTextRootObject):
        (WebCore::createMaskAndSwapContextForTextGradient):
        (WebCore::clipToTextMask):
        (WebCore::SVGPaintServerGradient::setup):
        (WebCore::SVGPaintServerGradient::renderPath):
        (WebCore::SVGPaintServerGradient::teardown):
        * svg/graphics/SVGPaintServerGradient.h:
        (WebCore::SVGPaintServerGradient::setGradientStops):
        (WebCore::SVGPaintServerGradient::gradientStops):
        * svg/graphics/SVGPaintServerLinearGradient.h:
        * svg/graphics/SVGPaintServerRadialGradient.h:
        * svg/graphics/cairo/SVGPaintServerGradientCairo.cpp: Removed.
        * svg/graphics/cg/SVGPaintServerCg.cpp:
        * svg/graphics/cg/SVGPaintServerGradientCg.cpp: Removed.
        * svg/graphics/qt/SVGPaintServerGradientQt.cpp: Removed.
        * svg/graphics/qt/SVGPaintServerLinearGradientQt.cpp: Removed.
        * svg/graphics/qt/SVGPaintServerRadialGradientQt.cpp: Removed.

        LayoutTests:

        Updated/Correcte the SVG pixel tests.
        We pass two more SVG tests now.

        * platform/mac-leopard/svg/W3C-SVG-1.1/pservers-grad-17-b-expected.checksum:
        * platform/mac-leopard/svg/W3C-SVG-1.1/pservers-grad-17-b-expected.png:
        * platform/mac/svg/batik/text/textEffect3-expected.checksum:
        * platform/mac/svg/batik/text/textEffect3-expected.png:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38952 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-03  Sam Weinig  <sam@webkit.org>

        Reviewed by Mark Rowe.

        Fix formatting.

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

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38951 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-03  Anders Carlsson  <andersca@apple.com>

        Reviewed by Dan Bernstein.

        <rdar://problem/6412234>
        Don't crash if we can't launch the plug-in host.

        * Plugins/Hosted/NetscapePluginHostManager.mm:
        (WebKit::NetscapePluginHostManager::instantiatePlugin):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38950 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Reviewed by Adam Roben.

As Adam Roben noticed, synchronize MessageSource enum with Console.js, and add a comment.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38949 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Reviewed by Alexey Proskuryakov.

Next steps to make LayoutTests work:
Never access the main frame from WMLRefreshElement/WMLCardElement, but
the current documents frame (WML layout tests will run in an <iframe>)

Prepare WMLErrorHandling for the case that no tokenizer is available anymore
-> report errors through Console::addMessage(), so they get logged in the
   expected layout test results.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38948 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-03  Tor Arne Vestbø  <tavestbo@trolltech.com>

        Reviewed by Simon Hausmann.

        Allow passing jsNull and jsUndefined to Qt plugins

        Currently limited to functions with QString and QVariant arguments,
        and properties of these types. Both jsNull and jsUndefined ends up
        as default-constructed QStrings and QVariants, which means you can
        check for isEmpty() and isValid() in the native plugin code.

        Based on patches by Jade Han <jade.han@nokia.com>

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38947 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-03  Trenton Schulz  <trenton.schulz@nokia.com>

        Reviewed by Simon Hausmann.

        Fix the build with Qt for Mac OS X.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38946 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Reviewed by Alexey Proskuryakov.

Fix error handling in WMLSetvarElement. If an invalid variable reference is contained
in the 'name' attribute of <setvar>, a 'WMLErrorInvalidVariab

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38945 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Reviewed by Alexey Proskuryakov.

Preparations to get WML layout tests going.
Make WMLPageState an OwnPtr to Page, and let Page create it on demand.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38944 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by Mark Rowe.

        https://bugs.webkit.org/show_bug.cgi?id=22627
        fast/workers/worker-terminate.html fails randomly

        The problem is that worker termination uses script timeouts, so an InterruptedExecutionError
        is raised, and it sometimes reaches the main thread.

        * dom/WorkerMessagingProxy.cpp:
        (WebCore::WorkerExceptionTask::create):
        (WebCore::WorkerExceptionTask::WorkerExceptionTask):
        (WebCore::WorkerExceptionTask::performTask):
        (WebCore::WorkerMessagingProxy::postWorkerException):
        * dom/WorkerMessagingProxy.h:
        Ignore exceptions that happen in terminated workers.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38943 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Reviewed by Alexey Proskuryakov.
Remove some unncessary includes.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38942 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Rubber-stamped by Cameron Zwarich and Adam Roben.

        Fix a typo: m_executionForbidded.

        * bindings/js/WorkerScriptController.cpp:
        (WebCore::WorkerScriptController::WorkerScriptController):
        (WebCore::WorkerScriptController::evaluate):
        (WebCore::WorkerScriptController::forbidExecution):
        * bindings/js/WorkerScriptController.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38941 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-03  Dean McNamee  <deanm@chromium.org>

        Bug 22623: Uninitialized memory access in cache parsing code
        <https://bugs.webkit.org/show_bug.cgi?id=22623>

        Reviewed by David Kilzer.

        Initialize m_haveParsedCacheControlHeader and m_haveParsedPragmaHeader.

        * platform/network/ResourceResponseBase.h:
        (WebCore::ResourceResponseBase::ResourceResponseBase):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38940 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Removed unchanged file from ChangeLog entry in r38918.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38939 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-03  Simon Hausmann  <hausmann@webkit.org>

        Reviewed by Tor Arne Vestbø.

        Fix the build of WebKit inside Qt by making sure that qmake generators
        that do not produce generated sources (i.e. cpp files) are not
        included in SOURCES.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38938 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-03 Tor Arne Vestbø <tavestbo@trolltech.com>

        Rubber-stamped by Simon Hausmann.

        Fix QWebFrame auto-test by giving it a little more time to load the image

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38937 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-03  Antti Koivisto  <antti@apple.com>

        Probable build fix.

        * WebCoreSources.bkl:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38936 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-03  Antti Koivisto  <antti@apple.com>

        Forgot to commit these.

        * dom/Node.cpp:
        (WebCore::forEachTagSelector):
        (WebCore::SelectorNeedsNamespaceResolutionFunctor::operator()):
        (WebCore::Node::querySelector):
        * dom/SelectorNodeList.cpp:
        (WebCore::createSelectorNodeList):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38935 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-03  Antti Koivisto  <antti@apple.com>

        Reviewed by Dan Bernstein and Mark Rowe.

        https://bugs.webkit.org/show_bug.cgi?id=22379
        Make CSSOM use less memory

        Reduce size of the CSSSelector by 3/8 by moving rarely used fields to a rare data
        struct. Browsing around with some instrumentation showed that ~0.1% of all selectors
        encountered had rare data.

        This also eliminates the CSSNthSelector subclass which will make possible to store
        CSSSelectors in an array instead of a linked list for futher memory savings.

        * css/CSSGrammar.y:
        * css/CSSNthSelector.cpp: Removed.
        * css/CSSNthSelector.h: Removed.
        * css/CSSParser.cpp:
        (WebCore::CSSParser::createFloatingSelector):
        * css/CSSParser.h:
        * css/CSSSelector.cpp:
        (WebCore::CSSSelector::specificity):
        (WebCore::CSSSelector::operator==):
        (WebCore::CSSSelector::selectorText):
        (WebCore::CSSSelector::setTagHistory):
        (WebCore::CSSSelector::attribute):
        (WebCore::CSSSelector::setAttribute):
        (WebCore::CSSSelector::setArgument):
        (WebCore::CSSSelector::setSimpleSelector):
        (WebCore::CSSSelector::parseNth):
        (WebCore::CSSSelector::matchNth):
        (WebCore::CSSSelector::RareData::parseNth):
        (WebCore::CSSSelector::RareData::matchNth):
        * css/CSSSelector.h:
        (WebCore::CSSSelector::CSSSelector):
        (WebCore::CSSSelector::~CSSSelector):
        (WebCore::CSSSelector::tagHistory):
        (WebCore::CSSSelector::hasAttribute):
        (WebCore::CSSSelector::argument):
        (WebCore::CSSSelector::simpleSelector):
        (WebCore::CSSSelector::RareData::RareData):
        (WebCore::CSSSelector::createRareData):
        (WebCore::CSSSelector::):
        * css/CSSStyleSelector.cpp:
        (WebCore::CSSStyleSelector::SelectorChecker::checkSelector):
        (WebCore::CSSStyleSelector::SelectorChecker::checkOneSelector):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38934 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-03  Dmitry Titov  <dimich@chromium.org>

        Reviewed by Alexey Proskuryakov.

