2009-10-12  Jan Michael Alonzo  <jmalonzo@webkit.org>

        Rubberstamped by Eric Seidel.

        [Gtk] Fix icu CFLAG for Darwin
        https://bugs.webkit.org/show_bug.cgi?id=29517

        Don't escape the srcdir variable. Also use $host instead of the
        os_foo variables.

        * autotools/webkit.m4:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49436 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-12  Jan Michael Alonzo  <jmalonzo@webkit.org>

        Reviewed by Gustavo Noronha.

        [Gtk] Use the SQLite3 headers from WebKitLibraries if sqlite3 is undetected
        https://bugs.webkit.org/show_bug.cgi?id=29518

        * configure.ac:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49435 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-11  Daniel Bates  <dbates@webkit.org>

        Reviewed by Adam Barth.

        https://bugs.webkit.org/show_bug.cgi?id=30242

        Fixes an issue where JavaScript URLs that are URL-encoded twice can
        bypass the XSSAuditor.

        JavaScript URLs that are completed by method Document::completeURL have added
        URL-encoded characters such that a direct comparison with the URL-decoded
        outgoing HTTP parameters is not sufficient. Instead, the URL-decoded outgoing
        HTTP parameters must be URL-decoded before comparison.

        Tests: http/tests/security/xssAuditor/iframe-javascript-url-twice-url-encode.html
               http/tests/security/xssAuditor/iframe-javascript-url-twice-url-encode2.html
               http/tests/security/xssAuditor/iframe-javascript-url-twice-url-encode3.html

        * bindings/ScriptControllerBase.cpp:
        (WebCore::ScriptController::executeIfJavaScriptURL): Modified to pass XSSAuditor
        the URL-decoded source code for the JavaScript URL.
        * page/XSSAuditor.cpp:
        (WebCore::isIllegalURICharacter): Minor syntactical change to the comment.
        (WebCore::XSSAuditor::CachingURLCanonicalizer::canonicalizeURL): Added
        parameter decodeURLEscapeSequencesTwice.
        (WebCore::XSSAuditor::canEvaluateJavaScriptURL):
        (WebCore::XSSAuditor::decodeURL): Ditto.
        (WebCore::XSSAuditor::findInRequest): Ditto.
        * page/XSSAuditor.h:
        (WebCore::XSSAuditor::CachingURLCanonicalizer::CachingURLCanonicalizer): Ditto.
2009-10-11  Daniel Bates  <dbates@webkit.org>

        Reviewed by Adam Barth.

        https://bugs.webkit.org/show_bug.cgi?id=30242

        Tests that JavaScript URLs that are twice URL encoded do not bypass the XSSAuditor.

        * http/tests/security/xssAuditor/iframe-javascript-url-twice-url-encode-expected.txt: Added.
        * http/tests/security/xssAuditor/iframe-javascript-url-twice-url-encode.html: Added.
        * http/tests/security/xssAuditor/iframe-javascript-url-twice-url-encode2-expected.txt: Added.
        * http/tests/security/xssAuditor/iframe-javascript-url-twice-url-encode2.html: Added.
        * http/tests/security/xssAuditor/iframe-javascript-url-twice-url-encode3-expected.txt: Added.
        * http/tests/security/xssAuditor/iframe-javascript-url-twice-url-encode3.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49434 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WebKit/mac: <rdar://problem/7294267> REGRESSION (r48492): Gmail voice and video chat
doesn't work
https://bugs.webkit.org/show_bug.cgi?id=30295

Reviewed by Darin Adler.

Test: plugins/setProperty.html

* Plugins/Hosted/ProxyInstance.mm:
(WebKit::ProxyInstance::setFieldValue): Fixed a typo. The function
should return early if m_instanceProxy is null.

LayoutTests: Test for <rdar://problem/7294267> REGRESSION (r48492): Gmail voice and
video chat doesn't work
https://bugs.webkit.org/show_bug.cgi?id=30295

Reviewed by Darin Adler.

* plugins/setProperty-expected.txt: Added.
* plugins/setProperty.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49433 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-11  Dominic Cooney  <dominicc@google.com>

        Reviewed by Adam Barth.

        Fixes a build break in the V8 bindings caused by revision 49420.
        https://bugs.webkit.org/show_bug.cgi?id=30294

        Revision 49420 introduces an IDL-defined CanvasActiveInfo
        type. This causes a build break in the V8 bindings where
        CanvasGraphicsContext3D naively refers to CanvasActiveInfo as a
        non-ref type.

        * WebCore.gypi: Include CanvasActiveInfo.idl in build.
        * bindings/scripts/CodeGeneratorV8.pm: CanvasActiveInfo is a ref
          ptr type.
        * bindings/v8/DOMObjectsInclude.h: Include generated
          CanvasActiveInfo files.
        * bindings/v8/DerivedSourcesAllInOne.cpp:
        * bindings/v8/V8Index.cpp:
        * bindings/v8/V8Index.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49432 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-11  Collin Jackson  <collinj@webkit.org>

        Reviewed by Adam Barth.

        Log mixed content warnings to console

        https://bugs.webkit.org/show_bug.cgi?id=30290

        * loader/FrameLoader.cpp:
        (WebCore::FrameLoader::checkIfDisplayInsecureContent):
        (WebCore::FrameLoader::checkIfRunInsecureContent):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49431 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-11  Adam Barth  <abarth@webkit.org>

        Unreviewed.  Add Collin to committers.py.

        * Scripts/modules/committers.py:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49430 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-11  Søren Gjesse  <sgjesse@chromium.org>

        Reviewed by Adam Barth.

        [V8] Add a context scope in JS listener destructor. The destructor code uses the cotext. See http://crbug.com/24200.
        https://bugs.webkit.org/show_bug.cgi?id=30250

        * bindings/v8/V8AbstractEventListener.cpp:
        (WebCore::V8AbstractEventListener::~V8AbstractEventListener):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49429 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-11  Lyon Chen  <lyon.chen@torchmobile.com>

        Reviewed by Alexey Proskuryakov.

        Correct the position of #endif sentence inside WebCore::Document::finishedParsing().

        https://bugs.webkit.org/show_bug.cgi?id=30244

        This change requires no test case as it fix a compiling error.

        * dom/Document.cpp:
        (WebCore::Document::finishedParsing):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49428 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-11  Dominic Cooney  <dominicc@google.com>

        Reviewed by Adam Barth.

        Sets a 500K stack limit for JavaScript workers in Chromium.
        https://bugs.webkit.org/show_bug.cgi?id=29797

        Runaway recursion in JavaScript workers crashes the Chromium
        worker process on OS X. This is because V8's default stack limit
        is 512K on ia32 or 1M on x64, but the worker process runs workers
        on a thread with the OS X default stack size--512K. Because there
        are already some C++ frames on the stack when V8 establishes its
        512K default stack limit, and V8 doesn't precisely enforce the
        stack limit, runaway recursion in V8 workers overflows the OS
        stack and segfaults, killing the worker process. This is described
        in Chromium bug 21653 <http://crbug.com/21653>.

        This patch sets the V8 stack limit for worker JavaScript in
        Chromium to a more conservative 500K on all platforms. This allows
        some "headroom" for the C++ stack in use when the limit is set up,
        and some "legroom" for the V8 helper functions which in practice
        briefly flout the V8 stack limit.

        Test: LayoutTests/fast/workers/use-machine-stack.html

        * bindings/v8/WorkerContextExecutionProxy.cpp:
        (WebCore::WorkerContextExecutionProxy::initV8IfNeeded):
        * bindings/v8/WorkerContextExecutionProxy.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49427 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Re-enable the JIT.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49426 268f45cc-cd09-0410-ab3c-d52691b4dbfc

wx build fixes, adding bindings to include/source dirs and fixing wxMac 2.9 compilation.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49425 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Allow [Reflect] on SVG elements.
https://bugs.webkit.org/show_bug.cgi?id=28936

Reviewed by Sam Weinig.

Update the JS binding generators to reference SVGNames instead of
HTMLNames, if [Reflect]ing an attribute on an SVG element.  Make
SVGElement::id use [Reflect].

Also make [Reflect] on an attribute with a setter exception work in ObjC
bindings.

WebCore:

Test: svg/dom/id-reflect.html

* bindings/scripts/CodeGenerator.pm: Add a function to determine the
appropriate C++ namespace for attribute name constants.
* bindings/scripts/CodeGeneratorObjC.pm: Generate ExceptionCode handling
code for [Reflect] on an attribute with a setter exception.
* bindings/scripts/CodeGeneratorCOM.pm: Generate "SVGNames" instead of
"HTMLNames" when appropriate.
* bindings/scripts/CodeGeneratorJS.pm: Ditto.
* bindings/scripts/CodeGeneratorV8.pm: Ditto.
* svg/SVGElement.cpp: Remove getter and setter methods for id.
* svg/SVGElement.h: Ditto.
* svg/SVGElement.idl: Add [Reflect] to id.

LayoutTests:

* svg/dom/script-tests/id-reflect.js: Added.
* svg/dom/id-reflect-expected.txt: Added.
* svg/dom/id-reflect.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49424 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Support for String.trim(), String.trimLeft() and String.trimRight() methods
https://bugs.webkit.org/show_bug.cgi?id=26590

Reviewed by Maciej Stachowiak.

Implement trim, trimLeft, and trimRight

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49423 268f45cc-cd09-0410-ab3c-d52691b4dbfc

REGRESSION (r49414): Rebaseline needed for Windows / Gtk Linux
https://bugs.webkit.org/show_bug.cgi?id=30279

Patch by Ryosuke Niwa <rniwa@webkit.org> on 2009-10-10
Reviewed by Eric Seidel.

