1. 01 Dec, 2009 9 commits
  2. 30 Nov, 2009 10 commits
    • enrica@apple.com's avatar
      WebCore: Can focus but not type into content editable block that contains only... · 429c6e2b
      enrica@apple.com authored
      WebCore: Can focus but not type into content editable block that contains only non-editable content.
      <rdar://problem/5982901>
      https://bugs.webkit.org/show_bug.cgi?id=31750
              
      Reviewed by Darin Adler.
      
      The goal is to change the way we choose a visible position
      after hit detection, by preferring a visually equivalent editable
      position if available. By doing this, it is possible to add content
      to an editable block that initially contains only non editable elements.
      
      Test: editing/selection/mixed-editability-10.html
      
      * WebCore.base.exp: Changed to match the new signature of downstream
      and upstream in the Position class.
      * dom/Position.cpp:
      (WebCore::Position::atEditingBoundary): Added.
      (WebCore::Position::upstream): Modified to allow to cross the boundary
      between editable and non editable content if required.
      (WebCore::Position::downstream): Modified to allow to cross the boundary
      between editable and non editable content if required.
      (WebCore::Position::isCandidate): Modified to qualify as candidates positions
      that are at the editability boundary.
      (WebCore::Position::getInlineBoxAndOffset): Modified to retrieve the inline box
      to be used in calculating the caret rectangle.
      * dom/Position.h:
      (WebCore::Position::):
      * dom/PositionIterator.cpp:
      (WebCore::PositionIterator::atEditingBoundary): Added.
      (WebCore::PositionIterator::isCandidate): Modified to qualify as candidates positions
      that are at the editability boundary.
      * dom/PositionIterator.h:
      * editing/htmlediting.cpp:
      (WebCore::firstEditablePositionAfterPositionInRoot): Modified to accept not only
      descendants of the editable container, but the container itself.
      (WebCore::lastEditablePositionBeforePositionInRoot): Modified to accept not only
      descendants of the editable container, but the container itself.
      * rendering/RenderObject.cpp:
      (WebCore::RenderObject::createVisiblePosition): Added logic to prefer an editable position,
      if available.
      * rendering/RenderText.cpp:
      (WebCore::RenderText::isAllCollapsibleWhitespace): Added.
      * rendering/RenderText.h:
      
      LayoutTests: Can focus but not type into content editable block that contains only non-editable content.
      <rdar://problem/5982901>
      https://bugs.webkit.org/show_bug.cgi?id=31750
      
      Reviewed by Darin Adler.
      
      * editing/selection/5825350-1-expected.txt:
      * editing/selection/5825350-1.html: Modified to use caret rectangle.
      * editing/selection/5825350-2-expected.txt:
      * editing/selection/5825350-2.html: Modified to use caret rectangle.
      * editing/selection/mixed-editability-10-expected.txt: Added.
      * editing/selection/mixed-editability-10.html: Added.
      * platform/mac/editing/deleting/5390681-2-expected.txt: Re-baselined
      to account for the new possible caret position.
      
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@51522 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      429c6e2b
    • bdakin@apple.com's avatar
      WebCore: Fix for https://bugs.webkit.org/show_bug.cgi?id=32000 Crash in · 70c2e451
      bdakin@apple.com authored
      Safari caused by extreme column-gap and column-width values 
      -and corresponding-
      <rdar://problem/7425433>
      
      Reviewed by Oliver Hunt.
      
      Prevent desiredColumnCount from being less than 1 since it is used 
      as a divisor.
      * rendering/RenderBlock.cpp:
      (WebCore::RenderBlock::calcColumnWidth):
      
      LayoutTests: Test for https://bugs.webkit.org/show_bug.cgi?id=32000 Crash in 
      Safari caused by extreme column-gap and column-width values 
      -and corresponding-
      <rdar://problem/7425433>
      
      Reviewed by Oliver Hunt.
      
