2010-09-09  Jer Noble  <jer.noble@apple.com>

        Reviewed by Eric Carlson.

        Adopt shared control drawing for <video> controls on Windows
        https://bugs.webkit.org/show_bug.cgi?id=45490

        * win/lib/WebKitSystemInterface.lib:
        * win/lib/WebKitSystemInterface_debug.lib:
2010-09-09  Jer Noble  <jer.noble@apple.com>

        Reviewed by Eric Carlson.

        Adopt shared control drawing for <video> controls on Windows
        https://bugs.webkit.org/show_bug.cgi?id=45490

        * DerivedSources.make: Remove the OS check around mediaControlsQuickTime.css.
        * rendering/RenderMediaControls.cpp:
        (WebCore::determineState): Use WKMediaControllerThemeState flags instead of SafariTheme ones.
        (WebCore::getUnzoomedRectAndAdjustCurrentContext): Added. Copied verbatim from RenderMediaMac.mm.
        (WebCore::RenderMediaControls::adjustMediaSliderThumbSize): Use wkMeasureMediaUIPart to adjust the thumb size.
        (WebCore::RenderMediaControls::paintMediaControlsPart): Use wkDrawMediaUIPart instead of paintThemePart.
        (WebCore::RenderMediaControls::volumeSliderOffsetFromMuteButton): Added. Copied verbatim from RenderMediaMac.mm.
        * rendering/RenderMediaControls.h:
        * rendering/RenderThemeWin.cpp:
        (WebCore::RenderThemeWin::adjustSliderThumbSize): Adjust both sliders and volume sliders.
        (WebCore::RenderThemeWin::extraMediaControlsStyleSheet): Added. Returns the mediaControlsQuickTime sheet.
        (WebCore::RenderThemeWin::paintMediaRewindButton): Added. Pass through to RenderMediaControls.
        (WebCore::RenderThemeWin::paintMediaControlsBackground): Added. Pass through to RenderMediaControls.
        (WebCore::RenderThemeWin::paintMediaVolumeSliderContainer): Added. Pass through to RenderMediaControls.
        (WebCore::RenderThemeWin::paintMediaVolumeSliderTrack): Added. Pass through to RenderMediaControls.
        (WebCore::RenderThemeWin::paintMediaVolumeSliderThumb): Added.  Pass through to RenderMediaControls.
        (WebCore::RenderThemeWin::volumeSliderOffsetFromMuteButton): Added. Pass through to RenderMediaControls.
        * rendering/RenderThemeWin.h:

2010-09-109  Peter Kasting  <pkasting@google.com>

        Reviewed by Darin Adler.

        Cleanup and simplification in advance of making middle clicks not send a
        click event.
        https://bugs.webkit.org/show_bug.cgi?id=22382

        * editing/DeleteButton.cpp:
        (WebCore::DeleteButton::defaultEventHandler): Removed unneeded code to
        check if the event is a MouseEvent. Any click event will do.
        * html/HTMLAnchorElement.cpp:
        (WebCore::appendServerMapMousePosition): Added. Factors out the
        modifications to the URL to add the mouse position to make the default
        event handler function easier to read.
        (WebCore::HTMLAnchorElement::defaultEventHandler): Reorganized to make
        the enter key logic separate from the link clicking logic and simplify
        the function.  This makes minor functional changes like not calling
        FrameLoader::urlSelected() for fake "keydown" events constructed from
        JavaScript.
        (WebCore::HTMLAnchorElement::isLiveLink): Refactored to use new shared
        implementation in treatLinkAsLiveForEventType().
        (WebCore::eventType): Factors out event type calculation (for
        treatLinkAsLiveForEventType()) from the default event handler to make it
        easier to read.
        (WebCore::HTMLAnchorElement::treatLinkAsLiveForEventType):
        Implementation of the portions of isLiveLink() that are common with the
        checks defaultEventHandler() wants to do.
        (WebCore::isEnterKeyKeydownEvent): Added. Shared by the default
        event handlers for all three anchor elements.
        (WebCore::isMiddleMouseButtonEvent): Added. Shared by the isLinkClick
        function below and some code in SVG. Later we can make this private to
        this source file once we remove the unneeded SVG code.
        (WebCore::isLinkClick): Added. Shared by the default event handlers for
        all three anchor elements.
        (WebCore::handleLinkClick): Ditto.
        * html/HTMLAnchorElement.h: Added the new functions.
        * svg/SVGAElement.cpp:
        (WebCore::SVGAElement::defaultEventHandler): Removed lots of unneeded
        logic and streamlined the code to more closely match the
        HTMLAnchorElement code and share functions with it.
        * wml/WMLAElement.cpp:
        (WebCore::WMLAElement::defaultEventHandler): Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67251 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Tony Chang  <tony@chromium.org>

        Unreviewed, fix chromium compile after r67238.

        shouldChangeSelection was removed from Frame.