* platform/gtk/editing/style/5228141-expected.txt: Same rebaseline done for Mac in r49414
* platform/win/editing/execCommand/toggle-compound-styles-expected.txt: two i tags are merged properly

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49422 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Fix paths in xcode.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49421 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Implement getActiveAttrib and getActiveUniform
https://bugs.webkit.org/show_bug.cgi?id=30276

Reviewed by Eric Carlson.

Implements the getActiveAttrib and getActiveUniform APIs.
Rather simple patch, adds CanvasActiveInfo definition and implementation
and adds forwarding to the GraphicsContext3D.

Test: fast/canvas/webgl/getActiveTest.html

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49420 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-10  Pavel Feldman  <pfeldman@chromium.org>

        Reviewed by Timothy Hatcher.

        Web Inspector: Adding new attributes in Element Panel is
        counterintuitive.

        https://bugs.webkit.org/show_bug.cgi?id=30057

        * inspector/front-end/ElementsTreeOutline.js:
        (WebInspector.ElementsTreeOutline.prototype._onmousemove):
        (WebInspector.ElementsTreeElement.prototype.set hovered):
        (WebInspector.ElementsTreeElement.prototype.toggleNewAttributeButton):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49419 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-10  Laszlo Gombos  <laszlo.1.gombos@nokia.com>

        Reviewed by Holger Freyther.

        Build fix if Netscape plugin support is turned off
        https://bugs.webkit.org/show_bug.cgi?id=30275

        In addition to fixing the build break with guards
        remove notImplemented() from PluginViewNone.cpp and
        make them just just be empty functions.

        * plugins/PluginView.cpp:
        (WebCore::PluginView::handleEvent):
        (WebCore::PluginView::PluginView):
        * plugins/PluginView.h:
        * plugins/PluginViewNone.cpp:
        (WebCore::PluginView::setFocus):
        (WebCore::PluginView::show):
        (WebCore::PluginView::hide):
        (WebCore::PluginView::paint):
        (WebCore::PluginView::handleKeyboardEvent):
        (WebCore::PluginView::handleMouseEvent):
        (WebCore::PluginView::setParent):
        (WebCore::PluginView::setNPWindowRect):
        (WebCore::PluginView::handlePostReadFile):
        (WebCore::PluginView::getValue):
        (WebCore::PluginView::getValueStatic):
        (WebCore::PluginView::invalidateRect):
        (WebCore::PluginView::invalidateRegion):
        (WebCore::PluginView::forceRedraw):
        (WebCore::PluginView::platformStart):
        (WebCore::PluginView::platformDestroy):
        (WebCore::PluginView::setParentVisible):
        (WebCore::PluginView::updatePluginWidget):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49418 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-10  Laszlo Gombos  <laszlo.1.gombos@nokia.com>

        Unreviewed.

        [Qt] Build fix for QtWebKit after r49415.

        Add HistoryController.cpp to the build.

        * WebCore.pro:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49417 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-09  Darin Fisher  <darin@chromium.org>

        Reviewed by Eric Seidel.

        [Chromium] KURLGoogle's protocolIs barfs on input containing hyphens
        https://bugs.webkit.org/show_bug.cgi?id=30262

        This is not testable from WebKit since WebKit never uses the member
        function form of protocolIs with input that can be controlled from a
        web page.  It always passes string literals lacking hyphens.

        * platform/KURLGoogle.cpp:
        (WebCore::lowerCaseEqualsASCII): Only assert if toASCIILower would
        transform the input character.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49416 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-10  Adam Barth  <abarth@webkit.org>

        Reviewed by Oliver Hunt.

        Move HistoryController to its own file
        https://bugs.webkit.org/show_bug.cgi?id=30272

        Purely mechanical.

        * GNUmakefile.am:
        * WebCore.gypi:
        * WebCore.vcproj/WebCore.vcproj:
        * WebCore.xcodeproj/project.pbxproj:
        * WebCoreSources.bkl:
        * loader/FrameLoader.cpp:
        * loader/FrameLoader.h:
        * loader/HistoryController.cpp: Added.
        * loader/HistoryController.h: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49415 268f45cc-cd09-0410-ab3c-d52691b4dbfc

ApplyStyleCommand removes presentational tags even when not necessary
https://bugs.webkit.org/show_bug.cgi?id=28091

Patch by Ryosuke Niwa <rniwa@webkit.org> on 2009-10-09
Reviewed by Eric Seidel.

WebCore:

This patch updates implicitlyStyledElementShouldBeRemovedWhenApplyingStyle to return true
(triggering removal of tag) only if the style implicitly added by the element is not present in
the new style to apply. It also changes surroundNodeRangeWithElement so that it merges the
surrounding element when possible. applyTextDecorationStyle is modified so that it does not add
style span when the style attribute is empty.

Test: editing/execCommand/toggle-style-3.html

* editing/ApplyStyleCommand.cpp:
(WebCore::ApplyStyleCommand::shouldRemoveTextDecorationTag): Returns true if specified text
decoration is not present in the style to apply
(WebCore::ApplyStyleCommand::implicitlyStyledElementShouldBeRemovedWhenApplyingStyle): No longer
returns true if the tag is used in new style
(WebCore::ApplyStyleCommand::applyTextDecorationStyle): Does not add style-span when redundant
(WebCore::ApplyStyleCommand::surroundNodeRangeWithElement): Merges the newly created element
with the surrounding identical elements
* editing/ApplyStyleCommand.h:

LayoutTests:

This patch adds test to make sure to preserve presentational tag
(u, s, strike, i, em, sup, sub, b, & strong) when the corresponding style is present in
the final computed style. i.e. avoid removing tag and re-applying the style later

* editing/execCommand/19653-2-expected.txt: Two b elements are merged.
* editing/execCommand/createLink-expected.txt: Two anchor elements are merged.
* editing/execCommand/script-tests/toggle-style-3.js: Added.
(testSingleToggle):
* editing/execCommand/strikethroughSelection-expected.txt: Redundant style spans are no longer added.
* editing/execCommand/toggle-style-3-expected.txt: Added.
* editing/execCommand/toggle-style-3.html: Added.
* editing/execCommand/unlink-expected.txt: Two anchor elements are merged.
* platform/mac/editing/style/5228141-expected.txt: Two b elements are merged.
* platform/mac/editing/style/style-boundary-001-expected.txt: Two b elements are merged.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49414 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-09  Adam Barth  <abarth@webkit.org>

        Reviewed by Darin Adler.

        Factor HistoryController out of FrameLoader
        https://bugs.webkit.org/show_bug.cgi?id=30246

        Update API call.

        * WebView.cpp:
        (WebView::loadBackForwardListFromOtherView):
2009-10-09  Adam Barth  <abarth@webkit.org>

        Reviewed by Darin Adler.

        Factor HistoryController out of FrameLoader
        https://bugs.webkit.org/show_bug.cgi?id=30246

        Update API call.

        * WebCoreSupport/WebFrameLoaderClient.mm:
        (WebFrameLoaderClient::restoreViewState):
        * WebView/WebView.mm:
        (-[WebView _loadBackForwardListFromOtherView:]):
2009-10-09  Adam Barth  <abarth@webkit.org>

        Reviewed by Darin Adler.

        Factor HistoryController out of FrameLoader
        https://bugs.webkit.org/show_bug.cgi?id=30246

        HistoryController is in charge of managing the current / previous /
        provisional HistoryItems.  The split isn't perfect, but it's a place to
        start.  I'll move HistoryController into its own file in another patch.

        * WebCore.base.exp:
        * loader/FrameLoader.cpp:
        (WebCore::HistoryController::HistoryController):
        (WebCore::HistoryController::~HistoryController):
        (WebCore::FrameLoader::FrameLoader):
        (WebCore::FrameLoader::closeURL):
        (WebCore::FrameLoader::begin):
        (WebCore::HistoryController::restoreDocumentState):
        (WebCore::HistoryController::setCurrentItem):
        (WebCore::HistoryController::setProvisionalItem):
        (WebCore::FrameLoader::loadURLIntoChildFrame):
        (WebCore::FrameLoader::canCachePageContainingThisFrame):
        (WebCore::FrameLoader::logCanCacheFrameDecision):
        (WebCore::FrameLoader::scrollToAnchor):
        (WebCore::FrameLoader::commitProvisionalLoad):
        (WebCore::FrameLoader::transitionToCommitted):
        (WebCore::FrameLoader::checkLoadCompleteForThisFrame):
        (WebCore::FrameLoader::didFirstLayout):
        (WebCore::HistoryController::updateForFrameLoadCompleted):
        (WebCore::FrameLoader::frameLoadCompleted):
        (WebCore::FrameLoader::detachFromParent):
        (WebCore::FrameLoader::receivedMainResourceError):
        (WebCore::FrameLoader::continueFragmentScrollAfterNavigationPolicy):
        (WebCore::FrameLoader::continueLoadAfterNavigationPolicy):
        (WebCore::HistoryController::addHistoryItemForFragmentScroll):
        (WebCore::FrameLoader::loadProvisionalItemFromCachedPage):
        (WebCore::FrameLoader::shouldTreatURLAsSameAsCurrent):
        (WebCore::HistoryController::createHistoryItem):
        (WebCore::FrameLoader::checkDidPerformFirstNavigation):
        (WebCore::HistoryController::addBackForwardItemClippedAtTarget):
        (WebCore::HistoryController::createHistoryItemTree):
        (WebCore::HistoryController::saveScrollPositionAndViewStateToItem):
        (WebCore::HistoryController::restoreScrollPositionAndViewState):
        (WebCore::HistoryController::invalidateCurrentItemCachedPage):
        (WebCore::HistoryController::saveDocumentState):
        (WebCore::FrameLoader::loadItem):
        (WebCore::HistoryController::urlsMatchItem):
        (WebCore::HistoryController::goToItem):
        (WebCore::HistoryController::recursiveGoToItem):
        (WebCore::HistoryController::childFramesMatchItem):
        (WebCore::HistoryController::updateForStandardLoad):
        (WebCore::HistoryController::updateForClientRedirect):
        (WebCore::HistoryController::updateForBackForwardNavigation):
        (WebCore::HistoryController::updateForReload):
        (WebCore::HistoryController::updateForRedirectWithLockedBackForwardList):
        (WebCore::HistoryController::updateForCommit):
        (WebCore::HistoryController::updateForAnchorScroll):
        (WebCore::HistoryController::saveDocumentAndScrollState):
        (WebCore::HistoryController::setCurrentItemTitle):
        (WebCore::FrameLoader::didChangeTitle):
        * loader/FrameLoader.h:
        (WebCore::HistoryController::current):
        (WebCore::HistoryController::provisional):
        (WebCore::FrameLoader::policyChecker):
        (WebCore::FrameLoader::history):
        (WebCore::FrameLoader::creatingInitialEmptyDocument):
        * page/Page.cpp:
        (WebCore::Page::goToItem):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49413 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Try to fix Windows build by adding needed exports.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49412 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WebKit/mac: <rdar://problem/7290671> Crash after navigating away from a page with subframe containing plug-in

Reviewed by Brady Eidson.