      * fast/multicol/huge-column-gap-crash-expected.txt: Added.
      * fast/multicol/huge-column-gap-crash.html: Added.
      
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@51517 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      70c2e451
    • ap@apple.com's avatar
      Reviewed by Oliver Hunt. · 53c40236
      ap@apple.com authored
              https://bugs.webkit.org/show_bug.cgi?id=31659
              Connection must be closed in case of Web Socket handshake error
      
              The network connection was closed, but the close event wasn't dispatched.
      
              Tested by websocket/tests/handshake-error.html, which is un-skipped now.
      
              * platform/network/cf/SocketStreamHandleCFNet.cpp: (WebCore::SocketStreamHandle::platformClose):
              Call client didHandle() method. This looks strange in CFNetwork implementation, because
              it's the client that asked to close the stream, so it shouldn't need the callback. It makes
              more sense in cross-process implementation, because closing is async there.
      
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@51516 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      53c40236
    • eric.carlson@apple.com's avatar
      2009-11-30 Eric Carlson <eric.carlson@apple.com> · 8fec90f7
      eric.carlson@apple.com authored
              Reviewed by Simon Fraser.
      
              All HTML5 media element events should be regular events
              https://bugs.webkit.org/show_bug.cgi?id=30513
      
              * html/HTMLMediaElement.cpp:
              (WebCore::HTMLMediaElement::selectMediaResource):
              (WebCore::HTMLMediaElement::noneSupported):
              (WebCore::HTMLMediaElement::mediaEngineError):
              (WebCore::HTMLMediaElement::setNetworkState):
              (WebCore::HTMLMediaElement::userCancelledLoad):
                  Call scheduleEvent instead of scheduleProgressEvent.
              (WebCore::HTMLMediaElement::progressEventTimerFired):
                  Call scheduleEvent instead of scheduleProgressEvent. Call renderer->updateFromElement
                  after scheduling a 'progress' event so the controller will update download
                  progress indicator.
              * html/HTMLMediaElement.h:
                  Remove scheduleProgressEvent prototype.
      
      2009-11-30  Eric Carlson  <eric.carlson@apple.com>
      
              Reviewed by Simon Fraser.
      
              All HTML5 media element events should be regular events
              https://bugs.webkit.org/show_bug.cgi?id=30513
              
              Remove progress event tests.
      
              * media/progress-event-at-least-one-expected.txt: Removed.
              * media/progress-event-at-least-one.html: Removed.
              * media/progress-event-expected.txt: Removed.
              * media/progress-event-total-expected.txt: Removed.
              * media/progress-event-total.html: Removed.
              * media/progress-event.html: Removed.
      
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@51511 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      8fec90f7
    • ap@apple.com's avatar
      It was me who reviewed the patch, not Eric! Alas, some tools we use cause... · 6d23a697
      ap@apple.com authored
      It was me who reviewed the patch, not Eric! Alas, some tools we use cause wrong reviewer information in Bugzilla.
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@51509 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      6d23a697
    • ap@apple.com's avatar
      2009-11-30 Yuzo Fujishima <yuzo@google.com> · 87995de7
      ap@apple.com authored
              Reviewed by Eric Seidel.
      
              In testing handshake-error case, web_socket_do_extra_handshake calls a non-existent method
              https://bugs.webkit.org/show_bug.cgi?id=31465
      
              * platform/mac/Skipped: websocket/tests/handshake-error.html now fails on Mac, added to
              Skipped list, and filed bug 31659 to investigate.
      
              * websocket/tests/handshake-error_wsh.py: Call write() on a correct object.
      
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@51508 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      87995de7
    • joepeck@webkit.org's avatar
      2009-11-30 Joseph Pecoraro <joepeck@webkit.org> · 723f16eb
      joepeck@webkit.org authored
              Reviewed by Pavel Feldman.
      