        * src/WebFrameImpl.cpp:
        (WebKit::WebFrameImpl::selectWordAroundPosition):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67250 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Update mac/Skipped to include the bug numbers for newly added skip entries.

* platform/mac/Skipped:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67249 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Mike Lawther  <mikelawther@chromium.org>

        Reviewed by Ojan Vafai.

        [Chromium] Rebaseline transitions/opacity-transition-zindex.html
        https://bugs.webkit.org/show_bug.cgi?id=45515

        Difference was off-by-one on the green channel.

        * platform/chromium-win/transitions/opacity-transition-zindex-expected.checksum: Added.
        * platform/chromium-win/transitions/opacity-transition-zindex-expected.png: Added.
        * platform/chromium/test_expectations.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67248 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WebKit2: Set the visible name for the web process
https://bugs.webkit.org/show_bug.cgi?id=45564
<rdar://problem/8416970>

Reviewed by Mark Rowe and Darin Adler.

* UIProcess/Launcher/mac/ProcessLauncherMac.mm:
(WebKit::processName):
Add a thread safe implementation of -[NSProcessInfo processName].

(WebKit::ProcessLauncher::launchProcess):
Pass the process name to the child process.

* WebProcess/mac/WebProcessMainMac.mm:
(WebKit::WebProcessMain):
Set the visible name.

WebKitLibraries: WebKitSystemInterface part of:
Set the visible name for the web process
https://bugs.webkit.org/show_bug.cgi?id=45564
<rdar://problem/8416970>

* WebKitSystemInterface.h:
* libWebKitSystemInterfaceLeopard.a:
* libWebKitSystemInterfaceSnowLeopard.a:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67247 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-109  Peter Kasting  <pkasting@google.com>

Reviewed by Darin Adler.

Cleanup and simplification in advance of making middle clicks not send a
click event.
https://bugs.webkit.org/show_bug.cgi?id=22382

* editing/DeleteButton.cpp:
(WebCore::DeleteButton::defaultEventHandler): Removed unneeded code to
check if the event is a MouseEvent. Any click event will do.
* html/HTMLAnchorElement.cpp:
(WebCore::appendServerMapMousePosition): Added. Factors out the
modifications to the URL to add the mouse position to make the default
event handler function easier to read.
(WebCore::HTMLAnchorElement::defaultEventHandler): Reorganized to make
the enter key logic separate from the link clicking logic and simplify
the function.  This makes minor functional changes like not calling
FrameLoader::urlSelected() for fake "keydown" events constructed from
JavaScript.
(WebCore::HTMLAnchorElement::isLiveLink): Refactored to use new shared
implementation in treatLinkAsLiveForEventType().
(WebCore::eventType): Factors out event type calculation (for
treatLinkAsLiveForEventType()) from the default event handler to make it
easier to read.
(WebCore::HTMLAnchorElement::treatLinkAsLiveForEventType):
Implementation of the portions of isLiveLink() that are common with the
checks defaultEventHandler() wants to do.
(WebCore::isEnterKeyKeydownEvent): Added. Shared by the default
event handlers for all three anchor elements.
(WebCore::isMiddleMouseButtonEvent): Added. Shared by the isLinkClick
function below and some code in SVG. Later we can make this private to
this source file once we remove the unneeded SVG code.
(WebCore::isLinkClick): Added. Shared by the default event handlers for
all three anchor elements.
(WebCore::handleLinkClick): Ditto.
* html/HTMLAnchorElement.h: Added the new functions.
* svg/SVGAElement.cpp:
(WebCore::SVGAElement::defaultEventHandler): Removed lots of unneeded
logic and streamlined the code to more closely match the
HTMLAnchorElement code and share functions with it.
* wml/WMLAElement.cpp:
(WebCore::WMLAElement::defaultEventHandler): Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67246 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Tony Gentilcore  <tonyg@chromium.org>

        Reviewed by Eric Seidel.

