GitLab

2009-04-22  Brady Eidson  <beidson@apple.com>

        Reviewed by Maciej Stachowiak

        <rdar://problem/6786961> - "Are you sure you want to resubmit this form?" nag displays twice.

        For most loads, we were consulting the navigation policy delegate twice.  Once from FrameLoader before
        the load started and once from MainResourceLoader in its willSendRequest callback.

        In the past we tried removing MainResourceLoader's call altogether.  This caused a regression where urls 
        that redirect to a url handled by an external application would no longer work in Safari.  It probably 
        also broke other WebKit apps in subtle ways.

        Changing MainResourceLoader to make the check only on redirects fixes both bugs.  We now only call the 
        policy delegate once for most standard loads, but we do correctly call it a second time for redirects.

        Tests: http/tests/misc/policy-delegate-called-twice.html
               http/tests/misc/redirect-to-external-url.html

        * loader/MainResourceLoader.cpp:
        (WebCore::MainResourceLoader::willSendRequest): Call the navigation policy delegate only for redirects.

WebKitTools:

2009-04-22  Brady Eidson  <beidson@apple.com>

        Reviewed by Maciej Stachowiak

        Add additional support for test:
        <rdar://problem/5689748> - Cannot redirect to protocols handled by external applications.

        * DumpRenderTree/mac/PolicyDelegate.mm:
        (-[PolicyDelegate webView:unableToImplementPolicyWithError:frame:]): Dump formatted information when
          a policy cannot be implemented.

        * DumpRenderTree/win/PolicyDelegate.cpp:
        (PolicyDelegate::unableToImplementPolicyWithError): Ditto.
        * DumpRenderTree/win/PolicyDelegate.h:

LayoutTests:

2009-04-22  Brady Eidson  <beidson@apple.com>

        Reviewed by Maciej Stachowiak

        With the fix for:
        <rdar://problem/6786961> - "Are you sure you want to resubmit this form?" nag displays twice.
        And also to make sure we never rebreak:
        <rdar://problem/5689748> - Cannot redirect to protocols handled by external applications.

        * http/tests/misc/redirect-to-external-url-expected.txt: Added.
        * http/tests/misc/redirect-to-external-url.html: Added.
        * http/tests/misc/resources/redirect-to-external-url-iframe.html: Added.
        * http/tests/misc/resources/redirect-to-external-url.php: Added.

        * http/tests/misc/policy-delegate-called-twice-expected.txt: Added.
        * http/tests/misc/policy-delegate-called-twice.html: Added.
        * http/tests/misc/resources/iframe-policy-1.html: Added.
        * http/tests/misc/resources/iframe-policy-2.html: Added.



git-svn-id: svn://svn.chromium.org/blink/trunk@42772 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Rubber-stamped by Gavin Barraclough.

* jit/ExecutableAllocatorFixedVMPool.cpp:
(JSC::FixedVMPoolAllocator::addToFreeList):
(JSC::FixedVMPoolAllocator::coalesceFreeSpace):
(JSC::FixedVMPoolAllocator::FixedVMPoolAllocator):
(JSC::FixedVMPoolAllocator::alloc):
(JSC::FixedVMPoolAllocator::free):
(JSC::FixedVMPoolAllocator::isWithinVMPool):

git-svn-id: svn://svn.chromium.org/blink/trunk@42771 bbb929c8-8fbe-4397-9dbb-9b2b20218538

attempt to free memory that didn't originate from the pool, or we attempt to
hand out a bogus address from alloc.

Rubber-stamped by Gavin Barraclough.

* jit/ExecutableAllocatorFixedVMPool.cpp:
(JSC::FixedVMPoolAllocator::release):
(JSC::FixedVMPoolAllocator::FixedVMPoolAllocator):
(JSC::FixedVMPoolAllocator::alloc):
(JSC::FixedVMPoolAllocator::free):
(JSC::FixedVMPoolAllocator::isWithinVMPool):

git-svn-id: svn://svn.chromium.org/blink/trunk@42770 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Mark Rowe.

        Fix for <rdar://problem/6800695>

        Add a hack to only allow navigating (via a link from a http page)
        to feed: urls (same for feeds: and feedsearch:) that map to http:
        or https: via their nested protocol. This includes both feed://example.com
        and feed:http://example.com.