At the time that -stop is called on a WebBaseNetscapePluginView that is contained within a subframe
the WebFrame's data source has been zeroed out.  This was causing -[WebBaseNetscapePluginView webFrame]
to return nil during -stop, preventing the plug-in view from unregistering itself from the PluginHalter.
We fix this by having -webFrame retrieve the frame directly rather than retrieving it from the data source.

* Plugins/WebBaseNetscapePluginView.mm:
(-[WebBaseNetscapePluginView dataSource]): Get the data source from the WebFrame.
(-[WebBaseNetscapePluginView webFrame]): Get the Frame directly from the element's document rather
than taking a windy route through the WebDataSource.

WebKitTools: Enable plug-in halting in DumpRenderTree.

We drop the plug-in halting delay to 1 second and opt in the delegate method to never halt plug-ins.
This is sufficient to ensure that the crash covered by <rdar://problem/7290671> no longer occurs.

* DumpRenderTree/mac/DumpRenderTree.mm:
(resetDefaultsToConsistentValues):
* DumpRenderTree/mac/UIDelegate.mm:
(-[UIDelegate webView:shouldHaltPlugin:]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49411 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Unreviewed - fixing an obvious typo in my previous check-in.

        https://bugs.webkit.org/show_bug.cgi?id=30260
        <rdar://problem/6447115> REGRESSION: Logging out from SAP doesn't work

        Fixes existing tests (which I thought I had run the previous time).

        * platform/network/CredentialStorage.cpp: (WebCore::originsWithCredentials): Return the
        set by reference.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49410 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Eliminated some legacy bytecode weirdness.

Patch by Geoffrey Garen <ggaren@apple.com> on 2009-10-09
Reviewed by Oliver Hunt.

Use vPC[x] subscripting instead of ++vPC to access instruction operands.
This is simpler, and often more efficient.

To support this, and to remove use of hard-coded offsets in bytecode and
JIT code generation and dumping, calculate jump offsets from the beginning
of an instruction, rather than the middle or end.

Also, use OPCODE_LENGTH instead of hard-coded constants for the sizes of
opcodes.

SunSpider reports no change in JIT mode, and a 1.01x speedup in Interpreter
mode.

* bytecode/CodeBlock.cpp:
(JSC::printConditionalJump):
(JSC::CodeBlock::dump):
* bytecompiler/BytecodeGenerator.cpp:
(JSC::BytecodeGenerator::emitJump):
(JSC::BytecodeGenerator::emitJumpIfTrue):
(JSC::BytecodeGenerator::emitJumpIfFalse):
(JSC::BytecodeGenerator::emitJumpIfNotFunctionCall):
(JSC::BytecodeGenerator::emitJumpIfNotFunctionApply):
(JSC::BytecodeGenerator::emitComplexJumpScopes):
(JSC::BytecodeGenerator::emitJumpScopes):
(JSC::BytecodeGenerator::emitNextPropertyName):
(JSC::BytecodeGenerator::emitCatch):
(JSC::BytecodeGenerator::emitJumpSubroutine):
(JSC::prepareJumpTableForImmediateSwitch):
(JSC::prepareJumpTableForCharacterSwitch):
(JSC::prepareJumpTableForStringSwitch):
(JSC::BytecodeGenerator::endSwitch):
* bytecompiler/Label.h:
(JSC::Label::setLocation):
(JSC::Label::bind):
* interpreter/Interpreter.cpp:
(JSC::Interpreter::resolve):
(JSC::Interpreter::resolveSkip):
(JSC::Interpreter::resolveGlobal):
(JSC::Interpreter::resolveBase):
(JSC::Interpreter::resolveBaseAndProperty):
(JSC::Interpreter::createExceptionScope):
(JSC::Interpreter::privateExecute):
* interpreter/Interpreter.h:
* jit/JIT.cpp:
(JSC::JIT::privateCompile):
* jit/JITArithmetic.cpp:
(JSC::JIT::emit_op_jnless):
(JSC::JIT::emitSlow_op_jnless):
(JSC::JIT::emit_op_jnlesseq):
(JSC::JIT::emitSlow_op_jnlesseq):
(JSC::JIT::emitBinaryDoubleOp):
* jit/JITOpcodes.cpp:
(JSC::JIT::emit_op_jmp):
(JSC::JIT::emit_op_loop):
(JSC::JIT::emit_op_loop_if_less):
(JSC::JIT::emitSlow_op_loop_if_less):
(JSC::JIT::emit_op_loop_if_lesseq):
(JSC::JIT::emitSlow_op_loop_if_lesseq):
(JSC::JIT::emit_op_loop_if_true):
(JSC::JIT::emitSlow_op_loop_if_true):
(JSC::JIT::emit_op_jfalse):
(JSC::JIT::emitSlow_op_jfalse):
(JSC::JIT::emit_op_jtrue):
(JSC::JIT::emitSlow_op_jtrue):
(JSC::JIT::emit_op_jeq_null):
(JSC::JIT::emit_op_jneq_null):
(JSC::JIT::emit_op_jneq_ptr):
(JSC::JIT::emit_op_jsr):
(JSC::JIT::emit_op_next_pname):
(JSC::JIT::emit_op_jmp_scopes):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49409 268f45cc-cd09-0410-ab3c-d52691b4dbfc

[v8] Need change to V8WorkerContextEventListener::reportError signature to match the base class.
https://bugs.webkit.org/show_bug.cgi?id=30264

Patch by David Levin <levin@chromium.org> on 2009-10-09
Reviewed by Eric Seidel.

The base class method changed in r48884.

Test: fast/worker/worker-script-error.html

* bindings/v8/V8WorkerContextEventListener.cpp:
(WebCore::V8WorkerContextEventListener::reportError):
* bindings/v8/V8WorkerContextEventListener.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49408 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Fix bad ChangeLog entry merge in r49019 by the commit queue script.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49407 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by Brady Eidson.

        https://bugs.webkit.org/show_bug.cgi?id=30260
        <rdar://problem/6447115> REGRESSION: Logging out from SAP doesn't work

        Tests: http/tests/xmlhttprequest/logout.html
               http/tests/xmlhttprequest/re-login-async.html
               http/tests/xmlhttprequest/re-login.html

        Fix several issues with existing credential handling code.

        * platform/network/CredentialStorage.cpp:
        (WebCore::pathToDefaultProtectionSpaceMap): Changed the data structure to a simpler one.
        (WebCore::originsWithCredentials): The reason for two-stage lookup above was that we didn't
        want to iterate paths for origins that never had credentials associated with them. Changed
        to use a separate HashSet for this.
        (WebCore::pathToDefaultProtectionSpaceMap): The concept of default per-path credentials didn't
        match the spec very well. UAs are supposed to deduce protection space from an URL, and then
        use whichever credentials are known for this protection space. So, OriginToDefaultBasicCredentialMap
        is now PathToDefaultProtectionSpaceMap.
        (WebCore::protectionSpaceMapKeyFromURL): Factored out a helper that extracts a directory
        URL from a given URL. These directory URLs are what we use as keys in PathToDefaultProtectionSpaceMap.
        (WebCore::CredentialStorage::set): Updated for above changes.
        (WebCore::findDefaultProtectionSpaceForURL): Factored out code iterating path length to find
        a prefix in OriginToDefaultBasicCredentialMap.
        (WebCore::CredentialStorage::set): Another version of set() can update credentials for a
        URL default protection space. It does nothing if the given URL doesn't correspond to a known
        protection space.
        (WebCore::CredentialStorage::get): Renamed from getDefaultAuthenticationCredential.

        * platform/network/CredentialStorage.h: Made the distinction between methods that use a known
        protection space and those that deduce one from URL more clear.

        * platform/network/mac/ResourceHandleMac.mm:
        (WebCore::ResourceHandle::start): Update credentials before starting the request for real.
        This makes the following pattern work:
          var req = new XMLHttpRequest("GET", "logout.html", "logout", "logout"); // wrong credentials
          req.send("");
          req.abort();
        Abort() is used here to avoid having UA present an auth dialog after getting a 401 response.
        Note that one cannot log in using the same method, because there isn't a known protection
        space for the URL yet in that case, so the added code has no effect.
        (WebCore::ResourceHandle::didReceiveAuthenticationChallenge): Use a correct persistence for
        calling receivedCredential(). This fixes logging in using an async XHR (withut abort(), of
        course), and matches sync case.
        (+[WebCoreSynchronousLoader loadRequest:allowStoredCredentials:returningResponse:error:]):
        Renamed getDefaultAuthenticationCredential() to get().

        * platform/network/cf/ResourceHandleCFNet.cpp:
        (WebCore::ResourceHandle::start):
        (WebCore::WebCoreSynchronousLoader::load):
        Same changes as in Mac code.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49406 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-09  Enrica Casucci  <enrica@apple.com>

        Reviewed by Adele Peterson.