        Add test for https://bugs.webkit.org/show_bug.cgi?id=22600,
        verifying that timeout callback is executed in the right scope.

        * fast/dom/Window/timeout-callback-scope-expected.txt: Added.
        * fast/dom/Window/timeout-callback-scope.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38933 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-03  Jian Li  <jianli@chromium.org>

        Reviewed by Alexey Proskuryakov.

        Remove unneeded included file in WorkerThread.cpp.
        https://bugs.webkit.org/show_bug.cgi?id=22613

        * dom/WorkerThread.cpp: No need to include JSWorkerContext.h.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38932 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by Maciej Stachowiak.

        https://bugs.webkit.org/show_bug.cgi?id=22543
        Consolidate ActiveDOMObject page cache interaction

        Test: http/tests/xmlhttprequest/abort-on-leaving-page.html

        * dom/ActiveDOMObject.cpp:
        (WebCore::ActiveDOMObject::canSuspend):
        (WebCore::ActiveDOMObject::suspend):
        (WebCore::ActiveDOMObject::resume):
        * dom/ActiveDOMObject.h:
        * dom/ScriptExecutionContext.cpp:
        (WebCore::ScriptExecutionContext::canSuspendActiveDOMObjects):
        (WebCore::ScriptExecutionContext::suspendActiveDOMObjects):
        (WebCore::ScriptExecutionContext::resumeActiveDOMObjects):
        * dom/ScriptExecutionContext.h:
        Added methods for suspending and resuming active objects.

        * dom/Worker.cpp:
        (WebCore::Worker::canSuspend):
        (WebCore::Worker::stop):
        * dom/Worker.h:
        Workers cannot be suspended yet. This is a change in behavior, as workers used to keep
        running until their owner was destroyed.

        * loader/FrameLoader.cpp:
        (WebCore::FrameLoader::stopLoading): No longer call stopActiveDOMObjects() here, because
        their activity is not necessarily loading.
        (WebCore::FrameLoader::canCachePage): Can only cache if all active objects can be suspended.
        Previously, stopLoading() cancelled outstanding XMLHttpRequests, which made the page
        uncacheable due to no-null main document error.
        (WebCore::FrameLoader::commitProvisionalLoad): Suspend or stop active objects, depending on
        whether the document will be cached.
        (WebCore::FrameLoader::frameDetached): Stop active objects to let them clean up before their
        context is destroyed.

        * xml/XMLHttpRequest.h: added canSuspend().
        * xml/XMLHttpRequest.cpp:
        (WebCore::XMLHttpRequest::canSuspend): Only XHRs that are not loading can be suspended.
        (WebCore::XMLHttpRequest::stop): Call abort() instead of internalAbort() to dispatch events
        for Firefox compatibility. This is a change in behavior that helps test that requests do
        get aborted.
        (WebCore::XMLHttpRequest::contextDestroyed): The request must have been stopped by now, so
        assert that.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38931 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-02  Cameron Zwarich  <zwarich@apple.com>

        Reviewed by Geoff Garen.

        Bug 22504: Crashes during code generation occur due to refing of ignoredResult()
        <https://bugs.webkit.org/show_bug.cgi?id=22504>

        Since ignoredResult() was implemented by casting 1 to a RegisterID*, any
        attempt to ref ignoredResult() results in a crash. This will occur in
        code generation of a function body where a node emits another node with
        the dst that was passed to it, and then refs the returned RegisterID*.

        To fix this problem, make ignoredResult() a member function of
        BytecodeGenerator that simply returns a pointe to a fixed RegisterID
        member of BytecodeGenerator.

        JavaScriptCore:

        * bytecompiler/BytecodeGenerator.h:
        (JSC::BytecodeGenerator::ignoredResult):
        * bytecompiler/RegisterID.h:
        * parser/Nodes.cpp:
        (JSC::NullNode::emitBytecode):
        (JSC::BooleanNode::emitBytecode):
        (JSC::NumberNode::emitBytecode):
        (JSC::StringNode::emitBytecode):
        (JSC::RegExpNode::emitBytecode):
        (JSC::ThisNode::emitBytecode):
        (JSC::ResolveNode::emitBytecode):
        (JSC::ObjectLiteralNode::emitBytecode):
        (JSC::PostfixResolveNode::emitBytecode):
        (JSC::PostfixBracketNode::emitBytecode):
        (JSC::PostfixDotNode::emitBytecode):
        (JSC::DeleteValueNode::emitBytecode):
        (JSC::VoidNode::emitBytecode):
        (JSC::TypeOfResolveNode::emitBytecode):
        (JSC::TypeOfValueNode::emitBytecode):
        (JSC::PrefixResolveNode::emitBytecode):
        (JSC::AssignResolveNode::emitBytecode):
        (JSC::CommaNode::emitBytecode):
        (JSC::ForNode::emitBytecode):
        (JSC::ForInNode::emitBytecode):
        (JSC::ReturnNode::emitBytecode):
        (JSC::ThrowNode::emitBytecode):
        (JSC::FunctionBodyNode::emitBytecode):
        (JSC::FuncDeclNode::emitBytecode):

        LayoutTests:

        * fast/js/ignored-result-ref-crash-expected.txt: Added.
        * fast/js/ignored-result-ref-crash.html: Added.
        * fast/js/resources/ignored-result-ref-crash.js: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38930 268f45cc-cd09-0410-ab3c-d52691b4dbfc

JavaScriptCore:

2008-12-02  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Cameron Zwarich.

        Fixed https://bugs.webkit.org/show_bug.cgi?id=22537
        REGRESSION (r38745): Assertion failure in jsSubstring() at ge.com

        The bug was that index would become greater than length, so our
        "end of input" checks, which all check "index == length", would fail.

        The solution is to check for end of input before incrementing index,
        to ensure that index is always <= length.

        As a side benefit, generateJumpIfEndOfInput can now use je instead of
        jg, which should be slightly faster.

        * wrec/WREC.cpp:
        (JSC::WREC::Generator::compileRegExp):
        * wrec/WRECGenerator.cpp:
        (JSC::WREC::Generator::generateJumpIfEndOfInput):

LayoutTests:

2008-12-02  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Cameron Zwarich.

        Test for https://bugs.webkit.org/show_bug.cgi?id=22537
        REGRESSION (r38745): Assertion failure in jsSubstring() at ge.com

        * fast/regex/alternative-length-miscalculation-expected.txt: Added.
        * fast/regex/alternative-length-miscalculation.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38929 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-02  Gavin Barraclough  <barraclough@apple.com>

        Reviewed by Geoffrey Garen.

        Plant shift right immediate instructions, which are awesome.
        https://bugs.webkit.org/show_bug.cgi?id=22610
        ~5% on the v8-crypto test.

        * jit/JIT.cpp:
        (JSC::JIT::privateCompileMainPass):
        (JSC::JIT::privateCompileSlowCases):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38928 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Bug 22606: Can <th> serves as the AXTitleUIElement for <td>?

<th> will serve as a title UI element when it is present in a table
that has not been exposed as an AXTable

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38927 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-02  Anders Carlsson  <andersca@apple.com>

        Reviewed by Dan Bernstein.

        Start processing messages sent to the client port.

        * Plugins/Hosted/NetscapePluginHostProxy.h:
        * Plugins/Hosted/NetscapePluginHostProxy.mm:
        Add a map from ports to plug-in proxies. Turn the set of instances into a map from
        pluginID to instance proxy.

        (WKPCStatusText):
        Look up the right instance proxy and call status().

        * Plugins/Hosted/NetscapePluginInstanceProxy.h:
        (WebKit::NetscapePluginInstanceProxy::pluginID):
        * Plugins/Hosted/NetscapePluginInstanceProxy.mm:
        (WebKit::NetscapePluginInstanceProxy::status):
        Implement this.

        * Plugins/Hosted/WebKitPluginClient.defs:
        Add the plug-in ID to StatusText.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38926 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-02  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Sam Weinig.

        Cleaned up SegmentedVector by abstracting segment access into helper
        functions.

        SunSpider reports no change.