        * loader/FrameLoader.cpp:
        (WebCore::isFeedWithNestedProtocolInHTTPFamily):
        (WebCore::FrameLoader::loadFrameRequest):



git-svn-id: svn://svn.chromium.org/blink/trunk@42769 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Anders Carlsson.

        Autogenerate the XMLHttpRequest constructor getter/setter for the
        WorkerContext and remove the now no longer needed JSWorkerContextBase.lut.h
        file.

        * DerivedSources.make: Remove JSWorkerContextBase.lut.h
        * WebCore.xcodeproj/project.pbxproj: Ditto.

        * bindings/js/JSWorkerContextBase.cpp:
        * bindings/js/JSWorkerContextBase.h: Remove code to support hand written
        code for XMLHttpRequest constructor lookup and shadowing.

        * bindings/js/JSWorkerContextCustom.cpp:
        (WebCore::JSWorkerContext::customGetOwnPropertySlot): Remove JSC prefixes.
        (WebCore::JSWorkerContext::xmlHttpRequest): Added. Custom implementation of
        cached constructor lookup.
        * workers/WorkerContext.idl: Make the XMLHttpRequestConstructor accessor not
        V8 specific.



git-svn-id: svn://svn.chromium.org/blink/trunk@42768 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Update WebKitSystemInterface with new method that maps CFNetwork error code to localized description.
        
        Reviewed by Darin Adler.

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

WebKit/win:
        <rdar://problem/5115298> Don't get error msg when download is interrupted and cannot resume due to server unavailability
        If we fail to get an error description back from CFNetwork, try to get the localized description
        based on the error code.

        Reviewed by Darin Adler.

        * WebError.cpp:
        (WebError::localizedDescription):



git-svn-id: svn://svn.chromium.org/blink/trunk@42767 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        * page/Navigator.idl: Touch this file to fix build since Geolocation
        configuration was recently turned off.



git-svn-id: svn://svn.chromium.org/blink/trunk@42766 bbb929c8-8fbe-4397-9dbb-9b2b20218538

git-svn-id: svn://svn.chromium.org/blink/trunk@42765 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Reviewed by Geoff Garen.


git-svn-id: svn://svn.chromium.org/blink/trunk@42764 bbb929c8-8fbe-4397-9dbb-9b2b20218538

2009-04-22  Justin Garcia  <justin.garcia@apple.com>

        Reviewed by Darin Adler.

        https://bugs.webkit.org/show_bug.cgi?id=25322
        Can't delete empty paragraph after a block image
        
        Empty paragraph removal would have been handled by mergeParagraphs, but we stopped short
        because of some code that avoided problems that would arise from Positions that would go
        bad during deletion.  
        
        We have several checks that prevent us from using bad Positions, this one in particular
        was added to avoid bad Positions that result from text removal.  So, I removed the check
        and started correctly updating m_downstreamEnd during text removal so that it doesn't go
        bad.  m_upstreamStart doesn't need to be updated during text removal, since only text
        after it will ever be removed.

        * editing/DeleteSelectionCommand.cpp:
        (WebCore::DeleteSelectionCommand::deleteTextFromNode):
        (WebCore::DeleteSelectionCommand::handleGeneralDelete):
        (WebCore::DeleteSelectionCommand::mergeParagraphs):

LayoutTests:

2009-04-22  Justin Garcia  <justin.garcia@apple.com>

        Reviewed by Darin Adler.
        
        https://bugs.webkit.org/show_bug.cgi?id=25322
        Can't delete empty paragraph after a block image

        * editing/deleting/25322-1-expected.txt: Added.
        * editing/deleting/25322-1.html: Added.
        * editing/deleting/25322-2-expected.txt: Added.
        * editing/deleting/25322-2.html: Added.
        * editing/deleting/25322-3-expected.txt: Added.
        * editing/deleting/25322-3.html: Added.



git-svn-id: svn://svn.chromium.org/blink/trunk@42763 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        * platform/mac/svg/W3C-SVG-1.1/animate-elem-36-t-expected.txt: Fixed one
        case Eric seems to have missed earlier today of RenderImage -> RenderSVGImage.



git-svn-id: svn://svn.chromium.org/blink/trunk@42762 bbb929c8-8fbe-4397-9dbb-9b2b20218538

<rdar://problem/6757346> SAP: Prevent default on mouseDown does not stop iframe from capturing subsequent mouse moves

Reviewed by Darin Adler.