              Web Inspector: Wrong console output for Regexp escape sequence
              https://bugs.webkit.org/show_bug.cgi?id=31538
      
              Updated inspector/console-format.html
      
              * inspector/front-end/ConsoleView.js:
              (WebInspector.ConsoleView.createDividerElement): style issues.
              (WebInspector.ConsoleView.createFilterElement): style issues.
              (WebInspector.ConsoleView): added _customFormatters table.
              (WebInspector.ConsoleView.prototype.updateMessageRepeatCount): style isses.
              (WebInspector.ConsoleView.prototype._incrementErrorWarningCount): style issues.
              (WebInspector.ConsoleView.prototype._format): simplified delegation to formatter logic.
              (WebInspector.ConsoleView.prototype._formatobject):
              (WebInspector.ConsoleView.prototype._formatnode):
              (WebInspector.ConsoleView.prototype._printArray):
              (WebInspector.ConsoleMessage.prototype._format): commented and broke down the algorithm into parts.
              (WebInspector.ConsoleMessage.prototype._formatWithSubstitutionString.append): handle substitution string formatting.
              (WebInspector.ConsoleMessage.prototype._formatIndividualValue): handling individual value formatting.
              (WebInspector.ConsoleCommandResult):
              * inspector/front-end/InjectedScript.js: simplified regex formatting.
              * inspector/front-end/inspector.js:
              (WebInspector.linkifyStringAsFragment): converted new RegExp to literal for performance benefits.
              * inspector/front-end/utilities.js:
              (Element.prototype.hasStyleClass): update inaccurate comment.
              (String.prototype.trimURL): converted new RegExp to literal for performance benefits.
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@51506 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      723f16eb
    • kov@webkit.org's avatar
      Skipping test that fails because of broken DND, and some tests · f4d1ccff
      kov@webkit.org authored
      requiring new results.
      
      * platform/gtk/Skipped:
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@51504 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      f4d1ccff
    • eric@webkit.org's avatar
      2009-11-30 Dirk Schulze <krit@webkit.org> · 5b0c5989
      eric@webkit.org authored
              Reviewed by Nikolas Zimmermann.
      
              Complex test of mask with every combination of maskUnits and maskContentUnits
              as well as different values for size and position of the mask and it's content.
      
              * platform/mac/svg/custom/mask-with-all-units-expected.checksum: Added.
              * platform/mac/svg/custom/mask-with-all-units-expected.png: Added.
              * platform/mac/svg/custom/mask-with-all-units-expected.txt: Added.
              * svg/custom/mask-with-all-units.svg: Added.
      2009-11-30  Dirk Schulze  <krit@webkit.org>
      
              Reviewed by Nikolas Zimmermann.
      
              REGRESSION (r49757): masking-mask-01-b.svg rendered incorrectly
              [https://bugs.webkit.org/show_bug.cgi?id=31980]
      
              It turns out, that we did not handle maskUnits="userSpaceOnUse"
              correctly. We just need to move the context of the maskImage
              if maskContentUnits="objectBoundingBox". The context gets scaled
              on this unit combination, so we have to substract the position of
              the targetRect. On userSpaceOnUse the position just depends on the
              position of the mask element.
              I added a test with some senseless and reasonable values for size
              and postion of the mask and it's contents. I also used every
              combination of maskUnits and maskContentUnits to be sure, that it
              is fixed this time.
      
              Test: svg/custom/mask-with-all-units.svg
      
              * svg/SVGMaskElement.cpp:
              (WebCore::SVGMaskElement::drawMaskerContent):
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@51496 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      5b0c5989
    • ossy@webkit.org's avatar
      [Qt] Put tests into skiplist because of missing features. · 1fd3cff0
      ossy@webkit.org authored
      * platform/qt/Skipped:
      - fast/dom/Window/window-postmessage-clone.html added
      - inspector/elements-img-tooltip.html added.
      