        * bytecompiler/SegmentedVector.h:
        (JSC::SegmentedVector::SegmentedVector):
        (JSC::SegmentedVector::~SegmentedVector):
        (JSC::SegmentedVector::size):
        (JSC::SegmentedVector::at):
        (JSC::SegmentedVector::operator[]):
        (JSC::SegmentedVector::last):
        (JSC::SegmentedVector::append):
        (JSC::SegmentedVector::removeLast):
        (JSC::SegmentedVector::grow):
        (JSC::SegmentedVector::clear):
        (JSC::SegmentedVector::deleteAllSegments):
        (JSC::SegmentedVector::segmentFor):
        (JSC::SegmentedVector::subscriptFor):
        (JSC::SegmentedVector::ensureSegmentsFor):
        (JSC::SegmentedVector::ensureSegment):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38925 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Reviewed by Kevin Ollivier.

Add HitTest to wxWebView (and wxWebFrame).

https://bugs.webkit.org/show_bug.cgi?id=22459

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38924 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Export new-ish JSCore convenience constructor functions

        Reviewed by Jon Honeycutt.

        * WebKit.vcproj/WebKit.def:
        * WebKit.vcproj/WebKit_debug.def:
        Export JSObjectMakeArray, JSObjectMakeDate, and JSObjectMakeRegExp.
        Remove duplicate export of JSObjectMakeFunction.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38923 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-02  Simon Fraser  <simon.fraser@apple.com>

        Reviewed by Dave Hyatt

        https://bugs.webkit.org/show_bug.cgi?id=22472

        Override absoluteClippedOverflowRect() in RenderReplaced to return a rect
        that is large enough to encompass the selection, so that the repainting of
        selected replaced elements works correctly.

        Test: fast/repaint/selected-replaced.html

        * rendering/RenderReplaced.cpp:
        (WebCore::RenderReplaced::selectionRect):
        (WebCore::RenderReplaced::localSelectionRect):
        (WebCore::RenderReplaced::absoluteClippedOverflowRect):
        * rendering/RenderReplaced.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38922 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Bug 22513: ZOOM: text selection does not send correct zoom bounds

When zoomed, text selection must send the zoom bounds in flipped screen coordinates.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38921 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-02  Anders Carlsson  <andersca@apple.com>

        Reviewed by Dan Bernstein.

        Create a client mach port and pass it to the plug-in host.

        * Plugins/Hosted/NetscapePluginHostManager.h:
        * Plugins/Hosted/NetscapePluginHostManager.mm:
        (WebKit::NetscapePluginHostManager::hostForPackage):
        (WebKit::NetscapePluginHostManager::spawnPluginHost):
        * Plugins/Hosted/NetscapePluginHostProxy.h:
        * Plugins/Hosted/NetscapePluginHostProxy.mm:
        (WebKit::NetscapePluginHostProxy::NetscapePluginHostProxy):
        * Plugins/Hosted/WebKitPluginHost.defs:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38920 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-02  Dean Jackson  <dino@apple.com>

        Reviewed by Sam Weinig.

        Remove support for 'now' from CSS (was
        erroneously added to transition and animation delay)
        https://bugs.webkit.org/show_bug.cgi?id=22571

        * css/CSSParser.cpp:
        (WebCore::CSSParser::parseAnimationDelay):
        * css/CSSStyleSelector.cpp:
        (WebCore::CSSStyleSelector::mapAnimationDelay):
        * css/CSSValueKeywords.in:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38919 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Build fix, no review.

        Fix the Mac and gtk builds:
        Don't use a float to hold a double.
        Remove a now duplicate symbol.

        * WebCore.xcodeproj/project.pbxproj:
        * page/Frame.cpp:
        (WebCore::Frame::selectionLayoutChanged):
        * platform/gtk/TemporaryLinkStubs.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38918 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-02  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Geoffrey Garen. (Patch by Cameron Zwarich <zwarich@apple.com>.)

        Fixed https://bugs.webkit.org/show_bug.cgi?id=22482
        REGRESSION (r37991): Occasionally see "Scene rendered incorrectly"
        message when running the V8 Raytrace benchmark

        Rolled out r37991. It didn't properly save xmm0, which is caller-save,
        before calling helper functions.

        SunSpider and v8 benchmarks show little change -- possibly a .2%
        SunSpider regression, possibly a .2% v8 benchmark speedup.

        * assembler/X86Assembler.h:
        (JSC::X86Assembler::):
        * bytecode/CodeBlock.cpp:
        (JSC::CodeBlock::dump):
        * bytecode/Instruction.h:
        (JSC::Instruction::):
        * bytecompiler/BytecodeGenerator.cpp:
        (JSC::BytecodeGenerator::emitUnaryOp):
        * bytecompiler/BytecodeGenerator.h:
        (JSC::BytecodeGenerator::emitToJSNumber):
        (JSC::BytecodeGenerator::emitTypeOf):
        (JSC::BytecodeGenerator::emitGetPropertyNames):
        * interpreter/Interpreter.cpp:
        (JSC::Interpreter::privateExecute):
        * interpreter/Interpreter.h:
        * jit/JIT.cpp:
        (JSC::JIT::privateCompileMainPass):
        (JSC::JIT::privateCompileSlowCases):
        * jit/JIT.h:
        * parser/Nodes.cpp:
        (JSC::UnaryOpNode::emitBytecode):
        (JSC::BinaryOpNode::emitBytecode):
        (JSC::EqualNode::emitBytecode):
        * parser/ResultType.h:
        (JSC::ResultType::isReusable):
        (JSC::ResultType::mightBeNumber):
        * runtime/JSNumberCell.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38917 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  Gavin Barraclough  <barraclough@apple.com>

        Reviewed by Geoffrey Garen.

        Remove unused (sampling only, and derivable) argument to JIT::emitCTICall.
        https://bugs.webkit.org/show_bug.cgi?id=22587

        * jit/JIT.cpp:
        (JSC::JIT::emitCTICall):
        (JSC::JIT::compileOpCall):
        (JSC::JIT::emitSlowScriptCheck):
        (JSC::JIT::compileBinaryArithOpSlowCase):
        (JSC::JIT::privateCompileMainPass):
        (JSC::JIT::privateCompileSlowCases):
        (JSC::JIT::privateCompile):
        * jit/JIT.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38916 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Bug 22596: Some elements don't report AXBlockQuoteLevel

Make sure all AX objects return a block quote level

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38915 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-02  Anders Carlsson  <andersca@apple.com>

        Reviewed by Sam Weinig.

        Let the plug-in views know if the plug-in host crashes.

        * Plugins/Hosted/NetscapePluginHostManager.h:
        * Plugins/Hosted/NetscapePluginHostManager.mm:
        (WebKit::NetscapePluginHostManager::instantiatePlugin):
        * Plugins/Hosted/NetscapePluginHostProxy.h:
        * Plugins/Hosted/NetscapePluginHostProxy.mm:
        (WebKit::NetscapePluginHostProxy::pluginHostDied):
        (WebKit::NetscapePluginHostProxy::addPluginInstance):
        (WebKit::NetscapePluginHostProxy::removePluginInstance):

        * Plugins/Hosted/NetscapePluginInstanceProxy.h:
        (WebKit::NetscapePluginInstanceProxy::create):
        * Plugins/Hosted/NetscapePluginInstanceProxy.mm:
        Keep a pointer to the host proxy.

        (WebKit::NetscapePluginInstanceProxy::NetscapePluginInstanceProxy):
        Add the instance to the host set.

        (WebKit::NetscapePluginInstanceProxy::~NetscapePluginInstanceProxy):
        Remove the instance form the host set.

        (WebKit::NetscapePluginInstanceProxy::pluginHostDied):
        Tell the plug-in view that the plug-in died.

        * Plugins/Hosted/WebHostedNetscapePluginView.h:
        * Plugins/Hosted/WebHostedNetscapePluginView.mm:
        (-[WebHostedNetscapePluginView createPlugin]):
        Pass the plug-in view to the instantiatePlugin.

        (-[WebHostedNetscapePluginView pluginHostDied]):
        Handle the plug-in host crashing.

        (-[WebHostedNetscapePluginView drawRect:]):
        Fill the plug-in view with a nice red shade if the plug-in crashes.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38914 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-02  Dimitri Glazkov  <dglazkov@chromium.org>

        Reviewed by Eric Seidel.

        Fix the inheritance chain for JSFunction.

        * runtime/JSFunction.cpp:
        (JSC::JSFunction::info): Add InternalFunction::info as parent class

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38913 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-02  Michael Moss  <mmoss@chromium.org>

        Reviewed by Eric Seidel.