Make mouseUp target the correct frame when the original
mousedown made the drag non-capturing.


git-svn-id: svn://svn.chromium.org/blink/trunk@42761 bbb929c8-8fbe-4397-9dbb-9b2b20218538

<rdar://problem/6757346> SAP: Prevent default on mouseDown does not stop iframe from capturing subsequent mouse moves

Reviewed by Darin Adler.

This is the first step of allowing drag events to match the behaviour
of mouse move events when the default action of the initial mouse down
is prevented.  Remaining issue is that the final mouse up event still
targets the original root frame.

Test: fast/events/mouse-drag-from-frame-to-other-frame.html


git-svn-id: svn://svn.chromium.org/blink/trunk@42760 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Adele Peterson

        <rdar://problem/6702882> QT movie's video fails to be displayed after switching between tabs
        <rdar://problem/6754957> Resizing Web browser window and hitting play will cause video blocks and artifacting

        When in a media document, MediaPlayerPrivateQTKit uses a QTMovieView which may get
        layer backing under some circumstances. However, drawing the view via
        displayRectIgnoringOpacity:inContext: bypasses any layer setup that AppKit normally performs.
        So when in the media document, we draw via displayRectIgnoringOpacity:.

        * platform/graphics/mac/MediaPlayerPrivateQTKit.mm:
        (WebCore::MediaPlayerPrivate::paint):

git-svn-id: svn://svn.chromium.org/blink/trunk@42759 bbb929c8-8fbe-4397-9dbb-9b2b20218538

* fast/dom/Window/window-properties-expected.txt:
* fast/dom/navigator-detached-no-crash-expected.txt:

git-svn-id: svn://svn.chromium.org/blink/trunk@42758 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Simon Fraser.

        https://bugs.webkit.org/show_bug.cgi?id=24849

        Add tests for media 'played' attribute (missed in r42619).

        * media/video-played-expected.txt: Added.
        * media/video-played.html: Added.

git-svn-id: svn://svn.chromium.org/blink/trunk@42757 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Oliver Hunt.

        Fix <rdar://problem/6792694>
        
        When we're trying to instantiate a plug-in and the plug-in host has died, we need to invalidate the
        instance so that it doesn't stick around and do bad things.
        
        * Plugins/Hosted/NetscapePluginHostManager.mm:
        (WebKit::NetscapePluginHostManager::instantiatePlugin):



git-svn-id: svn://svn.chromium.org/blink/trunk@42756 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Rubber stamped by Sam "Blackbeard" Weinig.

        Although pirates do spell the word 'generate' as 'genertate',
        webkit developers do not.  Fixerate.

        * yarr/RegexJIT.cpp:
        (JSC::Yarr::RegexGenerator::generateAssertionBOL):
        (JSC::Yarr::RegexGenerator::generateAssertionEOL):
        (JSC::Yarr::RegexGenerator::generateAssertionWordBoundary):
        (JSC::Yarr::RegexGenerator::generatePatternCharacterSingle):
        (JSC::Yarr::RegexGenerator::generatePatternCharacterPair):
        (JSC::Yarr::RegexGenerator::generatePatternCharacterFixed):
        (JSC::Yarr::RegexGenerator::generatePatternCharacterGreedy):
        (JSC::Yarr::RegexGenerator::generatePatternCharacterNonGreedy):
        (JSC::Yarr::RegexGenerator::generateCharacterClassSingle):
        (JSC::Yarr::RegexGenerator::generateCharacterClassFixed):
        (JSC::Yarr::RegexGenerator::generateCharacterClassGreedy):
        (JSC::Yarr::RegexGenerator::generateCharacterClassNonGreedy):
        (JSC::Yarr::RegexGenerator::generateTerm):



git-svn-id: svn://svn.chromium.org/blink/trunk@42755 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Sam "Blackbeard" Weinig.

        Improvements to YARR JIT.  This patch expands support in three key areas:
            * Add (temporary) support for falling back to PCRE for expressions not supported.
            * Add support for x86_64 and Windows.
            * Add support for singly quantified parentheses (? and ??), alternatives within
              parentheses, and parenthetical assertions.