        Implement HTML5 definition of script for-event neutering
        https://bugs.webkit.org/show_bug.cgi?id=45493

        * fast/dom/HTMLScriptElement/script-for-attribute-unexpected-execution-expected.txt:
        * fast/dom/HTMLScriptElement/script-tests/script-for-attribute-unexpected-execution.js: Beef up test with some more cases, change style to be more readable, and remove obsolete comments about supporting for-event.
2010-09-10  Tony Gentilcore  <tonyg@chromium.org>

        Reviewed by Eric Seidel.

        Implement HTML5 definition of script for-event neutering
        https://bugs.webkit.org/show_bug.cgi?id=45493

        * dom/ScriptElement.cpp:
        (WebCore::ScriptElementData::shouldExecuteAsJavaScript): Same behavior, but reads more like the spec. Removed obsolete comments about supporting for-event.
        * html/parser/HTMLScriptRunner.cpp:
        (WebCore::HTMLScriptRunner::executeScript): Moved to runScript.
        (WebCore::HTMLScriptRunner::runScript): Checking shouldExecuteAsJavaScript() here means that external scripts which fail the check will never be requested.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67245 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Nat Duca  <nduca@chromium.org>

        Reviewed by Darin Fisher.

        [chromium] Accelerated Compositing: screen garbage when scrolling
        https://bugs.webkit.org/show_bug.cgi?id=45092

        Split LayerRenderChromium::drawLayers into several different
        functions, responsible for preparing the backbuffer, updating the
        root texture, compositing and performing the final
        swapbuffers. This are then used by the new
        WebViewImpl::composite rendering path.

        * platform/graphics/chromium/LayerChromium.cpp:
        (WebCore::LayerChromium::setBounds):
        (WebCore::LayerChromium::setFrame):
        (WebCore::LayerChromium::setNeedsDisplay):
        (WebCore::LayerChromium::resetNeedsDisplay):
        * platform/graphics/chromium/LayerChromium.h:
        (WebCore::LayerChromium::dirtyRect):
        * platform/graphics/chromium/LayerRendererChromium.cpp:
        (WebCore::LayerRendererChromium::prepareToDrawLayers):
        (WebCore::LayerRendererChromium::updateRootLayerTextureRect):
        (WebCore::LayerRendererChromium::drawLayers):
        (WebCore::LayerRendererChromium::present):
        * platform/graphics/chromium/LayerRendererChromium.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67244 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Ryosuke Niwa  <rniwa@webkit.org>

        Reviewed by Eric Seidel.

        execCommand("fontsize") on certain selected html generates too many SPAN tags.
        https://bugs.webkit.org/show_bug.cgi?id=28968

        Added a test to ensure only one font element is added when applying font-size
        to multiple inline nodes.

        * editing/style/apply-font-size-to-multiple-nodes-expected.txt: Added.
        * editing/style/apply-font-size-to-multiple-nodes.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67243 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Tony Chang  <tony@chromium.org>

        Unreviewed, rolling out r67241.
        http://trac.webkit.org/changeset/67241
        https://bugs.webkit.org/show_bug.cgi?id=44709

        Accidentally committed.

        * Scripts/webkitpy/layout_tests/deduplicate_tests.py:
        * Scripts/webkitpy/layout_tests/deduplicate_tests_unittest.py:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67242 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Tony Chang  <tony@chromium.org>

        Reviewed by Ojan Vafai.

        deduplicate-tests should be runnable from any WebKit directory
        https://bugs.webkit.org/show_bug.cgi?id=44709

        * Scripts/webkitpy/layout_tests/deduplicate_tests.py:
        * Scripts/webkitpy/layout_tests/deduplicate_tests_unittest.py:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67241 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  David Holloway  <dhollowa@chromium.org>

        Reviewed by Adam Barth.