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@51482 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      1fd3cff0
  3. 29 Nov, 2009 3 commits
  4. 28 Nov, 2009 1 commit
  5. 27 Nov, 2009 3 commits
    • dbates@webkit.org's avatar
      2009-11-27 Daniel Bates <dbates@webkit.org> · 6b86d506
      dbates@webkit.org authored
              Reviewed by Adam Barth.
      
              https://bugs.webkit.org/show_bug.cgi?id=31940
      
              Makes the error messages more descriptive when we refuse to load an object/embed or
              refuse to load from the document base URL.
      
              * page/XSSAuditor.cpp:
              (WebCore::XSSAuditor::canLoadObject): Changed console message to be more descriptive.
              (WebCore::XSSAuditor::canSetBaseElementURL): Ditto.
      2009-11-27  Daniel Bates  <dbates@webkit.org>
      
              Reviewed by Adam Barth.
      
              https://bugs.webkit.org/show_bug.cgi?id=31940
      
              Rebased results because of new error messages.
      
              * http/tests/security/xssAuditor/base-href-control-char-expected.txt:
              * http/tests/security/xssAuditor/base-href-expected.txt:
              * http/tests/security/xssAuditor/base-href-null-char-expected.txt:
              * http/tests/security/xssAuditor/base-href-scheme-relative-expected.txt:
              * http/tests/security/xssAuditor/embed-tag-control-char-expected.txt:
              * http/tests/security/xssAuditor/embed-tag-expected.txt:
              * http/tests/security/xssAuditor/embed-tag-javascript-url-expected.txt:
              * http/tests/security/xssAuditor/embed-tag-null-char-expected.txt:
              * http/tests/security/xssAuditor/object-embed-tag-control-char-expected.txt:
              * http/tests/security/xssAuditor/object-embed-tag-expected.txt:
              * http/tests/security/xssAuditor/object-embed-tag-null-char-expected.txt:
              * http/tests/security/xssAuditor/object-src-inject-expected.txt:
              * http/tests/security/xssAuditor/object-tag-expected.txt:
              * http/tests/security/xssAuditor/object-tag-javascript-url-expected.txt:
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@51445 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      6b86d506
    • eric@webkit.org's avatar
      2009-11-27 Kenneth Russell <kbr@google.com> · 26865516
      eric@webkit.org authored
              Reviewed by Eric Seidel.
      
              REGRESSION(51418): fast/canvas/webgl/gl-object-get-calls.html is failing on the Leopard commit bot
              https://bugs.webkit.org/show_bug.cgi?id=31919
      
              Temporarily disabled portion of test which is failing on some
              machines but not others. Root cause will be investigated under
              https://bugs.webkit.org/show_bug.cgi?id=31843 .
      
              * fast/canvas/webgl/gl-object-get-calls-expected.txt:
              * fast/canvas/webgl/script-tests/gl-object-get-calls.js:
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@51441 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      26865516
    • hamaji@chromium.org's avatar
      2009-11-26 Shinichiro Hamaji <hamaji@chromium.org> · 63602224
      hamaji@chromium.org authored
              Unreviewed red bot fix.
      
              Skip tests introduced by the previous change as it seems to be failing.
      
              * platform/qt/Skipped:
              * platform/win/Skipped:
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@51430 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      63602224
  6. 26 Nov, 2009 4 commits
    • hamaji@chromium.org's avatar
      2009-11-26 Shinichiro Hamaji <hamaji@chromium.org> · 46d70cfa
      hamaji@chromium.org authored
              Reviewed by Dan Bernstein.
      
              Assertion failure in RenderBlock::positionForPointWithInlineChildren when running fast/inline/relative-positioned-overflow.html
              https://bugs.webkit.org/show_bug.cgi?id=29966
      
              A test case where a user drags the mouse from an empty inline
              element to above texts.
      