        * runtime/RegExp.cpp:
        (JSC::RegExp::match):
        * yarr/RegexJIT.cpp:
        (JSC::Yarr::RegexGenerator::storeToFrame):
        (JSC::Yarr::RegexGenerator::storeToFrameWithPatch):
        (JSC::Yarr::RegexGenerator::loadFromFrameAndJump):
        (JSC::Yarr::RegexGenerator::AlternativeBacktrackRecord::AlternativeBacktrackRecord):
        (JSC::Yarr::RegexGenerator::TermGenerationState::resetAlternative):
        (JSC::Yarr::RegexGenerator::TermGenerationState::resetTerm):
        (JSC::Yarr::RegexGenerator::TermGenerationState::jumpToBacktrack):
        (JSC::Yarr::RegexGenerator::TermGenerationState::plantJumpToBacktrackIfExists):
        (JSC::Yarr::RegexGenerator::TermGenerationState::addBacktrackJump):
        (JSC::Yarr::RegexGenerator::TermGenerationState::linkAlternativeBacktracks):
        (JSC::Yarr::RegexGenerator::TermGenerationState::propagateBacktrackingFrom):
        (JSC::Yarr::RegexGenerator::genertateAssertionBOL):
        (JSC::Yarr::RegexGenerator::genertateAssertionEOL):
        (JSC::Yarr::RegexGenerator::matchAssertionWordchar):
        (JSC::Yarr::RegexGenerator::genertateAssertionWordBoundary):
        (JSC::Yarr::RegexGenerator::genertatePatternCharacterSingle):
        (JSC::Yarr::RegexGenerator::genertatePatternCharacterPair):
        (JSC::Yarr::RegexGenerator::genertatePatternCharacterFixed):
        (JSC::Yarr::RegexGenerator::genertatePatternCharacterGreedy):
        (JSC::Yarr::RegexGenerator::genertatePatternCharacterNonGreedy):
        (JSC::Yarr::RegexGenerator::genertateCharacterClassSingle):
        (JSC::Yarr::RegexGenerator::genertateCharacterClassFixed):
        (JSC::Yarr::RegexGenerator::genertateCharacterClassGreedy):
        (JSC::Yarr::RegexGenerator::genertateCharacterClassNonGreedy):
        (JSC::Yarr::RegexGenerator::generateParenthesesDisjunction):
        (JSC::Yarr::RegexGenerator::generateParenthesesSingle):
        (JSC::Yarr::RegexGenerator::generateParentheticalAssertion):
        (JSC::Yarr::RegexGenerator::generateTerm):
        (JSC::Yarr::RegexGenerator::generateDisjunction):
        (JSC::Yarr::RegexGenerator::generateEnter):
        (JSC::Yarr::RegexGenerator::generateReturn):
        (JSC::Yarr::RegexGenerator::RegexGenerator):
        (JSC::Yarr::RegexGenerator::generate):
        (JSC::Yarr::RegexGenerator::compile):
        (JSC::Yarr::RegexGenerator::generationFailed):
        (JSC::Yarr::jitCompileRegex):
        (JSC::Yarr::executeRegex):
        * yarr/RegexJIT.h:
        (JSC::Yarr::RegexCodeBlock::RegexCodeBlock):
        (JSC::Yarr::RegexCodeBlock::~RegexCodeBlock):



git-svn-id: svn://svn.chromium.org/blink/trunk@42754 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Caused by the fix for <rdar://problem/6512520> (r42096).
        Unable to create a test case that works outside of Mail.

        <rdar://problem/6796069>

        Reviewed by Ada Chan.

        * editing/markup.cpp:
        (WebCore::createMarkup): Null check pastEnd before trying to
        call Range::compareBoundaryPoints.

git-svn-id: svn://svn.chromium.org/blink/trunk@42753 bbb929c8-8fbe-4397-9dbb-9b2b20218538

2009-04-22  Sam Weinig  <sam@webkit.org>

        Rubber-stamped by Darin Adler.

        Fix for <rdar://problem/6816957>
        Turn off Geolocation by default

        * Configurations/FeatureDefines.xcconfig:

WebCore:

2009-04-22  Sam Weinig  <sam@webkit.org>

        Rubber-stamped by Darin Adler.

        Fix for <rdar://problem/6816957>
        Turn off Geolocation by default

        * Configurations/FeatureDefines.xcconfig:

WebKit/mac:

2009-04-22  Sam Weinig  <sam@webkit.org>

        Rubber-stamped by Darin Adler.