        Change WebKit::WebFormElement::wasUserSubmitted to use WebCore::HTMLFormElement::submittedByUserGesture
        instead of WebCore::HTMLFormElement::submissionTrigger.
        Chromium AutoFill feature semantics changed with respect to what we mean by "user submitted" form.  A
        more accurate measure of the user manually submitting a form is via user gesture.
        https://bugs.webkit.org/show_bug.cgi?id=45128

        * src/WebFormElement.cpp:
        (WebKit::WebFormElement::wasUserSubmitted):
2010-09-10  David Holloway  <dhollowa@chromium.org>

        Reviewed by Adam Barth.

        Replace WebCore::HTMLFormElement::submissionTrigger with WebCore::HTMLFormElement::submittedByUserGesture.
        Chromium AutoFill feature semantics changed with respect to what we mean by "user submitted" form.  A
        more accurate measure of the user manually submitting a form is via user gesture.
        https://bugs.webkit.org/show_bug.cgi?id=45128

        No new tests as this is only used by the Chromium WebKit API.

        * html/HTMLFormElement.cpp:
        (WebCore::HTMLFormElement::HTMLFormElement):
        (WebCore::HTMLFormElement::prepareSubmit):
        (WebCore::HTMLFormElement::submit):
        (WebCore::HTMLFormElement::submittedByUserGesture):
        * html/HTMLFormElement.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67240 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        Remove mention of "part" from Document.cpp
        https://bugs.webkit.org/show_bug.cgi?id=45566

        This is an old KHTML name that's lingering around.

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

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67239 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  Darin Adler  <darin@apple.com>

        Reviewed by Adam Barth.

        Move functions from Frame to SelectionController as planned
        https://bugs.webkit.org/show_bug.cgi?id=45508

        * src/WebFrameImpl.cpp:
        (WebKit::WebFrameImpl::selectionBoundsRect):
        Call functions on selection().
2010-09-09  Darin Adler  <darin@apple.com>

        Reviewed by Adam Barth.

        Move functions from Frame to SelectionController as planned
        https://bugs.webkit.org/show_bug.cgi?id=45508

        * WebView.cpp:
        (WebView::selectionRect):
        (WebView::centerSelectionInVisibleArea):
        Call functions on selection().
2010-09-09  Darin Adler  <darin@apple.com>

        Reviewed by Adam Barth.

        Move functions from Frame to SelectionController as planned
        https://bugs.webkit.org/show_bug.cgi?id=45508

        * WebView/WebFrame.mm:
        (-[WebFrame _selectionGranularity]):
        (-[WebFrame _insertParagraphSeparatorInQuotedContent]):
        (-[WebFrame _typingStyle]):
        (-[WebFrame _replaceSelectionWithFragment:selectReplacement:smartReplace:matchStyle:]):
        * WebView/WebHTMLRepresentation.mm:
        (-[WebHTMLRepresentation currentForm]):
        * WebView/WebHTMLView.mm:
        (-[WebHTMLView jumpToSelection:]):
        (-[WebHTMLView centerSelectionInVisibleArea:]):
        (-[WebHTMLView _canSmartCopyOrDelete]):
        (-[WebHTMLView _lookUpInDictionaryFromMenu:]):
        (-[WebHTMLView selectionRect]):
        (-[WebHTMLView selectionTextRects]):
        (-[WebHTMLView selectionImageRect]):
        * WebView/WebView.mm:
        (-[WebView setEditable:]):
        Call functions on selection().
2010-09-09  Darin Adler  <darin@apple.com>

        Reviewed by Adam Barth.

        Move functions from Frame to SelectionController as planned
        https://bugs.webkit.org/show_bug.cgi?id=45508

        Also removed Frame::dragCaretController since it just called Page::dragCaretController.

        * editing/SelectionController.cpp:
        (WebCore::SelectionController::setSelection):
        (WebCore::SelectionController::modify):
        (WebCore::SelectionController::selectFrameElementInParentIfFullySelected):
        (WebCore::SelectionController::selectAll):
        (WebCore::SelectionController::focusedOrActiveStateChanged):
        (WebCore::SelectionController::notifyRendererOfSelectionChange):
        (WebCore::isFrameElement):
        (WebCore::SelectionController::setFocusedNodeIfNeeded):
        (WebCore::SelectionController::paintDragCaret):
        (WebCore::SelectionController::shouldDeleteSelection):
        (WebCore::SelectionController::selectionBounds):
        (WebCore::SelectionController::selectionTextRects):
        (WebCore::scanForForm):
        (WebCore::SelectionController::currentForm):
        (WebCore::SelectionController::revealSelection):
        (WebCore::SelectionController::setSelectionFromNone):
        (WebCore::SelectionController::shouldChangeSelection):
        * editing/SelectionController.h:
        Moved functions here from Frame.