        - Fix http/tests/navigation/reload-subframe-*frame.html on Windows.
        https://bugs.webkit.org/show_bug.cgi?id=20926

        Change 31264 fixed this on Mac (since moved to WebCore/loader/FrameLoader.cpp), but was never ported to Windows.

        * WebCoreSupport/WebFrameLoaderClient.cpp:
        (WebFrameLoaderClient::loadURLIntoChild):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38912 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-02  Dean McNamee  <deanm@chromium.org>

        Reviewed by Eric Seidel.

        Correctly handle a theme returning a non-blinking interval.  We should
        not set any timers when the interval is 0 (non-blinking), otherwise we
        repeatedly set and destroy a blinking timer, causing a paint and timer
        storm.  This is applicable to GTK and Windows, where a user can set a
        non-blinking caret in their system preferences.

        Renamed caretBlinkFrequency to the more accurate caretBlinkInterval.

        * page/Frame.cpp:
        (WebCore::Frame::selectionLayoutChanged):
        * platform/Theme.h:
        (WebCore::Theme::caretBlinkInterval):
        * platform/gtk/RenderThemeGtk.cpp:
        (WebCore::RenderThemeGtk::caretBlinkInterval):
        * platform/gtk/RenderThemeGtk.h:
        * rendering/RenderTheme.h:
        (WebCore::RenderTheme::caretBlinkInterval):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38911 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by Eric Seidel.

        https://bugs.webkit.org/show_bug.cgi?id=22538

        startsWith uses find which searches through the whole string if no match is found.
        Using reverseFind with an index of 0 has the benefit of only searching for the match
        at the beginning of the string.  This may only be a small benefit in the overall program,
        but it may help in some cases when the string is big.

        No observable change in behavior, so no test.

        * platform/text/StringImpl.h:
        (WebCore::StringImpl::startsWith):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38910 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by Darin Adler.

        Wrap a JSC-only hack in a USE(JSC) block to fix the v8 build.
        https://bugs.webkit.org/show_bug.cgi?id=21951

        * svg/SVGElementInstance.cpp:
        (WebCore::SVGElementInstance::forgetWrapper):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38909 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by Darin Adler.

        Add an ASSERT to try and catch the root cause of:
        https://bugs.webkit.org/show_bug.cgi?id=22168
        http://code.google.com/p/chromium/issues/detail?id=4122

        * editing/TextIterator.cpp:
        (WebCore::TextIterator::emitText):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38908 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by David Kilzer.

        Hack build-webkit --chromium to use Win32 python instead
        of using the default cygwin python.  Scons + CYGWIN tries
        to build with GCC by default, we force MSVC in our file
        but that just results in errors from Scons.  Until we can fix
        this latest round of errors, at least make build-webkit --chromium
        actually "work" under CYGWIN instead of forcing users to use
        a cmd shell directly.

        This is working around a Scons bug:
        http://scons.tigris.org/issues/show_bug.cgi?id=2266

        * Scripts/webkitdirs.pm:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38907 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-02  Brent Fulgham  <bfulgham@gmail.com>

        Reviewed by Adam Roben.

        Remove some CG-specific code from the Windows Cairo build.
        https://bugs.webkit.org/show_bug.cgi?id=22586

        No new test cases since this update should create no user-visible
        changes, and should be fully covered by the existing regression
        tests.

        * WebCore.vcproj/WebCore.vcproj: Remove FontDatabase from
        * platform/graphics/cairo/GraphicsContextCairo.cpp:
        (WebCore::focusRingColor): Add stub method
        * platform/win/TemporaryLinkStubs.cpp:
        (WebCore::populateFontDatabase): Add stub method.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38906 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  Anders Carlsson  <andersca@apple.com>

        Reviewed by Dan Bernstein.

        Recover if the plug-in host dies and we try to instantiate another plugin before we get the port death notification

        * Plugins/Hosted/NetscapePluginHostManager.h:
        * Plugins/Hosted/NetscapePluginHostManager.mm:
        (WebKit::NetscapePluginHostManager::pluginHostDied):
        Remove the plug-in host from the map.

        (WebKit::NetscapePluginHostManager::instantiatePlugin):
        NetscapePluginHostProxy is no longer refcounted.

        * Plugins/Hosted/NetscapePluginHostProxy.h:
        This is no longer refcounted. Add a set of plug-in instances (unused for now).

        * Plugins/Hosted/NetscapePluginHostProxy.mm:
        (WebKit::NetscapePluginHostProxy::NetscapePluginHostProxy):
        Create a death notification port.

        (WebKit::NetscapePluginHostProxy::pluginHostDied):
        Tell the manager that we're gone and delete ourselves.

        (WebKit::NetscapePluginHostProxy::deathPortCallback):
        New CFMachPort callback.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38905 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-02  Simon Hausmann  <hausmann@webkit.org>

        Rubber-stamped by Tor Arne Vestbø.

        Initial API documentation for QWebDatabase, QWebSecurityOrigin and
        QWebFrame::securityOrigin().

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38904 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01 Laszlo Gombos <laszlo.1.gombos@nokia.com>

        Reviewed by Tor Arne Vestbø.

        MinGW compilation fix for PluginPackageWin.cpp

        https://bugs.webkit.org/show_bug.cgi?id=22585

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38903 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Build fix for Windows projects with NOMINMAX defined globally

        * platform/win/COMPtr.h: Don't redefine NOMINMAX if it's already
        defined.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38902 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Top-Level:
    2008-12-02  Simon Hausmann  <hausmann@webkit.org>

            Reviewed by Tor Arne Vestbø.

            Removed JavaScriptCore from SUBDIRS as it's not built separately
            anymore.

WebCore:
    2008-12-02  Simon Hausmann  <hausmann@webkit.org>

            Reviewed by Tor Arne Vestbø.

            Build JavaScriptCore into libQtWebKit.so through a direct build
            instead of a static library. When linking a static library into
            a shared library qmake generates an incorrect .prl file, which
            causes all sorts of different build problems when linking against
            QtWebKit. Fixing this in qmake requires bigger changes that are
            currently not possible, so we need to work around this limitation
            for now. The advantages of the separate build did not outweight the
            build problems it caused.

JavaScriptCore:
    2008-12-02  Simon Hausmann  <hausmann@webkit.org>

            Reviewed by Tor Arne Vestbø.

            Fix ability to include JavaScriptCore.pri from other .pro files.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38901 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by Alexey Proskuryakov.

        https://bugs.webkit.org/show_bug.cgi?id=22588
        Move securityOrigin() from Document and WorkerContext into ScriptExecutionContext.

        No observable change in behavior, so no test.

        * dom/Document.cpp:
        (WebCore::Document::open):
        (WebCore::Document::domain):
        (WebCore::Document::setDomain):
        (WebCore::Document::initSecurityContext):
        (WebCore::Document::setSecurityOrigin):
        * dom/Document.h:
        * dom/ScriptExecutionContext.cpp:
        (WebCore::ScriptExecutionContext::setSecurityOrigin):
        * dom/ScriptExecutionContext.h:
        (WebCore::ScriptExecutionContext::securityOrigin):
        * dom/WorkerContext.cpp:
        (WebCore::WorkerContext::WorkerContext):
        * dom/WorkerContext.h:
        These changes are for the move of securityOrigin().

        * xml/XMLHttpRequest.cpp:
        (WebCore::XMLHttpRequest::createRequest):
        (WebCore::XMLHttpRequest::makeSimpleCrossSiteAccessRequest):
        (WebCore::XMLHttpRequest::makeCrossSiteAccessRequestWithPreflight):
        (WebCore::XMLHttpRequest::handleAsynchronousPreflightResult):
        (WebCore::XMLHttpRequest::setRequestHeader):
        (WebCore::XMLHttpRequest::getAllResponseHeaders):
        (WebCore::XMLHttpRequest::getResponseHeader):
        (WebCore::XMLHttpRequest::processSyncLoadResults):
        (WebCore::XMLHttpRequest::willSendRequest):
        (WebCore::XMLHttpRequest::accessControlCheck):
        (WebCore::XMLHttpRequest::didReceiveResponsePreflight):
        Change XMLHttpRequest to use ScriptExecutionContext instead of Document to access securityOrigin().

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38900 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-02  André Pönitz  <apoenitz@trolltech.com>

        Reviewed by Simon Hausmann.