              * editing/selection/last-empty-inline-expected.txt: Added.
              * editing/selection/last-empty-inline.html: Added.
              * platform/win/Skipped:
      2009-11-26  Shinichiro Hamaji  <hamaji@chromium.org>
      
              Reviewed by Dan Bernstein.
      
              Assertion failure in RenderBlock::positionForPointWithInlineChildren when running fast/inline/relative-positioned-overflow.html
              https://bugs.webkit.org/show_bug.cgi?id=29966
      
              When an empty inline element is clicked, the root inline box has
              no leaf children. Use the renderer of a normal child instead.
      
              This change resolves Windows port's assertion failure in a layout test.
              Also, this fixes the behavior when a user drags the mouse from an
              empty inline element to above texts.
      
              Test: editing/selection/last-empty-inline.html
      
              * rendering/RenderBlock.cpp:
              (WebCore::RenderBlock::positionForPointWithInlineChildren):
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@51429 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      46d70cfa
    • eric@webkit.org's avatar
      2009-11-26 Kinuko Yasuda <kinuko@chromium.com> · 782bba06
      eric@webkit.org authored
              Reviewed by Eric Seidel.
      
              Rewrite the test fast/parser/comment-in-textarea to use dumpAsText
              so that the expected results can be independent of platform differences.
              https://bugs.webkit.org/show_bug.cgi?id=31580
      
              * fast/parser/comment-in-textarea-expected.txt: Added.
              * fast/parser/comment-in-textarea.html:
              * fast/parser/script-tests/comment-in-textarea.js: Added.
              * platform/mac/fast/parser/comment-in-textarea-expected.checksum: Removed.
              * platform/mac/fast/parser/comment-in-textarea-expected.png: Removed.
              * platform/mac/fast/parser/comment-in-textarea-expected.txt: Removed.
              * platform/win/fast/parser/comment-in-textarea-expected.txt: Removed.
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@51427 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      782bba06
    • eric@webkit.org's avatar
      2009-11-26 Daniel Bates <dbates@webkit.org> · 650b16f3
      eric@webkit.org authored
              Reviewed by Pavel Feldman.
      
              https://bugs.webkit.org/show_bug.cgi?id=21554
      
              Tests that the tooltip text for an image is its dimensions.
      
              * inspector/elements-img-tooltip-expected.txt: Added.
              * inspector/elements-img-tooltip.html: Added.
      2009-11-26  Daniel Bates  <dbates@webkit.org>
      
              Reviewed by Pavel Feldman.
      
              https://bugs.webkit.org/show_bug.cgi?id=21554
      
              Implements support for hovering over <img> src to display the height and width of that image
              in a tooltip. Displays both the displayable and natural dimensions of the image.
      
              Test: inspector/elements-img-tooltip.html
      
              * inspector/front-end/ElementsTreeOutline.js:
              (WebInspector.ElementsTreeElement.prototype.createTooltipForImageNode): Added.
              (WebInspector.ElementsTreeElement.prototype._updateTitle.callback):
              (WebInspector.ElementsTreeElement.prototype._updateTitle):
              (WebInspector.ElementsTreeElement.prototype._nodeTitleInfo):
              * inspector/front-end/ObjectProxy.js:
              (WebInspector.ObjectProxy.getPropertiesAsync): Added.
              * inspector/front-end/inspector.js:
              (WebInspector.linkifyURLAsNode): Added tooltipText argument.
              (WebInspector.linkifyURL): Ditto.
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@51420 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      650b16f3
    • zecke@webkit.org's avatar
      [Qt] Add automatic test case for plugins that use QWidget · 87659e2a
      zecke@webkit.org authored
      https://bugs.webkit.org/show_bug.cgi?id=31203
      
      Change the RenderTreeAsText to look into the WebCore::Widget
      and print the state of the platformWidget. Change the DRT to
      use QtUiTools to be able to initiate QProgressBar and such
      as a plugin. Add the test case, add the result, update the
      other results due the new test.
      