        Undoing a indent removes text instead of it's formatting.
        <rdar://problem/7169206>
        https://bugs.webkit.org/show_bug.cgi?id=28722
        Added test case for https://bugs.webkit.org/show_bug.cgi?id=23995

        * editing/execCommand/19653-1-expected.txt: Re-baselined.
        * editing/execCommand/indent-div-inside-list-expected.txt: Re-baselined.
        * editing/execCommand/indent-nested-blockquotes-expected.txt: Re-baselined.
        * editing/execCommand/indent-nested-div-expected.txt: Re-baselined.
        * editing/execCommand/indent-second-paragraph-in-blockquote-expected.txt: Re-baselined.
        * editing/execCommand/indent-with-style-expected.txt: Added.
        * editing/execCommand/indent-with-style.html: Added.
        * editing/undo/undo-indent-expected.txt: Added.
        * editing/undo/undo-indent.html: Added.
        * platform/mac/editing/execCommand/4641880-2-expected.txt: Re-baselined.
        * platform/mac/editing/execCommand/4916541-expected.txt: Re-baselined.
        * platform/mac/editing/execCommand/5138441-expected.txt: Re-baselined.
        * platform/mac/editing/execCommand/5481523-expected.txt: Re-baselined.
        * platform/mac/editing/execCommand/indent-selection-expected.txt: Re-baselined.
2009-10-09  Enrica Casucci  <enrica@apple.com>

        Reviewed by Adele Peterson.

        Undoing a indent removes text instead of it's formatting.
        <rdar://problem/7169206>
        https://bugs.webkit.org/show_bug.cgi?id=28722
        Restored the behavior prior to http://trac.webkit.org/changeset/46143. This change implemented the
        indent command making simple DOM operation, without using moveParagraph which cause undo to be broken.

        Added editing/undo/undo-indent.html test.

        * editing/IndentOutdentCommand.cpp:
        (WebCore::IndentOutdentCommand::prepareBlockquoteLevelForInsertion): Restored.
        (WebCore::IndentOutdentCommand::tryIndentingAsListItem): Merged changes.
        (WebCore::IndentOutdentCommand::indentIntoBlockquote): Restored.
        (WebCore::IndentOutdentCommand::indentRegion): Restored and added call to updateLayout to
        ensure TextIterator could correctly calculate the position starting from the index.
        * editing/IndentOutdentCommand.h: Restored.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49405 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-09  Rimas Kudelis  <rq@akl.lt>

        Rubber-stamped by Gustavo Noronha

        https://bugs.webkit.org/show_bug.cgi?id=30054
        Localization of WebKit GTK into Lithuanian

        * lt.po: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49404 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-09  Jens Alfke  <snej@chromium.org>

        Reviewed by Darin Adler.

        Optimization: Many StringImpl transformations are no-ops and should just return 'this'
        https://bugs.webkit.org/show_bug.cgi?id=30186

        Optimized StringImpl methods lower(), stripWhiteSpace() and simplifyWhiteSpace() to
        detect no-ops and return this instead of creating a new instance.
        Empirical testing shows that the majority of calls to these methods are no-ops, making
        this worthwhile even if (in the case of lower()) the non-no-op case is slightly slowed.
        Upper() is very rarely a no-op, so it wasn't worthwhile to optimize it.

        * platform/text/StringImpl.cpp:
        (WebCore::StringImpl::lower):
        (WebCore::StringImpl::upper): Just add a comment explaining why this wasn't optimized
        (WebCore::StringImpl::stripWhiteSpace):
        (WebCore::StringImpl::simplifyWhiteSpace):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49403 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-09  Dirk Schulze  <krit@webkit.org>

        Reviewed by Oliver Hunt.

        SVG Filter feGaussianBlur implementation is missing
        [https://bugs.webkit.org/show_bug.cgi?id=28141]

        This is the implementation of GaussianBlur filter for SVG.

        There is already a test for feGaussianBlur
        Test: svg/W3C-SVG-1.1/filters-gauss-01-b-w3c.svg

        Test: svg/filters/feGaussianBlur.svg

        * platform/graphics/filters/FilterEffect.cpp:
        (WebCore::FilterEffect::FilterEffect):
        * platform/graphics/filters/FilterEffect.h:
        (WebCore::FilterEffect::isAlphaImage):
        (WebCore::FilterEffect::setIsAlphaImage):
        * platform/graphics/filters/SourceAlpha.cpp:
        (WebCore::SourceAlpha::apply):
        * svg/graphics/filters/SVGFEGaussianBlur.cpp:
        (WebCore::boxBlur):
        (WebCore::FEGaussianBlur::apply):

        Test for feGaussianBlur implementation

        * platform/mac/svg/filters/feGaussianBlur-expected.checksum: Added.
        * platform/mac/svg/filters/feGaussianBlur-expected.png: Added.
        * platform/mac/svg/filters/feGaussianBlur-expected.txt: Added.
        * svg/filters/feGaussianBlur.svg: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49402 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-09  Philippe Normand  <pnormand@igalia.com>

        Reviewed by Gustavo Noronha.

        [GTK] QoS support in the video sink
        https://bugs.webkit.org/show_bug.cgi?id=29959

        Removed the async buffer queue from the sink. Synchronize the
        render method of the sink using a g_timeout_add() combined with a
        gcond triggered when the buffer has been rendered.
        Also fixed the video sink reference handling in the player, now
        that the idle is not there anymore to mess up things.

        * platform/graphics/gtk/MediaPlayerPrivateGStreamer.cpp:
        (WebCore::MediaPlayerPrivate::~MediaPlayerPrivate):
        (WebCore::MediaPlayerPrivate::createGSTPlayBin):
        * platform/graphics/gtk/VideoSinkGStreamer.cpp:
        (webkit_video_sink_init):
        (webkit_video_sink_timeout_func):
        (webkit_video_sink_render):
        (webkit_video_sink_dispose):
        (unlock_buffer_mutex):
        (webkit_video_sink_unlock):
        (webkit_video_sink_stop):
        (webkit_video_sink_class_init):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49401 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-08  Dirk Schulze  <krit@webkit.org>

        Reviewed by Nikolas Zimmermann.

        feMorphology filter is not implemented
        [https://bugs.webkit.org/show_bug.cgi?id=5863]

        Adds SVGFEMorphologyElement, for the SVG filter. It also fixes a bug
        with the attribute "operator". It was only accessible via "_operator"
        on JavaScript.

        Test: svg/dom/operatorAttribute.xhtml

        * DerivedSources.cpp:
        * DerivedSources.make:
        * GNUmakefile.am:
        * WebCore.SVG.Filters.exp:
        * WebCore.gypi:
        * WebCore.pro:
        * WebCore.xcodeproj/project.pbxproj:
        * bindings/objc/DOM.mm:
        (WebCore::createElementClassMap):
        * bindings/objc/DOMSVG.h:
        * bindings/scripts/CodeGeneratorJS.pm:
        * bindings/scripts/CodeGeneratorObjC.pm:
        * page/DOMWindow.idl:
        * svg/SVGAllInOne.cpp:
        * svg/SVGFECompositeElement.idl:
        * svg/SVGFEMorphologyElement.cpp: Added.
        (WebCore::SVGFEMorphologyElement::SVGFEMorphologyElement):
        (WebCore::SVGFEMorphologyElement::~SVGFEMorphologyElement):
        (WebCore::SVGFEMorphologyElement::setRadius):
        (WebCore::SVGFEMorphologyElement::parseMappedAttribute):
        (WebCore::SVGFEMorphologyElement::build):
        * svg/SVGFEMorphologyElement.h: Added.
        * svg/SVGFEMorphologyElement.idl: Added.
        * svg/graphics/filters/SVGFEMorphology.cpp:
        (WebCore::operator<<):
        * svg/graphics/filters/SVGFEMorphology.h:
        (WebCore::):
        * svg/svgtags.in:

        Reviewed by Nikolas Zimmermann.

        Attribute "operator" schould only be accessible by "operator", not by "_operator"

        * svg/dom/operatorAttribute-expected.txt: Added.
        * svg/dom/operatorAttribute.html: Added.
        * svg/dom/script-tests/operatorAttribute.js: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49400 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-09  Yury Semikhatsky  <yurys@chromium.org>

        Reviewed by Dimitri Glazkov.

        Fix Chromium breakage.

        https://bugs.webkit.org/show_bug.cgi?id=30104

        * inspector/front-end/inspector.js:
        (WebInspector.addCookieDomain): check that Storage panel is present before calling its methods.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49399 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Migrated some code that didn't belong out of Structure.

Patch by Geoffrey Garen <ggaren@apple.com> on 2009-10-09
Reviewed by Sam Weinig.

SunSpider says maybe 1.03x faster.

* runtime/JSCell.h: Nixed Structure::markAggregate, and made marking of
a Structure's prototype the direct responsility of the object using it.
(Giving Structure a mark function was misleading because it implied that
all live structures get marked during GC, when they don't.)

* runtime/JSGlobalObject.cpp:
(JSC::markIfNeeded):
(JSC::JSGlobalObject::markChildren): Added code to mark prototypes stored
on the global object. Maybe this wasn't necessary, but now we don't have
to wonder.

* runtime/JSObject.cpp:
(JSC::JSObject::getPropertyNames):
(JSC::JSObject::getOwnPropertyNames):
(JSC::JSObject::getEnumerableNamesFromClassInfoTable):
* runtime/JSObject.h:
(JSC::JSObject::markChildrenDirect):
* runtime/PropertyNameArray.h:
* runtime/Structure.cpp:
* runtime/Structure.h:
(JSC::Structure::setEnumerationCache):
(JSC::Structure::enumerationCache): Moved property name gathering code
from Structure to JSObject because having a Structure iterate its JSObject
was a layering violation. A JSObject is implemented using a Structure; not
the other way around.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49398 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-09  Joe Ligman  <joseph.ligman@nokia.com>

        Reviewed by Simon Hausmann.