        Disable the creation of debug information for the Qt build when done
        inside Qt. With 670 mb it was slowing down gdb start significantly
        for third-party applications that usually don't need it.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38899 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WebCore:

2008-12-01  Beth Dakin  <bdakin@apple.com>

        Reviewed by Dan Bernstein.

        Fix for https://bugs.webkit.org/show_bug.cgi?id=13736 REGRESSION
        (r19811): Using the down arrow in a textarea gets "stuck" at the
        end of a wrapped line
        And corresponding: <rdar://problem/5347931>

        The basic problem here is that Position::getInlineBoxAndOffset()
        failed to look beyond a single renderer. This patch looks for a
        better match beyond the first renderer when the affinity is
        downstream and we failed to find a "perfect" match.
        * dom/Position.cpp:
        (WebCore::isNonTextLeafChild):
        (WebCore::searchAheadForBetterMatch):
        (WebCore::Position::getInlineBoxAndOffset):

        This is a fix I made based on code inspection. It looks like the
        old code here and skipped over the parent as a possible match.
        * rendering/RenderObject.cpp:
        (WebCore::RenderObject::nextInPreOrderAfterChildren):

LayoutTests:

2008-12-01  Beth Dakin  <bdakin@apple.com>

        Reviewed by Dan Bernstein.

        Tests for https://bugs.webkit.org/show_bug.cgi?id=13736 REGRESSION
        (r19811): Using the down arrow in a textarea gets "stuck" at the
        end of a wrapped line
        And corresponding: <rdar://problem/5347931>

        * editing/selection/wrapped-line-caret-1.html: Added.
        * editing/selection/wrapped-line-caret-2.html: Added.
        * platform/mac/editing/selection/wrapped-line-caret-1-
        expected.checksum: Added.
        * platform/mac/editing/selection/wrapped-line-caret-1-expected.png:
        Added.
        * platform/mac/editing/selection/wrapped-line-caret-1-expected.txt:
        Added.
        * platform/mac/editing/selection/wrapped-line-caret-2-
        expected.checksum: Added.
        * platform/mac/editing/selection/wrapped-line-caret-2-expected.png:
        Added.
        * platform/mac/editing/selection/wrapped-line-caret-2-expected.txt:
        Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38898 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  Brent Fulgham  <bfulgham@gmail.com>

        Reviewed by Adam Roben.

        Add WML related files to Visual Studio projects.
        https://bugs.webkit.org/show_bug.cgi?id=22561

        * WebCore.vcproj/WebCore.vcproj:
          1. Add files from the wml directory to the set of windows files.
          2. Extend include paths with new wml directory.
          3. Add new autogenerated WML files to DerivedSources.
          4. Alphabetize preprocesor includes (holdover from earlier debugging).

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38897 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Build fix.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38896 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Revise node/selection image fix.
        Moved updateLayout call so selection rect is fetched after the layout.

        Reviewed by Adam Roben.

        * page/win/FrameCGWin.cpp:
        (WebCore::imageFromRect):
        (WebCore::imageFromSelection):
        (WebCore::Frame::nodeImage):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38895 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WebCore:

2008-12-01  Steve Falkenburg  <sfalken@apple.com>

        Support needed to implement renderedImage for Windows.
        https://bugs.webkit.org/show_bug.cgi?25648

        Reviewed by Adam Roben.

        * page/Frame.h:
        * page/win/FrameCGWin.cpp:
        (WebCore::imageFromRect):
        (WebCore::imageFromSelection):
        (WebCore::Frame::nodeImage):
        * page/win/FrameCairoWin.cpp:
        (WebCore::imageFromNode):

WebKit/win:

2008-12-01  Steve Falkenburg  <sfalken@apple.com>

        Implement renderedImage for Windows.
        https://bugs.webkit.org/show_bug.cgi?25648

        Reviewed by Adam Roben.

        * DOMCoreClasses.cpp:
        (DOMElement::renderedImage):
        * DOMCoreClasses.h:
        * Interfaces/DOMPrivate.idl:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38894 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  Simon Fraser  <simon.fraser@apple.com>

        Reviewed by Dan Bernstein

        https://bugs.webkit.org/show_bug.cgi?id=22581

        Fix the painting of the caps lock indicator for transformed text inputs,
        by replacing a call to absoluteContentBox() with code that computes the
        painting rect for the input contents.

        * rendering/RenderTextControl.cpp:
        (WebCore::RenderTextControl::paint):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38893 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Not reviewed. Forgot to include updated layout tests in last commit.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38892 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Cameron Zwarich, with help from Gavin Barraclough.

        Fixed https://bugs.webkit.org/show_bug.cgi?id=22583.

        Refactored regular expression parsing to parse sequences of characters
        as a single unit, in preparation for optimizing sequences of characters.

        SunSpider reports no change.

        * JavaScriptCore.xcodeproj/project.pbxproj:
        * wrec/Escapes.h: Added. Set of classes for representing an escaped
        token in a pattern.

        * wrec/Quantifier.h:
        (JSC::WREC::Quantifier::Quantifier): Simplified this constructor slightly,
        to match the new Escape constructor.

        * wrec/WRECGenerator.cpp:
        (JSC::WREC::Generator::generatePatternCharacterSequence):
        * wrec/WRECGenerator.h: Added an interface for generating a sequence
        of pattern characters at a time. It doesn't do anything special yet.

        * wrec/WRECParser.cpp:
        (JSC::WREC::Parser::consumeGreedyQuantifier):
        (JSC::WREC::Parser::consumeQuantifier): Renamed "parse" to "consume" in
        these functions, to match "consumeEscape."

        (JSC::WREC::Parser::parsePatternCharacterSequence): New function for
        iteratively aggregating a sequence of characters in a pattern.

        (JSC::WREC::Parser::parseCharacterClassQuantifier):
        (JSC::WREC::Parser::parseBackreferenceQuantifier): Renamed "parse" to
        "consume" in these functions, to match "consumeEscape."

        (JSC::WREC::Parser::parseCharacterClass): Refactored to use the common
        escape processing code in consumeEscape.

        (JSC::WREC::Parser::parseEscape): Refactored to use the common
        escape processing code in consumeEscape.

        (JSC::WREC::Parser::consumeEscape): Factored escaped token processing
        into a common function, since we were doing this in a few places.

        (JSC::WREC::Parser::parseTerm): Refactored to use the common
        escape processing code in consumeEscape.

        * wrec/WRECParser.h:
        (JSC::WREC::Parser::consumeOctal): Refactored to use a helper function
        for reading a digit.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38891 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Reviewed by George Staikos.

Add two new helper functions to WMLElement, parseValueSubstitutingVariableReferences/parseValueForbiddingVariableReferences.

Convert all elements to parse their attribute values using these new helper functions. This simplifies the
actual attribute parsing code in all WMLElement derived classes, as most WML attribute values either report
an error if there's a variable reference used in the attribute value, or if it's invalid.

Split the WMLErrorInvalidVariableReference error which covered both 'invalid syntax' / 'wrong location'
in two seperated error codes: WMLErrorInvalidVariableReference / WMLErrorInvalidVariableReferenceLocation.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38890 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Not reviewed, add reviewers full name.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38889 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Reviewed by George.
Fix switching active cards within a document. No way to test, until <go> support is implemented.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38888 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  Cameron Zwarich  <zwarich@apple.com>

        Reviewed by Oliver Hunt.

        Bug 20340: SegmentedVector segment allocations can lead to unsafe use of temporary registers
        <https://bugs.webkit.org/show_bug.cgi?id=20340>

        SegmentedVector currently frees segments and reallocates them when used
        as a stack. This can lead to unsafe use of pointers into freed segments.

        In order to fix this problem, SegmentedVector will be changed to only
        grow and never shrink. Also, rename the reserveCapacity() member
        function to grow() to match the actual usage in BytecodeGenerator, where
        this function is used to allocate a group of registers at once, rather
        than merely saving space for them.

        * bytecompiler/BytecodeGenerator.cpp:
        (JSC::BytecodeGenerator::BytecodeGenerator): Use grow() instead of
        reserveCapacity().
        * bytecompiler/SegmentedVector.h:
        (JSC::SegmentedVector::SegmentedVector):
        (JSC::SegmentedVector::last):
        (JSC::SegmentedVector::append):
        (JSC::SegmentedVector::removeLast):
        (JSC::SegmentedVector::grow): Renamed from reserveCapacity().
        (JSC::SegmentedVector::clear):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38887 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  Pierre-Olivier Latour  <pol@apple.com>

        Reviewed by Mark Rowe.