      WebCore:
      * rendering/RenderTreeAsText.cpp:
      (WebCore::operator<<): Add special case for RenderPart
      
      WebKitTools:
      * DumpRenderTree/qt/DumpRenderTree.cpp:
      (WebCore::WebPage::createPlugin):
      * DumpRenderTree/qt/DumpRenderTree.h:
      * DumpRenderTree/qt/DumpRenderTree.pro:
      
      LayoutTest:
      * platform/gtk/Skipped:
      * platform/mac/Skipped:
      * platform/qt/plugins/netscape-dom-access-expected.txt: Added
      * platform/qt/plugins/qt-qwidget-plugin-expected.txt: Added
      * platform/win/Skipped:
      * plugins/qt-qwidget-plugin.html: Added.
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@51410 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      87659e2a
  7. 25 Nov, 2009 7 commits
  8. 24 Nov, 2009 3 commits
    • cmarrin@apple.com's avatar
      Implement accelerated compositing · 93972964
      cmarrin@apple.com authored
      https://bugs.webkit.org/show_bug.cgi?id=27314
      
      Initial implementation, but disabled in this checkin.
      To enable ENABLE_3D_RENDERING and USE_ACCELERATED_COMPOSITING
      must be turned on. Also d3d9.lib and d3d9x.lib must be added
      to the build.
      
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@51371 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      93972964
    • cmarrin@apple.com's avatar
    • darin@chromium.org's avatar
      2009-11-24 Darin Fisher <darin@chromium.org> · acc8f482
      darin@chromium.org authored
              Reviewed by Adam Barth.
      
              [Chromium] Renderer hang when using www.expedia.com
              https://bugs.webkit.org/show_bug.cgi?id=31822
      
              Test: http/tests/cache/subresource-failover-to-network.html
      
              * loader/FrameLoader.cpp:
              (WebCore::FrameLoader::loadResourceSynchronously): Use the originalRequest
              when inheriting cache policy.  This matches SubresourceLoader::create.
      
      2009-11-24  Darin Fisher  <darin@chromium.org>
      
              Reviewed by Adam Barth.
      
              [Chromium] Renderer hang when using www.expedia.com
              https://bugs.webkit.org/show_bug.cgi?id=31822
      
              This test verifies that a synchronous XMLHttpRequest, generated from a
              page that is the result of a form submission, loads properly when the
              user navigates back to the page.
      
              When navigating back to a page that resulted from a form submission,
              the page is loaded with the ReturnCacheDataDontLoad cache policy.  It
              is important that subresources (including XMLHttpRequest instances) do
              not inherit this cache policy.
      
              * http/tests/cache/resources/echo-no-store.cgi: Added.
              * http/tests/cache/resources/subresource-failover-to-network.cgi: Added.
              * http/tests/cache/subresource-failover-to-network-expected.txt: Added.
              * http/tests/cache/subresource-failover-to-network.html: Added.
      
      2009-11-24  Darin Fisher  <darin@chromium.org>
      
              Reviewed by Adam Barth.
      
              [Chromium] Renderer hang when using www.expedia.com
              https://bugs.webkit.org/show_bug.cgi?id=31822
      
              * src/FrameLoaderClientImpl.cpp:
              (WebKit::FrameLoaderClientImpl::dispatchWillSendRequest):
              Assign ReturnCacheDataDontLoad cache policy at the last moment
              to match Safari.  This way the cache policy does not get stored
              in DocumentLoader's originalRequest preventing it from being
              inherited by subresource requests.
              * src/ResourceHandle.cpp:
              (WebCore::ResourceHandle::willLoadFromCache): Remove assignment
              of ReturnCacheDataDontLoad cache policy so that it doesn't end
              up applying to the DocumentLoader's originalRequest.
      
      
      
      git-svn-id: http://svn.webkit.org/repository/webkit/trunk@51350 268f45cc-cd09-0410-ab3c-d52691b4dbfc
      acc8f482