        Fix for <rdar://problem/6816957>
        Turn off Geolocation by default

        * Configurations/FeatureDefines.xcconfig:

WebKitTools:

2009-04-22  Sam Weinig  <sam@webkit.org>

        Rubber-stamped by Darin Adler.

        Fix for <rdar://problem/6816957>
        Turn off Geolocation by default

        * Scripts/build-webkit:



git-svn-id: svn://svn.chromium.org/blink/trunk@42752 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Change RenderSVGImage to dump as RenderSVGImage instead of RenderImage

        * rendering/RenderSVGImage.h:
        (WebCore::RenderSVGImage::renderName):

git-svn-id: svn://svn.chromium.org/blink/trunk@42751 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        it was never added to the pending map in the first place.

        <rdar://problem/6814144>

        Reviewed by Darin Adler.

        * page/DOMWindow.cpp:
        (WebCore::allowsPendingBeforeUnloadListeners): Renamed from shouldAddPendingBeforeUnloadListener.
        (WebCore::DOMWindow::addEventListener): Call allowsPendingBeforeUnloadListeners now.
        (WebCore::DOMWindow::removeEventListener): Call allowsPendingBeforeUnloadListeners before
        removing the before unload listener.
        (WebCore::DOMWindow::clearAttributeEventListener): Ditto.

git-svn-id: svn://svn.chromium.org/blink/trunk@42750 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Ariya Hidayat.

        Made windowed plugins move/resize synchronized with the painting, so
        that windowed plugins are not moved before the rest of the parent
        frame during scrolling.

        * plugins/qt/PluginContainerQt.cpp:
        (PluginContainerQt::PluginContainerQt):
        (PluginContainerQt::requestGeometry):
        (PluginContainerQt::adjustGeometry):
        * plugins/qt/PluginContainerQt.h:
        * plugins/qt/PluginViewQt.cpp:
        (WebCore::PluginView::updatePluginWidget):
        (WebCore::PluginView::paint):

git-svn-id: svn://svn.chromium.org/blink/trunk@42749 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Ariya Hidayat.

        Fix various bugs in the X11 Qt plugin focus handling.

        Notify the focus controller on activation/deactivation of the plugin and
        set focus on the plugin's element when it receives the window system
        focus.

git-svn-id: svn://svn.chromium.org/blink/trunk@42748 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Ariya Hidayat.

        https://bugs.webkit.org/show_bug.cgi?id=25023
        Delete reply in QNetworkReplyHandler::abort() to avoid leak.

        * platform/network/qt/QNetworkReplyHandler.cpp:
        (WebCore::QNetworkReplyHandler::abort):

git-svn-id: svn://svn.chromium.org/blink/trunk@42747 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Allow the UI delegate to control cursor-setting.

        Reviewed by Ada Chan.

        WebCore/:
        * page/Chrome.cpp:
        (WebCore::Chrome::setCursor):
        Pass the call to the client.

        * page/Chrome.h:

        * page/ChromeClient.h:
        (WebCore::ChromeClient::setCursor):
        Stubbed; returns false.

        * platform/Cursor.h:
        Define PlatformCursorHandle for each platform.

        * platform/win/WidgetWin.cpp:
        Reordered some includes. Added an include of Page.h.
        (WebCore::Widget::setCursor):
        Fixed a typo in the comment about ignoreNextSetCursor. Get the Page, and
        call its Chrome's setCursor() method with the passed cursor. Fall back
        to SetCursor() if the Page is not available.

        * plugins/win/PluginViewWin.cpp:
        (WebCore::PluginView::handleMouseEvent):
        Remove an unused variable. Cleaned up some whitespace.

        WebKit/win/:
        * Interfaces/IWebUIDelegatePrivate.idl:
        Added a function to the IWebUIDelegatePrivate5 interface.

        * WebCoreSupport/WebChromeClient.cpp:
        (WebChromeClient::setCursor):
        Try to call the UI delegate's webViewSetCursor() method. If it fails,
        fall back to calling SetCursor() with the passed cursor.