        Updated the reference images for the <video> layout tests as the ones checked-in previously
        do not pass on the pixel bots.

        https://bugs.webkit.org/show_bug.cgi?id=21911

        * platform/mac/media/video-aspect-ratio-expected.checksum:
        * platform/mac/media/video-aspect-ratio-expected.png:
        * platform/mac/media/video-controls-rendering-expected.checksum:
        * platform/mac/media/video-controls-rendering-expected.png:
        * platform/mac/media/video-display-toggle-expected.checksum:
        * platform/mac/media/video-display-toggle-expected.png:
        * platform/mac/media/video-layer-crash-expected.checksum:
        * platform/mac/media/video-layer-crash-expected.png:
        * platform/mac/media/video-transformed-expected.checksum:
        * platform/mac/media/video-transformed-expected.png:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38886 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Fix logic error in LegacyWebArchive::create() from r38884.

        * loader/archive/cf/LegacyWebArchive.cpp:
        (WebCore::LegacyWebArchive::create): Only continue if the
        subresource is actually added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38885 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Bug 22466: REGRESSION (35867): Many resources missing when saving webarchive of webkit.org

        <https://bugs.webkit.org/show_bug.cgi?id=22466>
        <rdar://problem/6403593>

        Reviewed by Brady Eidson.

WebCore:

        Test: http/tests/webarchive/test-preload-resources.html

        * loader/archive/cf/LegacyWebArchive.cpp:
        (WebCore::LegacyWebArchive::create): Check the WebCore cache for
        resources if DocumentLoader::subresource() doesn't return them.
        Note that the DocumentLoader::subresource() method returned
        preloaded resources before r35867, but this caused a regression in
        Mail.

WebKitTools:

        * DumpRenderTree/mac/DumpRenderTree.mm:
        (normalizeHTTPResponseHeaderFields): Added.  Normalizes Date, Etag,
        Keep-Alive, Last-Modified and Server header fields to prevent false
        positive test failures.
        (convertWebResourceResponseToDictionary): Call
        normalizeHTTPResponseHeaderFields() to noramlize HTTP response
        header fields.

LayoutTests:

        * http/tests/webarchive/resources/test-preload-resources.css: Added.
        * http/tests/webarchive/test-preload-resources-expected.webarchive: Added.
        * http/tests/webarchive/test-preload-resources.html: Added.
        * platform/qt/Skipped: Added http/tests/webarchive and grouped the
        skipped list of other webarchive tests together.
        * platform/win/Skipped: Added http/tests/webarchive.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38884 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  Anders Carlsson  <andersca@apple.com>

        Reviewed by Dan Bernstein.

        Move instantiatePlugin to NetscapePluginHostManager.

        * Plugins/Hosted/NetscapePluginHostManager.h:
        * Plugins/Hosted/NetscapePluginHostManager.mm:
        (WebKit::NetscapePluginHostManager::instantiatePlugin):
        * Plugins/Hosted/NetscapePluginHostProxy.h:
        * Plugins/Hosted/NetscapePluginHostProxy.mm:
        * Plugins/Hosted/WebHostedNetscapePluginView.mm:
        (-[WebHostedNetscapePluginView createPlugin]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38883 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  Anders Carlsson  <andersca@apple.com>

        Reviewed by Dan Bernstein.

        Pass the plug-in host port directly to the instance proxy.

        * Plugins/Hosted/NetscapePluginHostProxy.mm:
        (WebKit::NetscapePluginHostProxy::instantiatePlugin):
        * Plugins/Hosted/NetscapePluginInstanceProxy.h:
        (WebKit::NetscapePluginInstanceProxy::create):
        * Plugins/Hosted/NetscapePluginInstanceProxy.mm:
        (WebKit::NetscapePluginInstanceProxy::NetscapePluginInstanceProxy):
        (WebKit::NetscapePluginInstanceProxy::resize):
        (WebKit::NetscapePluginInstanceProxy::destroy):
        (WebKit::NetscapePluginInstanceProxy::focusChanged):
        (WebKit::NetscapePluginInstanceProxy::windowFocusChanged):
        (WebKit::NetscapePluginInstanceProxy::windowFrameChanged):
        (WebKit::NetscapePluginInstanceProxy::startTimers):
        (WebKit::NetscapePluginInstanceProxy::mouseEvent):
        (WebKit::NetscapePluginInstanceProxy::stopTimers):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38882 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  Julien Chaffraix  <jchaffraix@webkit.org>

        Reviewed by Eric Seidel.

        Bug 22564: Make HTML elements' constructors take a QualifiedName
        https://bugs.webkit.org/show_bug.cgi?id=22564

        - Modified the remaining HTML elements' constructors to take a QualifiedName.

        - Added an assertion that the QualifiedName given corresponds to the element constructed.

        * bindings/js/JSAudioConstructor.cpp:
        (WebCore::constructAudio):
        * html/HTMLAudioElement.cpp:
        (WebCore::HTMLAudioElement::HTMLAudioElement):
        * html/HTMLAudioElement.h:
        * html/HTMLBRElement.cpp:
        (WebCore::HTMLBRElement::HTMLBRElement):
        * html/HTMLBRElement.h:
        * html/HTMLElement.cpp:
        (WebCore::HTMLElement::setInnerText):
        * html/HTMLElementFactory.cpp:
        (WebCore::brConstructor):
        (WebCore::quoteConstructor):
        (WebCore::marqueeConstructor):
        (WebCore::audioConstructor):
        (WebCore::videoConstructor):
        (WebCore::sourceConstructor):
        * html/HTMLMarqueeElement.cpp:
        (WebCore::HTMLMarqueeElement::HTMLMarqueeElement):
        * html/HTMLMarqueeElement.h:
        * html/HTMLQuoteElement.cpp:
        (WebCore::HTMLQuoteElement::HTMLQuoteElement):
        * html/HTMLQuoteElement.h:
        * html/HTMLSourceElement.cpp:
        (WebCore::HTMLSourceElement::HTMLSourceElement):
        * html/HTMLSourceElement.h:
        * html/HTMLVideoElement.cpp:
        (WebCore::HTMLVideoElement::HTMLVideoElement):
        * html/HTMLVideoElement.h:
        * rendering/RenderTextControl.cpp:
        (WebCore::RenderTextControl::updateFromElement):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38881 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  Pamela Greene  <pam@chromium.org>

        Reviewed by Darin Adler.

        https://bugs.webkit.org/show_bug.cgi?id=22464
        Add a test for a potential crash when performing a same-origin
        check on a subframe reloaded with about:blank.

        * fast/js/same-origin-subframe-about-blank-expected.txt: Added.
        * fast/js/same-origin-subframe-about-blank.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38880 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  Pierre-Olivier Latour  <pol@apple.com>

        Reviewed by Dan Bernstein.

        Fixed <video> layout tests so that DRT is notified they are done only when the video frames
        are effectively rendered in the page. This requires using a timeout since there is no <video>
        event that fires when the first frame has been drawn (the closest even fires when the element
        is ready to render and will eventually draw at some unknown point in the future).
        Thanks to this change, these tests now pass in --pixel mode as well.

        https://bugs.webkit.org/show_bug.cgi?id=21911

        * media/video-aspect-ratio.html:
        * media/video-controls-rendering.html:
        * media/video-display-toggle.html:
        * media/video-layer-crash.html:
        * media/video-paint-test.js:
        (init):
        * media/video-transformed.html:
        * platform/mac/media/video-aspect-ratio-expected.checksum:
        * platform/mac/media/video-aspect-ratio-expected.png:
        * platform/mac/media/video-controls-rendering-expected.checksum:
        * platform/mac/media/video-controls-rendering-expected.png:
        * platform/mac/media/video-display-toggle-expected.checksum:
        * platform/mac/media/video-display-toggle-expected.png:
        * platform/mac/media/video-layer-crash-expected.checksum:
        * platform/mac/media/video-layer-crash-expected.png:
        * platform/mac/media/video-transformed-expected.checksum:
        * platform/mac/media/video-transformed-expected.png:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38879 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  Julien Chaffraix  <jchaffraix@webkit.org>

        Reviewed by Antti Koivisto.