        Sets Qt::WA_InputMethodEnabled and Qt::ImhHiddenText for password fields in EditorClientQt
        setInputMethodState. This change is needed so widgets such as the s60 software
        input panel can receive input method events for password fields.
        It's up to the Qt platform to determine which widget will receive input method
        events when these flags are set.
        Also added implementation for setInputMethodEnabled and setInputMethodHint
        to QGraphicsWebViewPrivate and QWebViewPrivate. This change removes the direct
        dependency on QWebView and uses QWebPageClient.
        Added autotest to tst_qwebpage.cpp

        https://bugs.webkit.org/show_bug.cgi?id=30023

        * Api/qgraphicswebview.cpp:
        (QGraphicsWebViewPrivate::setInputMethodEnabled):
        (QGraphicsWebViewPrivate::setInputMethodHint):
        * Api/qwebview.cpp:
        (QWebViewPrivate::setInputMethodEnabled):
        (QWebViewPrivate::setInputMethodHint):
        * WebCoreSupport/EditorClientQt.cpp:
        (WebCore::EditorClientQt::setInputMethodState):
        * tests/qwebpage/tst_qwebpage.cpp:
        (tst_QWebPage::inputMethods):

2009-10-09  Joe Ligman  <joseph.ligman@nokia.com>

        Reviewed by Simon Hausmann.

        [Qt] Added pure virtual methods setInputMethodEnabled and setInputMethodHint to QWebPageClient

        https://bugs.webkit.org/show_bug.cgi?id=30023

        * platform/qt/QWebPageClient.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49397 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Requested by Mark Rowe.

Patch by Darin Adler <darin@apple.com> on 2009-10-09
Moved DOM test that was in fast/js to fast/dom where it belongs.

* fast/dom/Document/invalid-domain-change-throws-exception-expected.txt:
Copied from fast/js/invalid-domain-change-throws-exception-expected.txt.
* fast/dom/Document/invalid-domain-change-throws-exception.html:
Copied from fast/js/invalid-domain-change-throws-exception.html.
Regenerated with the make-script-test-wrappers.
* fast/dom/Document/script-tests/invalid-domain-change-throws-exception.js:
Copied from fast/js/resources/invalid-domain-change-throws-exception.js.

* fast/js/invalid-domain-change-throws-exception-expected.txt: Removed.
* fast/js/invalid-domain-change-throws-exception.html: Removed.
* fast/js/resources/invalid-domain-change-throws-exception.js: Removed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49396 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Fix ChangeLog patch reviewer to be accurate.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49395 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Implement beforeload for images.  ImageLoadEventSender has been refactored into a more generic
ImageEventSender that can be used by both load and beforeload events.  If the document has any
beforeload listeners, then the installation of images onto the renderer becomes asynchronous
and will be held up until the beforeload event can fire at a later date.

Reviewed by Adam Roben.

Both beforeload and load events now fire at the end of the tokenizer write() methods, so that
in the typical parsing case we don't have to put off the beforeload/load events until after
a layout or paint might already have happened.  This lets beforeload/load not cause extra
layouts and repaints.

* dom/ContainerNode.cpp:
(WebCore::ContainerNode::dispatchBeforeLoadEvent):
* dom/Document.cpp:
(WebCore::Document::implicitClose):
(WebCore::Document::addListenerTypeIfNeeded):
* dom/Document.h:
(WebCore::Document::):
* dom/XMLTokenizer.cpp:
(WebCore::XMLTokenizer::write):
* html/HTMLImageElement.cpp:
(WebCore::HTMLImageElement::attach):
* html/HTMLInputElement.cpp:
(WebCore::HTMLInputElement::attach):
* html/HTMLTokenizer.cpp:
(WebCore::HTMLTokenizer::write):
* loader/ImageLoader.cpp:
(WebCore::ImageBeforeLoadEventSender::ImageBeforeLoadEventSender):
(WebCore::ImageLoadEventSender::ImageLoadEventSender):
(WebCore::beforeLoadEventSender):
(WebCore::ImageLoader::ImageLoader):
(WebCore::ImageLoader::~ImageLoader):
(WebCore::ImageLoader::setImage):
(WebCore::ImageLoader::setLoadingImage):
(WebCore::ImageLoader::updateFromElement):
(WebCore::ImageLoader::notifyFinished):
(WebCore::ImageLoader::dispatchPendingBeforeLoadEvent):
(WebCore::ImageLoader::dispatchPendingEvents):
(WebCore::ImageEventSender::ImageEventSender):
(WebCore::ImageEventSender::dispatchEventSoon):
(WebCore::ImageEventSender::cancelEvent):
(WebCore::ImageEventSender::dispatchPendingEvents):
(WebCore::ImageEventSender::timerFired):
* loader/ImageLoader.h:
(WebCore::ImageLoader::haveFiredBeforeLoadEvent):
* wml/WMLImageElement.cpp:
(WebCore::WMLImageElement::attach):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49394 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-09  Pavel Feldman  <pfeldman@chromium.org>

        Reviewed by Dimitri Glazkov.

        Web Inspector: Wrap Database into ScriptObject in v8 bindings.

        https://bugs.webkit.org/show_bug.cgi?id=30174

        No tests due to no Script* test infrastructure.

        * bindings/v8/ScriptObjectQuarantine.cpp:
        (WebCore::getQuarantinedScriptObject):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49393 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Add SPI to specifically enable visited link tracking.

Reviewed by Dan Bernstein.

* History/WebHistory.mm:
(+[WebHistory _setVisitedLinkTrackingEnabled:]):
* History/WebHistoryPrivate.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49392 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-09  Girish Ramakrishnan  <girish@forwardbias.in>

        Reviewed by Simon Hausmann.

        [Qt] Windowless Plugins : Fix crash when using QWebPage without QWebView.

        'client' is 0 when we have no view.

        https://bugs.webkit.org/show_bug.cgi?id=30251

        * plugins/qt/PluginViewQt.cpp:
        (WebCore::PluginView::paint):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49391 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Set the svn:mime-type property of recently-added PNG files to image/png.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49390 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-09  Girish Ramakrishnan  <girish@forwardbias.in>

        Reviewed by Simon Hausmann.

        [Qt] Plugins : Add missing setCurrentPlugin(0)

        https://bugs.webkit.org/show_bug.cgi?id=30248

        * plugins/qt/PluginViewQt.cpp:
        (WebCore::PluginView::dispatchNPEvent):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49389 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-09  Jakub Wieczorek  <faw217@gmail.com>

        Reviewed by Simon Hausmann.

        [Qt] beforeload event does not fire on script elements in XML
        https://bugs.webkit.org/show_bug.cgi?id=30235

        r49194 was lacking a change to XMLTokenizerQt.cpp, equivalent to the one
        made in XMLTokenizerLibxml2.cpp.

        * dom/XMLTokenizerQt.cpp:
        (WebCore::XMLTokenizer::parseEndElement):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49388 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-09  Zoltan Horvath  <zoltan@webkit.org>

        Reviewed by Darin Adler.

        Allow custom memory allocation control for WebCore's CSSParserValueList
        https://bugs.webkit.org/show_bug.cgi?id=30249

        Inherits CSSParserValueList class from FastAllocBase because it has
        been instantiated by 'new' in WebCore/css/CSSGrammar.y:1271.

        * css/CSSParserValues.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49387 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-09  Girish Ramakrishnan  <girish@forwardbias.in>

        Reviewed by Simon Hausmann.

        [Qt] Windowless Plugins : Create Pixmap only when size changes.

        https://bugs.webkit.org/show_bug.cgi?id=30214

        * plugins/qt/PluginViewQt.cpp:
        (WebCore::PluginView::updatePluginWidget):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49386 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Remove the WebKitPluginHalterEnabledPreferenceKey in favor of checking
for the existence of a PluginHalterDelegate.

This fixes a leak of WebPluginHalterClients:
https://bugs.webkit.org/show_bug.cgi?id=30119.

WebCore:

Reviewed by Dan Bernstein.

* WebCore.base.exp:
Removed the export of WebCore::Settings::setPluginHalterEnabled().

* loader/EmptyClients.h:
(WebCore::EmptyPluginHalterClient::enabled):

* page/Page.cpp:
(WebCore::Page::Page):
Remove initialization of m_pluginHalterClient, which was removed. If a
non-null PluginHalterClient was passed, create the PluginHalter, and
set its allowed run time.

* page/Page.h:
Removed pluginHalterEnabledStateChanged() and m_pluginHalterClient.

* page/PluginHalter.cpp:
(WebCore::PluginHalter::didStartPlugin):
Check whether the PluginHalterClient is enabled.
(WebCore::PluginHalter::didStopPlugin):
Ditto.

* page/PluginHalter.h:
Made m_client an OwnPtr.

* page/PluginHalterClient.h:
Added a function to return the enabled state.

* page/Settings.cpp:
(WebCore::Settings::Settings):
Remove initialization of removed member.

* page/Settings.h:
Removed settings for the enabled state of the PluginHalter; we now use
the existence of a WebPluginHalterDelegate to determine whether the
PluginHalter is enabled.

WebKit/mac:

* WebCoreSupport/WebPluginHalterClient.h:
Add declaration for new function.

* WebCoreSupport/WebPluginHalterClient.mm:
(WebPluginHalterClient::enabled):
Check whether the UIDelegate responds to shouldHaltPlugin.

* WebView/WebPreferenceKeysPrivate.h:
Remove the "enabled" preference key.

* WebView/WebPreferences.mm:
(+[WebPreferences initialize]):
Remove the initialization of the pref.

* WebView/WebPreferencesPrivate.h:
Remove the getter/setter for this pref.