        * WebCoreSupport/WebChromeClient.h:

        * WebView.cpp:
        (WebViewWndProc):
        Call the Chrome's setCursor() method with the last set cursor.

git-svn-id: svn://svn.chromium.org/blink/trunk@42746 bbb929c8-8fbe-4397-9dbb-9b2b20218538

git-svn-id: svn://svn.chromium.org/blink/trunk@42745 bbb929c8-8fbe-4397-9dbb-9b2b20218538

git-svn-id: svn://svn.chromium.org/blink/trunk@42744 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Reviewed by Maciej Stachowiak

Apply our vm reentry caching logic to String.replace with global
regexes.


git-svn-id: svn://svn.chromium.org/blink/trunk@42743 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        * fast/encoding/mailto-always-utf-8-expected.txt:



git-svn-id: svn://svn.chromium.org/blink/trunk@42742 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Jon Honeycutt.

        - Mac part of fixing for <rdar://problem/6755137> Action dictionary for
          policy decision is missing keys when full-page zoom is used

        * WebCoreSupport/WebFrameLoaderClient.mm:
        (WebFrameLoaderClient::actionDictionary): Use absoluteLocation() instead
        of pageX() and pageY(), which are adjusted for zoom.

WebKit/win:

        Reviewed by Jon Honeycutt.

        - Windows part of fixing <rdar://problem/6755137> Action dictionary for
          policy decision is missing keys when full-page zoom is used

        * WebActionPropertyBag.cpp:
        (WebActionPropertyBag::Read): Use absoluteLocation() instead of
        clientX() and clientY(), which are adjusted for zoom.

LayoutTests:

        - test for <rdar://problem/6755137> Action dictionary for policy
          decision is missing keys when full-page zoom is used

        * fast/loader/policy-delegate-action-hit-test-zoomed-expected.txt: Added.
        * fast/loader/policy-delegate-action-hit-test-zoomed.html: Added.



git-svn-id: svn://svn.chromium.org/blink/trunk@42741 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Jon Honeycutt.

        - implement DOMNode::nodeName()

        * DOMCoreClasses.cpp:
        (DOMNode::nodeName):

WebKitTools:

        Reviewed by Jon Honeycutt.

        - dump the node in the element dictionary in the action information
          passed to the policy delegate

        * DumpRenderTree/mac/PolicyDelegate.mm:
        (-[PolicyDelegate webView:decidePolicyForNavigationAction:request:frame:decisionListener:]):
        * DumpRenderTree/win/PolicyDelegate.cpp:
        (dumpPath):
        (PolicyDelegate::decidePolicyForNavigationAction):

LayoutTests:

        Reviewed by Jon Honeycutt.

        - updated result that shows the node in the element dictionary in the
          action information passed to the policy delegate

        * http/tests/security/feed-urls-from-remote-expected.txt:



git-svn-id: svn://svn.chromium.org/blink/trunk@42740 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Dimitri Glazkov.

        Add a method to purge the Chromium plugin list cache.
        https://bugs.webkit.org/show_bug.cgi?id=25318

        * plugins/chromium/PluginDataChromium.cpp:
        (WebCore::resetChromiumPluginCache):

git-svn-id: svn://svn.chromium.org/blink/trunk@42739 bbb929c8-8fbe-4397-9dbb-9b2b20218538

Reviewed by Darin Adler.

We need to ensure that XMLTokenizer sets the current DocLoader before calling in to
any libxml2 methods that may trigger a load.  The presence of a DocLoader indicates
that the load was originated by WebCore's use of libxml2 and that we should enforce
the same-origin policy on it.  XMLTokenizer::initializeParserContext,
XMLTokenizer::doWrite and XMLTokenizer::doEnd were three methods that were not setting
the current DocLoader when they should have.

The XMLTokenizerScope class is introduced to simplify the pattern of saving, setting and
restoring the current DocLoader and libxml2 error handlers.  The DocLoader and error handlers
are saved and set when the scope is allocated, and restored to their previous values when
the scope is exited.

Test: http/tests/security/xss-DENIED-xml-external-entity.xhtml

* GNUmakefile.am:
* WebCore.pro:
* WebCore.scons:
* WebCore.vcproj/WebCore.vcproj:
* WebCore.xcodeproj/project.pbxproj:
* dom/XMLTokenizer.cpp:
* dom/XMLTokenizer.h:
* dom/XMLTokenizerLibxml2.cpp:
(WebCore::matchFunc):
(WebCore::shouldAllowExternalLoad):
(WebCore::openFunc):
(WebCore::XMLTokenizer::doWrite):
(WebCore::XMLTokenizer::initializeParserContext):
(WebCore::XMLTokenizer::doEnd):
(WebCore::xmlDocPtrForString):
* dom/XMLTokenizerScope.cpp:
(WebCore::XMLTokenizerScope::XMLTokenizerScope):
(WebCore::XMLTokenizerScope::~XMLTokenizerScope):
* dom/XMLTokenizerScope.h:
* xml/XSLStyleSheet.cpp:
(WebCore::XSLStyleSheet::parseString):
* xml/XSLTProcessor.cpp:

git-svn-id: svn://svn.chromium.org/blink/trunk@42738 bbb929c8-8fbe-4397-9dbb-9b2b20218538

2009-04-21  Sam Weinig  <sam@webkit.org>

        Reviewed by Dan Bernstein.