        Bug 22441: Bridge the gap between the generated ElementFactory and HTMLElementFactory
        https://bugs.webkit.org/show_bug.cgi?id=22441

        Add an assertion to the modified HTML elements' constructor
        to check that the QualifiedName parameter is the one that
        matches the Element's.

        * html/HTMLBaseElement.cpp:
        (WebCore::HTMLBaseElement::HTMLBaseElement):
        * html/HTMLBlockquoteElement.cpp:
        (WebCore::HTMLBlockquoteElement::HTMLBlockquoteElement):
        * html/HTMLBodyElement.cpp:
        (WebCore::HTMLBodyElement::HTMLBodyElement):
        * html/HTMLButtonElement.cpp:
        (WebCore::HTMLButtonElement::HTMLButtonElement):
        * html/HTMLDListElement.cpp:
        (WebCore::HTMLDListElement::HTMLDListElement):
        * html/HTMLDirectoryElement.cpp:
        (WebCore::HTMLDirectoryElement::HTMLDirectoryElement):
        * html/HTMLDivElement.cpp:
        (WebCore::HTMLDivElement::HTMLDivElement):
        * html/HTMLFieldSetElement.cpp:
        (WebCore::HTMLFieldSetElement::HTMLFieldSetElement):
        * html/HTMLFormElement.cpp:
        (WebCore::HTMLFormElement::HTMLFormElement):
        * html/HTMLFrameElement.cpp:
        (WebCore::HTMLFrameElement::HTMLFrameElement):
        * html/HTMLFrameSetElement.cpp:
        (WebCore::HTMLFrameSetElement::HTMLFrameSetElement):
        * html/HTMLHeadElement.cpp:
        (WebCore::HTMLHeadElement::HTMLHeadElement):
        * html/HTMLHtmlElement.cpp:
        (WebCore::HTMLHtmlElement::HTMLHtmlElement):
        * html/HTMLIFrameElement.cpp:
        (WebCore::HTMLIFrameElement::HTMLIFrameElement):
        * html/HTMLInputElement.cpp:
        (WebCore::HTMLInputElement::HTMLInputElement):
        * html/HTMLIsIndexElement.cpp:
        (WebCore::HTMLIsIndexElement::HTMLIsIndexElement):
        * html/HTMLKeygenElement.cpp:
        (WebCore::HTMLKeygenElement::HTMLKeygenElement):
        * html/HTMLLIElement.cpp:
        (WebCore::HTMLLIElement::HTMLLIElement):
        * html/HTMLLabelElement.cpp:
        (WebCore::HTMLLabelElement::HTMLLabelElement):
        * html/HTMLLegendElement.cpp:
        (WebCore::HTMLLegendElement::HTMLLegendElement):
        * html/HTMLLinkElement.cpp:
        (WebCore::HTMLLinkElement::HTMLLinkElement):
        * html/HTMLMenuElement.cpp:
        (WebCore::HTMLMenuElement::HTMLMenuElement):
        * html/HTMLMetaElement.cpp:
        (WebCore::HTMLMetaElement::HTMLMetaElement):
        * html/HTMLOListElement.cpp:
        (WebCore::HTMLOListElement::HTMLOListElement):
        * html/HTMLOptGroupElement.cpp:
        (WebCore::HTMLOptGroupElement::HTMLOptGroupElement):
        * html/HTMLOptionElement.cpp:
        (WebCore::HTMLOptionElement::HTMLOptionElement):
        * html/HTMLSelectElement.cpp:
        (WebCore::HTMLSelectElement::HTMLSelectElement):
        * html/HTMLStyleElement.cpp:
        (WebCore::HTMLStyleElement::HTMLStyleElement):
        * html/HTMLTextAreaElement.cpp:
        (WebCore::HTMLTextAreaElement::HTMLTextAreaElement):
        * html/HTMLTitleElement.cpp:
        (WebCore::HTMLTitleElement::HTMLTitleElement):
        * html/HTMLUListElement.cpp:
        (WebCore::HTMLUListElement::HTMLUListElement):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38878 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Disable WREC for x86_64 since memory allocated by the system allocator is not marked executable,
which causes 64-bit debug builds to crash.  Once we have a dedicated allocator for executable
memory we can turn this back on.

Rubber-stamped by Anders Carlsson.

* wtf/Platform.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38877 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  Chris Marrin  <cmarrin@apple.com>

        Reviewed by Darin Adler.

        https://bugs.webkit.org/show_bug.cgi?id=22046

        Fixed another case of crashing because the animation object is getting deleted when
        it's in the middle of a callback. I actually moved where I am retaining the pointer
        up out of the lower level AnimationBase code and into the timer callback that calls
        them. So now the pointer is valid throughout the entire sequence of callback code.

        The testcase for https://bugs.webkit.org/show_bug.cgi?id=22052 also exhibits a
        crash which this patch fixes.

        * page/animation/AnimationBase.cpp:
        (WebCore::AnimationBase::updateStateMachine):
        (WebCore::AnimationBase::animationTimerCallbackFired):
        * page/animation/CompositeAnimation.cpp:
        (WebCore::CompositeAnimationPrivate::setAnimationStartTime):
        (WebCore::CompositeAnimationPrivate::setTransitionStartTime):
        (WebCore::CompositeAnimationPrivate::styleAvailable):
        * page/animation/ImplicitAnimation.cpp:
        (WebCore::ImplicitAnimation::sendTransitionEvent):
        * page/animation/KeyframeAnimation.cpp:
        (WebCore::KeyframeAnimation::sendAnimationEvent):
        (WebCore::KeyframeAnimation::resumeOverriddenAnimations):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38876 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  Antti Koivisto  <antti@apple.com>

        Reviewed by Maciej Stachowiak.

        Restore inline buffer after vector is shrunk back below its inline capacity.

        * wtf/Vector.h:
        (WTF::):
        (WTF::VectorBuffer::restoreInlineBufferIfNeeded):
        (WTF::::shrinkCapacity):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38875 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  Anders Carlsson  <andersca@apple.com>

        Try to fix the Tiger build.

        * Plugins/WebNetscapePluginView.mm:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38874 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by Adam Roben.

        NULL-check documentFrameView() and topDocumentFrameView() it's possible
        for either of these methods to return NULL, and this was seen to cause
        crashes in Chromium.
        https://bugs.webkit.org/show_bug.cgi?id=22572

        * AccessibleBase.cpp:
        (AccessibleBase::get_accParent):
        (AccessibleBase::accLocation):
        (AccessibleBase::accHitTest):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38873 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  Anders Carlsson  <andersca@apple.com>

        Rename _layer to _pluginLayer where I forgot to do so.

        * Plugins/Hosted/WebHostedNetscapePluginView.mm:
        (-[WebHostedNetscapePluginView destroyPlugin]):
        * Plugins/WebNetscapePluginView.h:
        * Plugins/WebNetscapePluginView.mm:
        (-[WebNetscapePluginView createPlugin]):
        (-[WebNetscapePluginView setLayer:]):
        (-[WebNetscapePluginView destroyPlugin]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38872 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Versioning.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38871 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Fix ChangeLog

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38869 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  Anders Carlsson  <andersca@apple.com>

        Reviewed by Adam Roben.

        Make sure to re-insert layers as needed so they won't be lost when the layer backed
        view is removed from the view hierarchy.

        * Plugins/Hosted/WebHostedNetscapePluginView.h:
        Add _layer ivar.

        * Plugins/Hosted/WebHostedNetscapePluginView.mm:
        (-[WebHostedNetscapePluginView createPlugin]):
        Don't add the layer here.

        (-[WebHostedNetscapePluginView setLayer:]):
        Instead, add it here.

        (-[WebHostedNetscapePluginView destroyPlugin]):
        Set _layer to 0.

        * Plugins/WebNetscapePluginView.mm:
        (-[WebNetscapePluginView createPlugin]):
        Don't add the layer here.

        (-[WebNetscapePluginView setLayer:]):
        Do it here.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38868 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  Pamela Greene  <pam@chromium.org>

        Reviewed by Darin Adler.

        https://bugs.webkit.org/show_bug.cgi?id=22463
        Add a test for accessing navigator.cookieEnabled after navigation.

        * fast/dom/navigator-cookieEnabled-no-crash-expected.txt: Added.
        * fast/dom/navigator-cookieEnabled-no-crash.html: Added.
        * fast/dom/resources/test-successful-destination.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38867 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-11-30  Antti Koivisto  <antti@apple.com>

        Reviewed by Mark Rowe.