* WebView/WebView.mm:
(-[WebView _preferencesChangedNotification:]):
Remove propagation of the pref.

WebKit/win:

* Interfaces/IWebPreferencesPrivate.idl:
Remove the getter and setter.

* WebCoreSupport/WebPluginHalterClient.cpp:
(WebPluginHalterClient::enabled):
Return true if the delegate exists.

* WebCoreSupport/WebPluginHalterClient.h:
Declare implementation of enabled().

* WebPreferenceKeysPrivate.h:
Removed the pref key.

* WebPreferences.cpp:
(WebPreferences::initializeDefaultSettings):
Removed initialization of the pref.

* WebPreferences.h:
Removed declaration of the getter/setter.

* WebView.cpp:
(WebView::notifyPreferencesChanged):
Don't propagate the pref.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49385 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Attempt to fix the GTK release build.

* GNUmakefile.am: Include Grammar.cpp in release builds now that
AllInOneFile.cpp is gone.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49384 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-09  Gabor Loki  <loki@inf.u-szeged.hu>

        Rubber-stamped by Eric Seidel.

        Add ARM JIT support for Gtk port (disabled by default)
        https://bugs.webkit.org/show_bug.cgi?id=30228

        * GNUmakefile.am:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49383 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-08  Alex Milowski  <alex@milowski.com>

        Reviewed by Darin Adler.

        Added a basic description of the layout and rendering project and added links to the blog entries on rendering.

        * projects/layout/index.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49382 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-08  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        Move executeScript from FrameLoader to ScriptController
        https://bugs.webkit.org/show_bug.cgi?id=30200

        Update API call.

        * Api/qwebframe.cpp:
        (QWebFrame::evaluateJavaScript):
2009-10-08  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        Move executeScript from FrameLoader to ScriptController
        https://bugs.webkit.org/show_bug.cgi?id=30200

        Update API call.

        * webkit/webkitwebview.cpp:
        (webkit_web_view_execute_script):
2009-10-08  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        Move executeScript from FrameLoader to ScriptController
        https://bugs.webkit.org/show_bug.cgi?id=30200

        Update API call.

        * WebView.cpp:
        (WebView::stringByEvaluatingJavaScriptFromString):
2009-10-08  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        Move executeScript from FrameLoader to ScriptController
        https://bugs.webkit.org/show_bug.cgi?id=30200

        Update API call.

        * WebFrame.cpp:
        (wxWebFrame::RunScript):
2009-10-08  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        Move executeScript from FrameLoader to ScriptController
        https://bugs.webkit.org/show_bug.cgi?id=30200

        Update API call.

        * WebView/WebFrame.mm:
        (-[WebFrame _stringByEvaluatingJavaScriptFromString:forceUserGesture:]):
        * WebView/WebView.mm:
        (-[WebView aeDescByEvaluatingJavaScriptFromString:]):
2009-10-08  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        Move executeScript from FrameLoader to ScriptController
        https://bugs.webkit.org/show_bug.cgi?id=30200

        These methods have virtually no interaction with FrameLoader.  They
        really seem like they ought to belong to ScriptController.

        * WebCore.base.exp:
        * bindings/js/ScheduledAction.cpp:
        (WebCore::ScheduledAction::execute):
        * bindings/js/ScriptController.cpp:
        (WebCore::ScriptController::ScriptController):
        (WebCore::ScriptController::executeScript):
        (WebCore::ScriptController::executeIfJavaScriptURL):
        (WebCore::ScriptController::evaluate):
        * bindings/js/ScriptController.h:
        * bindings/v8/ScriptController.cpp:
        (WebCore::ScriptController::ScriptController):
        (WebCore::ScriptController::executeScript):
        (WebCore::ScriptController::executeIfJavaScriptURL):
        * bindings/v8/ScriptController.h:
        * dom/XMLTokenizer.cpp:
        (WebCore::XMLTokenizer::notifyFinished):
        * dom/XMLTokenizerLibxml2.cpp:
        (WebCore::XMLTokenizer::endElementNs):
        * dom/XMLTokenizerQt.cpp:
        (WebCore::XMLTokenizer::parseEndElement):
        * html/HTMLTokenizer.cpp:
        (WebCore::HTMLTokenizer::scriptExecution):
        * loader/FrameLoader.cpp:
        (WebCore::FrameLoader::changeLocation):
        (WebCore::FrameLoader::urlSelected):
        (WebCore::FrameLoader::requestFrame):
        (WebCore::FrameLoader::submitForm):
        (WebCore::FrameLoader::replaceDocument):
        * loader/FrameLoader.h:
        * plugins/PluginView.cpp:
        (WebCore::PluginView::performRequest):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49372 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Tiger build fix: added a few more variable initializations.

Patch by Geoffrey Garen <ggaren@apple.com> on 2009-10-08
* runtime/StringPrototype.cpp:
(JSC::stringProtoFuncReplace):
(JSC::stringProtoFuncSearch):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49371 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Qt build fix: added missing #include.

Patch by Geoffrey Garen <ggaren@apple.com> on 2009-10-08
* jsc.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49370 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Ask the History Delegate to populate the visited links hash.
<rdar://problem/7285293> and https://webkit.org/b/29904

Reviewed by Dan Bernstein.

WebKit/mac:

* History/WebHistory.mm:
(+[WebHistory _removeAllVisitedLinks]):
* History/WebHistoryPrivate.h:
* WebCoreSupport/WebChromeClient.mm:
(WebChromeClient::populateVisitedLinks):
* WebView/WebDelegateImplementationCaching.h:
* WebView/WebDelegateImplementationCaching.mm:
(CallHistoryDelegate):
* WebView/WebHistoryDelegate.h:
* WebView/WebView.mm:
(-[WebView _cacheHistoryDelegateImplementations]):
(-[WebView addVisitedLinks:]):
* WebView/WebViewPrivate.h:

WebKitTools:

Add the ability for LayoutTestController to clear all visited links.
Also lets the History Delegate dump visited links, but only if this test specifically cleared them.

* DumpRenderTree/LayoutTestController.cpp:
(LayoutTestController::LayoutTestController):
(removeAllVisitedLinksCallback):
(LayoutTestController::staticFunctions):
* DumpRenderTree/LayoutTestController.h:
* DumpRenderTree/mac/LayoutTestControllerMac.mm:
(LayoutTestController::removeAllVisitedLinks):

* DumpRenderTree/mac/HistoryDelegate.mm:
(-[HistoryDelegate populateVisitedLinksForWebView:]):

* DumpRenderTree/gtk/LayoutTestControllerGtk.cpp:
(LayoutTestController::removeAllVisitedLinks):
* DumpRenderTree/win/LayoutTestControllerWin.cpp:
(LayoutTestController::removeAllVisitedLinks):
* DumpRenderTree/wx/LayoutTestControllerWx.cpp:
(LayoutTestController::removeAllVisitedLinks):

LayoutTests:

* http/tests/globalhistory/history-delegate-basic-visited-links-expected.txt: Added.
* http/tests/globalhistory/history-delegate-basic-visited-links.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49369 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Tiger build fix: initialize variable whose initialization the compiler
can't otherwise figure out.

Patch by Geoffrey Garen <ggaren@apple.com> on 2009-10-08
* runtime/RegExpObject.cpp:
(JSC::RegExpObject::match):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49368 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Windows build fix: updated exports.

Patch by Geoffrey Garen <ggaren@apple.com> on 2009-10-08
* JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.def:
* JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore_debug.def:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49367 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Tiger build fix: fixed file name case.

Patch by Geoffrey Garen <ggaren@apple.com> on 2009-10-08
* JavaScriptCore.xcodeproj/project.pbxproj:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49366 268f45cc-cd09-0410-ab3c-d52691b4dbfc

At long last, I pronounce the death of AllInOneFile.cpp.

Patch by Geoffrey Garen <ggaren@apple.com> on 2009-10-08
Reviewed by Maciej Stachowiak.

SunSpider reports a 1.01x speedup.

* AllInOneFile.cpp: Removed.
* GNUmakefile.am:
* JavaScriptCore.exp:
* JavaScriptCore.gypi:
* JavaScriptCore.xcodeproj/project.pbxproj: Added missing project files
to compilation stages.

* parser/Grammar.y:
* parser/Lexer.cpp:
* parser/Lexer.h:
(JSC::jscyylex):
* runtime/ArrayConstructor.cpp:
(JSC::constructArrayWithSizeQuirk):
* runtime/Collector.h:
* runtime/JSCell.cpp:
(JSC::JSCell::operator new):
* runtime/JSCell.h:
(JSC::JSCell::operator new):
* runtime/JSGlobalObject.cpp:
(JSC::JSGlobalObject::operator new):
* runtime/JSNumberCell.h:
(JSC::JSNumberCell::operator new):
* runtime/JSString.cpp:
* runtime/JSString.h:
(JSC::jsString):
(JSC::jsSubstring):
(JSC::jsOwnedString):
* runtime/RegExpConstructor.cpp:
* runtime/RegExpConstructor.h:
(JSC::RegExpConstructorPrivate::RegExpConstructorPrivate):
(JSC::RegExpConstructorPrivate::lastOvector):
(JSC::RegExpConstructorPrivate::tempOvector):
(JSC::RegExpConstructorPrivate::changeLastOvector):
(JSC::RegExpConstructor::performMatch):
* runtime/StringPrototype.cpp:
(JSC::stringProtoFuncMatch):
* yarr/RegexJIT.cpp:
* yarr/RegexJIT.h:
(JSC::Yarr::executeRegex): Inlined a few things that Shark said
were hot, on the presumption that AllInOneFile.cpp used to inline them
automatically.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49365 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-08  Daniel Bates  <dbates@webkit.org>

        Reviewed by Adam Roben.

        https://bugs.webkit.org/show_bug.cgi?id=30175

        The Windows DRT equivalent of DoDragDrop (i.e. UIDelegate::doDragDrop) does not return
        the OLE drag-and-drop return value like the function it emulates. Currently,
        UIDelegate::doDragDrop returns a hard-coded S_OK. Hence, the caller cannot determine
        whether the drag-and-drop operation was successful or was cancelled.