        Fix for <rdar://problem/5874009>
        Add port 3659 (apple-sasl / PasswordServer) to the list of blocked ports.

        * platform/network/ResourceHandle.cpp:
        (WebCore::portAllowed):

LayoutTests:

2009-04-21  Sam Weinig  <sam@webkit.org>

        Reviewed by Dan Bernstein.

        Update test and results for <rdar://problem/5874009>
        Add port 3659 (apple-sasl / PasswordServer) to the list of blocked ports.

        * platform/mac-tiger/security/block-test-expected.txt:
        * platform/mac/security/block-test-expected.txt:
        * security/block-test.html:



git-svn-id: svn://svn.chromium.org/blink/trunk@42737 bbb929c8-8fbe-4397-9dbb-9b2b20218538

git-svn-id: svn://svn.chromium.org/blink/trunk@42736 bbb929c8-8fbe-4397-9dbb-9b2b20218538

2009-04-21  Justin Garcia  <justin.garcia@apple.com>

        Reviewed by Dan Bernstein.

        https://bugs.webkit.org/show_bug.cgi?id=25320
        Crash when changing block styles in a root that contains no visible content (but contains nodes with renderers)

        * editing/CompositeEditCommand.cpp:
        (WebCore::CompositeEditCommand::moveParagraphContentsToNewBlockIfNecessary): If the enclosing block is the 
        root editable element and it contains no visible content, create a new block but don't try and move 
        content into it, since there's nothing for moveParagraphs to move.

LayoutTests:

2009-04-21  Justin Garcia  <justin.garcia@apple.com>
        
        Reviewed by Dan Bernstein.

        https://bugs.webkit.org/show_bug.cgi?id=25320
        Crash when changing block styles in a root that contains no visible content (but contains nodes with renderers)

        * editing/execCommand/25320-expected.txt: Added.
        * editing/execCommand/25320.html: Added.
        * platform/mac/fast/forms/caret-rtl-expected.checksum: Replaced.
        * platform/mac/fast/forms/caret-rtl-expected.txt: Replaced.



git-svn-id: svn://svn.chromium.org/blink/trunk@42735 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Reviewed by Cameron Zwarich and Oliver Hunt.
        
        Re-Fixed <rdar://problem/6406045> REGRESSION: Stack overflow on PowerPC on
        fast/workers/use-machine-stack.html (22531)
        
        SunSpider reports no change.
        
        Use a larger recursion limit on the main thread (because we can, and
        there's some evidence that it may improve compatibility), and a smaller
        recursion limit on secondary threads (because they tend to have smaller
        stacks).

        * interpreter/Interpreter.cpp:
        (JSC::Interpreter::execute):
        (JSC::Interpreter::prepareForRepeatCall):
        * interpreter/Interpreter.h:
        (JSC::): Ditto. I wrote the recursion test slightly funny, so that the
        common case remains a simple compare to constant.

        * runtime/ArrayPrototype.cpp:
        (JSC::arrayProtoFuncToString):
        (JSC::arrayProtoFuncToLocaleString):
        (JSC::arrayProtoFuncJoin): Conservatively, set the array recursion limits
        to the lower, secondary thread limit. We can do something fancier if
        compatibility moves us, but this seems sufficient for now.



git-svn-id: svn://svn.chromium.org/blink/trunk@42734 bbb929c8-8fbe-4397-9dbb-9b2b20218538

        Minor clean-ups to WebKitCreateInstance change.
        
        Reviewed by Jon Honeycutt.

        * WebKitCOMAPI.cpp:
        (classFactory):
        (WebKitCreateInstance):



git-svn-id: svn://svn.chromium.org/blink/trunk@42733 bbb929c8-8fbe-4397-9dbb-9b2b20218538