        Try to return free pages in the current thread cache too.

        * wtf/FastMalloc.cpp:
        (WTF::TCMallocStats::releaseFastMallocFreeMemory):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38866 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  Pamela Greene  <pam@chromium.org>

        Reviewed by Sam Weinig.

        https://bugs.webkit.org/show_bug.cgi?id=22365
        Add a test to verify behavior of DOM node constructors.

        * fast/dom/dom-constructors-expected.txt: Added.
        * fast/dom/dom-constructors.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38865 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Bug 22559: Report builds left to test in bisect-builds (like git-bisect)

        <https://bugs.webkit.org/show_bug.cgi?id=22559>

        Reviewed by Adam Roben.

        * Scripts/bisect-builds: Updated status message to report the
        maximum number of builds left to test after the current one.
        (max): Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38864 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  Pamela Greene  <pam@chromium.org>

        Reviewed by Darin Adler.

        https://bugs.webkit.org/show_bug.cgi?id=22358
        Add a test for crash on "Copy" after showing a tooltip.

        * editing/pasteboard/copy-crash-expected.txt: Added.
        * editing/pasteboard/copy-crash.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38863 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Rubber-stamped by Mark Rowe.

        https://bugs.webkit.org/show_bug.cgi?id=22531
        Stack overflow on PowerPC on fast/workers/use-machine-stack.html

        Disabling the test until the problem can be investigated.

        * fast/workers/use-machine-stack.html: Removed.
        * fast/workers/use-machine-stack.html-disabled: Copied from fast/workers/use-machine-stack.html.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38862 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  Xan Lopez  <xan@gnome.org>

        Reviewed by Holger Freyther.

        http://bugs.webkit.org/show_bug.cgi?id=22553
        Remove unneeded GObject casts.

        Remove unneeded casts to GObject in functions that take a gpointer
        argument.

        * WebCoreSupport/ChromeClientGtk.cpp:
        (WebKit::ChromeClient::setWindowRect):
        (WebKit::ChromeClient::createWindow):
        (WebKit::ChromeClient::setToolbarsVisible):
        (WebKit::ChromeClient::toolbarsVisible):
        (WebKit::ChromeClient::setStatusbarVisible):
        (WebKit::ChromeClient::statusbarVisible):
        (WebKit::ChromeClient::setScrollbarsVisible):
        (WebKit::ChromeClient::scrollbarsVisible):
        (WebKit::ChromeClient::setMenubarVisible):
        (WebKit::ChromeClient::menubarVisible):
        (WebKit::ChromeClient::setToolTip):
        * WebCoreSupport/InspectorClientGtk.cpp:
        (WebKit::InspectorClient::createPage):
        * webkit/webkitwebframe.cpp:
        * webkit/webkitwebview.cpp:
        * webkit/webkitwebwindowfeatures.cpp:
        (_WebKitWebWindowFeaturesPrivate::webkit_web_window_features_new_from_core_features):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38861 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Make it clearer in ChangeLog entry that the last change only affected debug builds.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38860 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  David Levin  <levin@chromium.org>

        Reviewed by Alexey Proskuryakov.

        https://bugs.webkit.org/show_bug.cgi?id=22567
        Make HashTable work as expected with respect to threads. Specifically, it has class-level
        thread safety and constant methods work on constant objects without synchronization.

        No observable change in behavior, so no test.

        * wtf/HashTable.cpp:
        (WTF::hashTableStatsMutex):
        (WTF::HashTableStats::~HashTableStats):
        (WTF::HashTableStats::recordCollisionAtCount):
        Guarded variable access with a mutex.

        * wtf/HashTable.h:
        (WTF::::lookup):
        (WTF::::lookupForWriting):
        (WTF::::fullLookupForWriting):
        (WTF::::add):
        (WTF::::reinsert):
        (WTF::::remove):
        (WTF::::rehash):
        Changed increments of static variables to use atomicIncrement.

        (WTF::::invalidateIterators):
        (WTF::addIterator):
        (WTF::removeIterator):
        Guarded mutable access with a mutex.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38859 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  Tor Arne Vestbø  <tavestbo@trolltech.com>

        Reviewed by Simon Hausmann.

        [Qt/Mac] Initialize NPAPI plugins before getting their entry points

        The old behavior (calling NP_GetEntryPoints before NP_Initialize) was
        copied from Windows, but caused Silverlight on Mac to crash when loaded.

        Apparently the call order of NP_Initialize and NP_GetEntryPoints is
        reversed on Mac. See https://bugzilla.mozilla.org/show_bug.cgi?id=344425

Reported-by: Peter Johnson <peter@zattoo.com>
git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38858 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-11-29  Gavin Barraclough  <barraclough@apple.com>

        Reviewed by Camron Zwarich.

        Enable WREC on PLATFORM(X86_64).  This change predominantly requires changes to the
        WREC::Generator::generateEnter method to support the x86-64 ABI, and addition of
        support for a limited number of quadword operations in the X86Assembler.

        This patch will cause the JS heap to be allocated with RWX permissions on 64-bit Mac
        platforms.  This is a regression with respect to previous 64-bit behaviour, but is no
        more permissive than on 32-bit builds.  This issue should be addressed at some point.
        (This is tracked by bug #21783.)

        https://bugs.webkit.org/show_bug.cgi?id=22554
        Greater than 4x speedup on regexp-dna, on x86-64.

        * assembler/MacroAssembler.h:
        (JSC::MacroAssembler::addPtr):
        (JSC::MacroAssembler::loadPtr):
        (JSC::MacroAssembler::storePtr):
        (JSC::MacroAssembler::pop):
        (JSC::MacroAssembler::push):
        (JSC::MacroAssembler::move):
        * assembler/X86Assembler.h:
        (JSC::X86Assembler::):
        (JSC::X86Assembler::movq_rr):
        (JSC::X86Assembler::addl_i8m):
        (JSC::X86Assembler::addl_i32r):
        (JSC::X86Assembler::addq_i8r):
        (JSC::X86Assembler::addq_i32r):
        (JSC::X86Assembler::movq_mr):
        (JSC::X86Assembler::movq_rm):
        * wrec/WREC.h:
        * wrec/WRECGenerator.cpp:
        (JSC::WREC::Generator::generateEnter):
        (JSC::WREC::Generator::generateReturnSuccess):
        (JSC::WREC::Generator::generateReturnFailure):
        * wtf/Platform.h:
        * wtf/TCSystemAlloc.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38857 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-12-01  Cameron Zwarich  <zwarich@apple.com>

        Reviewed by Sam Weinig.

        Preliminary work for bug 20340: SegmentedVector segment allocations can lead to unsafe use of temporary registers
        <https://bugs.webkit.org/show_bug.cgi?id=20340>

        SegmentedVector currently frees segments and reallocates them when used
        as a stack. This can lead to unsafe use of pointers into freed segments.

        In order to fix this problem, SegmentedVector will be changed to only
        grow and never shrink, with the sole exception of clearing all of its
        data, a capability that is required by Lexer. This patch changes the
        public interface to only allow for these capabilities.

        * bytecompiler/BytecodeGenerator.cpp:
        (JSC::BytecodeGenerator::BytecodeGenerator): Use reserveCapacity()
        instead of resize() for m_globals and m_parameters.
        * bytecompiler/SegmentedVector.h:
        (JSC::SegmentedVector::resize): Removed.
        (JSC::SegmentedVector::reserveCapacity): Added.
        (JSC::SegmentedVector::clear): Added.
        (JSC::SegmentedVector::shrink): Removed.
        (JSC::SegmentedVector::grow): Removed.
        * parser/Lexer.cpp:
        (JSC::Lexer::clear): Use clear() instead of resize(0).

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38856 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Bug 22510: Crash at WebCore::AccessibilityRenderObject::activeDescendant() on Google Reader with ARIA
https://bugs.webkit.org/show_bug.cgi?id=22510

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38855 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2008-11-30  Sam Weinig  <sam@webkit.org>

        Reviewed by Mark Rowe.

        Renames jumps to m_jumps in JumpList.

        * assembler/MacroAssembler.h:
        (JSC::MacroAssembler::JumpList::link):
        (JSC::MacroAssembler::JumpList::linkTo):
        (JSC::MacroAssembler::JumpList::append):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@38854 268f45cc-cd09-0410-ab3c-d52691b4dbfc