        This patch fixes this issue by having UIDelegate::doDragDrop return the OLE drag-and-drop
        return value according to whether the drop operation was successful or not.

        * DumpRenderTree/win/EventSender.cpp:
        (doMouseUp): Added parameter oleDragAndDropReturnValue.
        (replaySavedEvents): Ditto.
        * DumpRenderTree/win/EventSender.h:
        * DumpRenderTree/win/UIDelegate.cpp:
        (UIDelegate::doDragDrop): Modified to return OLE drag-and-drop return value.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49336 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Rolled out 49331 because it still doesn't build on Windows

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49335 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Windows build fix: added missing #include.

Patch by Geoffrey Garen <ggaren@apple.com> on 2009-10-08
* interpreter/CallFrame.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49334 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Windows build fix: added missing #include.

JavaScriptCore:

Patch by Geoffrey Garen <ggaren@apple.com> on 2009-10-08
* runtime/PropertyNameArray.h:

WebCore:

Patch by Geoffrey Garen <ggaren@apple.com> on 2009-10-08
* ForwardingHeaders/runtime/StructureChain.h: Copied from WebCore/ForwardingHeaders/runtime/PropertyNameArray.h.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49333 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Windows build fix: added missing #include.

Patch by Geoffrey Garen <ggaren@apple.com> on 2009-10-08
* runtime/NumberPrototype.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49332 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Migrated some code that didn't belong out of Structure.

Patch by Geoffrey Garen <ggaren@apple.com> on 2009-10-08
Reviewed by Sam Weinig.

SunSpider says maybe 1.03x faster.

* runtime/JSCell.h: Nixed Structure::markAggregate, and made marking of
a Structure's prototype the direct responsility of the object using it.
(Giving Structure a mark function was misleading because it implied that
all live structures get marked during GC, when they don't.)

* runtime/JSGlobalObject.cpp:
(JSC::markIfNeeded):
(JSC::JSGlobalObject::markChildren): Added code to mark prototypes stored
on the global object. Maybe this wasn't necessary, but now we don't have
to wonder.

* runtime/JSObject.cpp:
(JSC::JSObject::getPropertyNames):
(JSC::JSObject::getOwnPropertyNames):
(JSC::JSObject::getEnumerableNamesFromClassInfoTable):
* runtime/JSObject.h:
(JSC::JSObject::markChildrenDirect):
* runtime/PropertyNameArray.h:
* runtime/Structure.cpp:
* runtime/Structure.h:
(JSC::Structure::setEnumerationCache):
(JSC::Structure::enumerationCache): Moved property name gathering code
from Structure to JSObject because having a Structure iterate its JSObject
was a layering violation. A JSObject is implemented using a Structure; not
the other way around.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49331 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Revert 49329: broke Windows Debug (Build)

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49330 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-08  Adam Langley  <agl@google.com>

        Reviewed by Eric Seidel.

        Currently, Skia clip paths are 1-bit. This patch makes our path
        clipping anti-aliased for non-canvas drawing.

        http://code.google.com/p/chromium/issues/detail?id=5927
        https://bugs.webkit.org/show_bug.cgi?id=28820
        http://www.imperialviolet.org/2009/09/02/anti-aliased-clipping.html

        Already covered by layout tests. New baselines will be needed in the
        Chromium tree.

        * html/canvas/CanvasRenderingContext2D.cpp:
        (WebCore::CanvasRenderingContext2D::clip):
        * platform/graphics/GraphicsContext.h:
        * platform/graphics/cairo/GraphicsContextCairo.cpp:
        (WebCore::GraphicsContext::canvasClip):
        * platform/graphics/cg/GraphicsContextCG.cpp:
        (WebCore::GraphicsContext::canvasClip):
        * platform/graphics/haiku/GraphicsContextHaiku.cpp:
        (WebCore::GraphicsContext::canvasClip):
        * platform/graphics/qt/GraphicsContextQt.cpp:
        (WebCore::GraphicsContext::canvasClip):
        * platform/graphics/skia/GraphicsContextSkia.cpp:
        (WebCore::GraphicsContext::addInnerRoundedRectClip):
        (WebCore::GraphicsContext::clip):
        (WebCore::GraphicsContext::canvasClip):
        (WebCore::GraphicsContext::clipPath):
        * platform/graphics/skia/PlatformContextSkia.cpp:
        (PlatformContextSkia::clipPathAntiAliased):
        (PlatformContextSkia::restore):
        (PlatformContextSkia::applyAntiAliasedClipPaths):
        * platform/graphics/skia/PlatformContextSkia.h:
        * platform/graphics/win/GraphicsContextWin.cpp:
        (WebCore::GraphicsContext::canvasClip):
        * platform/graphics/wince/GraphicsContextWince.cpp:
        (WebCore::GraphicsContext::canvasClip):
        * platform/graphics/wx/GraphicsContextWx.cpp:
        (WebCore::GraphicsContext::canvasClip):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49329 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Rolled out r49322 and friends because they killed the build

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49328 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Fixes <https://bugs.webkit.org/show_bug.cgi?id=30237>.
DOM Content and Load lines aren't shown if a page is loaded when the inspector is up.

Reviewed by Timothy Hatcher.

Make sure to call update script object if we have an InspectorFrontend to
call it on when we get one of the new events.

* inspector/InspectorController.cpp:
(WebCore::InspectorController::mainResourceFiredDOMContentEvent):
(WebCore::InspectorController::mainResourceFiredLoadEvent):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49327 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Build fix: updated for removal of Structure::markAggregate().

Patch by Geoffrey Garen <ggaren@apple.com> on 2009-10-08
* bindings/js/JSDOMGlobalObject.cpp:
(WebCore::JSDOMGlobalObject::markChildren):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49326 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Removed some conflict markers from the ChangeLog

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49325 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Windows build fix: Added missing #include.

Patch by Geoffrey Garen <ggaren@apple.com> on 2009-10-08
* runtime/JSGlobalData.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49324 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Take one branch instead of two to test for JSValue().

Patch by Geoffrey Garen <ggaren@apple.com> on 2009-10-01
Reviewed by Sam Weinig.

1.1% SunSpider speedup.

* jit/JITCall.cpp:
(JSC::JIT::compileOpCall):
* jit/JITOpcodes.cpp:
(JSC::JIT::emit_op_to_jsnumber):
(JSC::JIT::emit_op_create_arguments):
* jit/JITPropertyAccess.cpp:
(JSC::JIT::emitSlow_op_get_by_val):
(JSC::JIT::emit_op_put_by_val): Test for the empty value tag, instead
of testing for the cell tag with a 0 payload.

* runtime/JSValue.cpp:
(JSC::JSValue::description): Added support for dumping the new empty value,
and deleted values, in debug builds.

* runtime/JSValue.h:
(JSC::JSValue::JSValue()): Construct JSValue() with the empty value tag.

(JSC::JSValue::JSValue(JSCell*)): Convert null pointer to the empty value
tag, to avoid having two different c++ versions of null / empty.

(JSC::JSValue::operator bool): Test for the empty value tag, instead
of testing for the cell tag with a 0 payload.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49323 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Make AtomicString create its StringImpl via create(), not the constructor,
so it gets allocated in a single heap block, saving memory and CPU cycles.
This eliminates two StringImpl constructors, making the remaining ones
unambiguous, so the "AdoptBuffer" parameter is no longer needed.
Added const attribute to UChar* in StringImpl constructor, eliminating the
need for several const_casts in calls to it.
StringImpl also unfriends AtomicString (OMG drama!!!)
https://bugs.webkit.org/show_bug.cgi?id=30141

Patch by Jens Alfke <snej@chromium.org> on 2009-10-08
Reviewed by Darin Adler.

* platform/text/AtomicString.cpp:
(WebCore::CStringTranslator::translate): Call StringImpl::create().
(WebCore::UCharBufferTranslator::translate): Ditto.
(WebCore::HashAndCharactersTranslator::translate): Ditto.
* platform/text/StringImpl.cpp:
(WebCore::StringImpl::StringImpl): Remove unnecessary AdoptBuffer param.
(WebCore::StringImpl::adopt): Ditto.
(WebCore::StringImpl::createUninitialized): Ditto.
(WebCore::StringImpl::create): Ditto.
(WebCore::StringImpl::crossThreadString): Ditto.
* platform/text/StringImpl.h:
(WebCore::StringImpl::setHash): Used by AtomicString when creating StringImpls.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49322 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-08  Aaron Boodman  <aa@chromium.org>

        Reviewed by Eric Seidel.

        r49278 added support for reusing isolated worlds, and the
        corresponding change to Chromium's LayoutTestController changed
        to expect two arguments. The first argument is the world that
        the script should run in.

        This change fixes the tests by updating them all to pass 0
        for the world id. The specific world id used for these tests
        doesn't really matter because they don't test world interaction.

        https://bugs.webkit.org/show_bug.cgi?id=30231

        * http/tests/security/isolatedWorld/all-window-properties.html:
        * http/tests/security/isolatedWorld/all-window-prototypes.html:
        * http/tests/security/isolatedWorld/body-properties.html:
        * http/tests/security/isolatedWorld/body-prototype.html:
        * http/tests/security/isolatedWorld/click-event.html:
        * http/tests/security/isolatedWorld/document-open.html:
        * http/tests/security/isolatedWorld/document-properties.html:
        * http/tests/security/isolatedWorld/document-prototype.html:
        * http/tests/security/isolatedWorld/global-variables.html:
        * http/tests/security/isolatedWorld/image-properties.html:
        * http/tests/security/isolatedWorld/image-prototype.html:
        * http/tests/security/isolatedWorld/location-properties.html:
        * http/tests/security/isolatedWorld/location-prototype.html:
        * http/tests/security/isolatedWorld/number-prototype.html:
        * http/tests/security/isolatedWorld/object-prototype.html:
        * http/tests/security/isolatedWorld/storage-properties.html:
        * http/tests/security/isolatedWorld/storage-prototype.html:
        * http/tests/security/isolatedWorld/string-prototype.html:
        * http/tests/security/isolatedWorld/window-properties.html:
        * http/tests/security/isolatedWorld/window-setTimeout-function.html:
        * http/tests/security/isolatedWorld/window-setTimeout-string.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49321 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WebCore: Remove the shouldLoadMediaElementURL frame loader client function.

Reviewed by Sam Weinig.

* loader/FrameLoader.cpp:
(WebCore::FrameLoader::willLoadMediaElementURL):

WebKit/mac: Remove the shouldLoadMediaElementURL implementation.

Reviewed by Sam Weinig.

* WebCoreSupport/WebFrameLoaderClient.h:
* WebCoreSupport/WebFrameLoaderClient.mm:
* WebView/WebPolicyDelegatePrivate.h:

WebKit/win: Remove the shouldLoadMediaElementURL implementation.

Reviewed by Sam Weinig.

* Interfaces/IWebPolicyDelegatePrivate.idl:
* WebCoreSupport/WebFrameLoaderClient.cpp:
(WebFrameLoaderClient::shouldUsePluginDocument):
* WebCoreSupport/WebFrameLoaderClient.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49320 268f45cc-cd09-0410-ab3c-d52691b4dbfc

SVGURIReference Build problem for V8.  An additional stale reference to SVGURIReference bindings
which wasn't exposed until a clean build.
https://bugs.webkit.org/show_bug.cgi?id=30217

Patch by John Gregg <johnnyg@google.com> on 2009-10-08
Reviewed by Dimitri Glazkov.

No new tests, just a build fix.

* bindings/v8/V8Index.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49319 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-08  Alejandro G. Castro  <alex@igalia.com>

        Reviewed by Xan Lopez.

        [GTK] Added support for a parameter setting the button that was
        pressed in the mouseDown function.
        https://bugs.webkit.org/show_bug.cgi?id=30220

        * WebKitTools/DumpRenderTree/gtk/EventSender.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49318 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-08  Nikolas Zimmermann  <nzimmermann@rim.com>

        Not reviewed. Sort XCode project file.

        * WebCore.xcodeproj/project.pbxproj:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49317 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-08  Alejandro G. Castro  <alex@igalia.com>

        Reviewed by Xan Lopez.

        [GTK] Added a line to the bottom of the expected result to match
        the output of the test.
        https://bugs.webkit.org/show_bug.cgi?id=30220

        * LayoutTests/platform/gtk/editing/pasteboard/middle-click-onpaste-
        expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49316 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-08  Patrick Mueller  <Patrick_Mueller@us.ibm.com>

        Reviewed by Timothy Hatcher.

        unselectable resources in resource panel
        https://bugs.webkit.org/show_bug.cgi?id=30079

        manual test added

        Also changed the way DnD for resources in the Resources panel is
        handled.

        * inspector/front-end/ResourcesPanel.js:
        (WebInspector.ResourceSidebarTreeElement.prototype.onattach):
        (WebInspector.ResourceSidebarTreeElement.prototype.ondragstart):
        * manual-tests/inspector/duplicate-resource-urls.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49315 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Fixes <http://webkit.org/b/30233>
Local Storage and Cookies show Blank Titles on Local Files.

Reviewed by Timothy Hatcher.

If a local storage or cookies sidebar item has no domain, set the
title to "Local Files", because that is what it represents.

* English.lproj/localizedStrings.js:
* inspector/front-end/StoragePanel.js:
(WebInspector.DOMStorageSidebarTreeElement.prototype.get mainTitle):
(WebInspector.CookieSidebarTreeElement.prototype.get mainTitle):
(WebInspector.CookieSidebarTreeElement.prototype.set mainTitle):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49314 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Fix failing media layout tests.  Don't make beforeload block loads of objects that aren't in the
document, since video/audio (and images eventually too) can load when not in the document yet.

Reviewed by Oliver Hunt.

* dom/ContainerNode.cpp:
(WebCore::ContainerNode::dispatchBeforeLoadEvent):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49313 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-08  Nikolas Zimmermann  <nzimmermann@rim.com>

        Reviewed by Eric Seidel.

        Move SVGFitToViewBox ANIMATED_* macros in classes that inherit from it
        https://bugs.webkit.org/show_bug.cgi?id=30230

        This patch also devirtualizes viewBoxToViewTransform(), is it's superflous.
        viewBoxToViewTransform() is now a simple static heper function in SVGFitToViewBox.
        As a result, the SVGSVGElement::viewBoxToViewTransform() function now resuses the same logic.

        As side-effect this patch fixes svg/custom/linking-a-03-b-transform.svg, the return
        statement in SVGSVGElement::viewBoxToViewTransform() was clearly wrong.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49312 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Correct a typo that fixes the resizing of columns in Cookie view when you
resize the Inspector window.

Reviewed by Jon Honeycutt.

* inspector/front-end/CookieItemsView.js:
(WebInspector.CookieItemsView.prototype.resize):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49311 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Fixes <https://bugs.webkit.org/show_bug.cgi?id=29811>
Text in Inspector's Styles gear menu is invisible.

Reviewed by Adam Roben.

Adds a style rule for the option text and hr's in the gear menu,
to prevent them from being transparent.

* inspector/front-end/inspector.css:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49310 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Fix build problem with SVGURIReference in V8.
https://bugs.webkit.org/show_bug.cgi?id=30217

Patch by John Gregg <johnnyg@google.com> on 2009-10-08
Reviewed by David Levin.

No new tests, just a build fix.

* WebCore.gypi:
* bindings/v8/DerivedSourcesAllInOne.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49309 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Make beforeload work on <video>/<audio> elements.

Reviewed by Eric Carlson.

Added fast/dom/beforeload/video-before-load.html

* html/HTMLMediaElement.cpp:
(WebCore::HTMLMediaElement::parseMappedAttribute):
(WebCore::HTMLMediaElement::selectMediaResource):
(WebCore::HTMLMediaElement::selectNextSourceChild):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49308 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-08  Xan Lopez  <xlopez@igalia.com>

        Skip another failing test with reference to its bug.

        * platform/gtk/Skipped:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49307 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-08  Nikolas Zimmermann  <nzimmermann@rim.com>

    Reviewed by Eric Seidel.

    Move SVGExternalResourcesRequired ANIMATED_* macros in classes that inherit from it
    https://bugs.webkit.org/show_bug.cgi?id=30218

    The long term goal is to kill the need for the virtual contextElement() function in SVG code.
    No change in functionality, thus no new tests.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49306 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Make IWebInspectorPrivate be accessed in a more standard way

Fixes <http://webkit.org/b/30215> Make IWebInspectorPrivate be
accessed in a more standard way

Reviewed by John Sullivan and Tim Hatcher.

WebKit/win:

IWebViewPrivate::inspectorPrivate has been removed. Callers should
instead use IWebViewPrivate::inspector and then use QueryInterface to
get the IWebInspectorPrivate interface. This matches how our other COM
classes work.

* Interfaces/IWebViewPrivate.idl: Removed inspectorPrivate.

* WebInspector.cpp:
(WebInspector::QueryInterface): Support querying to
IWebInspectorPrivate.

* WebView.cpp:
* WebView.h:
Removed inspectorPrivate.

WebKitTools:

Use QueryInterface to get IWebInspectorPrivate

* DumpRenderTree/win/LayoutTestControllerWin.cpp:
(LayoutTestController::evaluateInWebInspector): Get the IWebInspector
by calling IWebViewPrivate::inspector, then use QueryInterface to get
to the IWebInspectorPrivate interface.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49305 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Move the new IWebViewPrivate::inspectorPrivate function after all functions that existed when Safari 4.0.3 was released

Fixes <http://webkit.org/b/30085> REGRESSION (r49091): run-safari
crashes in Safari.dll

Reviewed by John Sullivan.

* Interfaces/IWebViewPrivate.idl: Moved inspectorPrivate to the end of
the interface, so it won't modify the vtable that Safari 4.0.3 sees.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49304 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-08  Dirk Schulze  <krit@webkit.org>

        Reviewed by Darin Adler.

        SVG feComposite: operator over mixes up inputs
        [https://bugs.webkit.org/show_bug.cgi?id=30205]

        SVG feComposite operator over mixed up the inputs.
        The pixel-test svg/filters/feComposite.svg can already test this.
        No new test result is needed as long as filters are disabled.

        * platform/graphics/filters/FEComposite.cpp:
        (WebCore::FEComposite::apply):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49303 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-08  Xan Lopez  <xlopez@igalia.com>

        Skip flakey transition test.

        * platform/gtk/Skipped:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49302 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Reviewed by George Staikos.

Move SVGURIReference ANIMATED_* macros in classes that inherit from it
https://bugs.webkit.org/show_bug.cgi?id=30184

The long term goal is to kill the need for the virtual contextElement() function in SVG code.
No change in functionality, thus no new tests.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49301 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-10-08  Jocelyn Turcotte  <jocelyn.turcotte@nokia.com>

        Reviewed by Simon Hausmann.

        [Qt] Re-add the QApplication::syncX() line for plugin windows removed
        by r49169. Also changed the location of the sync just after the
        creation instead of just before sending it to the plugin.
        https://bugs.webkit.org/show_bug.cgi?id=25053

        * plugins/qt/PluginViewQt.cpp:
        (WebCore::PluginView::platformStart):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@49300 268f45cc-cd09-0410-ab3c-d52691b4dbfc