        * page/Frame.cpp:
        * page/Frame.h:
        Moved functions to SelectionController.
        Reorganized header to be easier to read and a bit more logical.

        * WebCore.exp.in: Updated.

        * dom/Document.cpp:
        (WebCore::Document::nodeChildrenWillBeRemoved):
        (WebCore::Document::nodeWillBeRemoved):
        Call Page::dragCaretController directly.

        * dom/Element.cpp:
        (WebCore::Element::updateFocusAppearance):
        * dom/InputElement.cpp:
        (WebCore::InputElement::updateFocusAppearance):
        * editing/ApplyStyleCommand.cpp:
        (WebCore::ApplyStyleCommand::editingStyleAtPosition):
        * editing/DeleteSelectionCommand.cpp:
        (WebCore::DeleteSelectionCommand::calculateTypingStyleAfterDelete):
        * editing/Editor.cpp:
        (WebCore::Editor::canSmartCopyOrDelete):
        (WebCore::Editor::textDirectionForSelection):
        (WebCore::Editor::appliedEditing):
        (WebCore::Editor::insertTextWithoutSendingTextEvent):
        (WebCore::Editor::advanceToNextMisspelling):
        (WebCore::Editor::markAllMisspellingsAndBadGrammarInRanges):
        (WebCore::Editor::revealSelectionAfterEditingOperation):
        (WebCore::Editor::transpose):
        (WebCore::Editor::changeSelectionAfterCommand):
        (WebCore::Editor::computeAndSetTypingStyle):
        (WebCore::Editor::selectionComputedStyle):
        (WebCore::Editor::styleForSelectionStart):
        (WebCore::Editor::findString):
        * editing/EditorCommand.cpp:
        (WebCore::executeDelete):
        * editing/InsertLineBreakCommand.cpp:
        (WebCore::InsertLineBreakCommand::doApply):
        * editing/InsertTextCommand.cpp:
        (WebCore::InsertTextCommand::input):
        * editing/RemoveFormatCommand.cpp:
        (WebCore::RemoveFormatCommand::doApply):
        * editing/ReplaceSelectionCommand.cpp:
        (WebCore::ReplaceSelectionCommand::doApply):
        * editing/TypingCommand.cpp:
        (WebCore::TypingCommand::insertTextRunWithoutNewlines):
        (WebCore::TypingCommand::deleteKeyPressed):
        (WebCore::TypingCommand::forwardDeleteKeyPressed):
        * html/HTMLTextAreaElement.cpp:
        (WebCore::HTMLTextAreaElement::updateFocusAppearance):
        * page/ContextMenuController.cpp:
        (WebCore::ContextMenuController::contextMenuItemSelected):
        * page/DragController.cpp:
        (WebCore::DragController::concludeEditDrag):
        (WebCore::dragLocForSelectionDrag):
        * page/EventHandler.cpp:
        (WebCore::EventHandler::selectClosestWordFromMouseEvent):
        (WebCore::EventHandler::selectClosestWordOrLinkFromMouseEvent):
        (WebCore::EventHandler::handleMousePressEventTripleClick):
        (WebCore::EventHandler::handleMousePressEventSingleClick):
        (WebCore::EventHandler::updateSelectionForMouseDrag):
        (WebCore::EventHandler::handleMouseReleaseEvent):
        * page/FocusController.cpp:
        (WebCore::FocusController::advanceFocusInDocumentOrder):
        * page/chromium/EventHandlerChromium.cpp:
        (WebCore::EventHandler::passMousePressEventToSubframe):
        * page/chromium/FrameChromium.cpp:
        (WebCore::Frame::dragImageForSelection):
        * page/mac/FrameMac.mm:
        (WebCore::Frame::selectionImage):
        * platform/mac/ClipboardMac.mm:
        (WebCore::ClipboardMac::writeRange):
        * rendering/RenderBlock.cpp:
        (WebCore::RenderBlock::paintCaret):
        Call functions on selection().

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67238 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Simon Fraser  <simon.fraser@apple.com>

        Re-enable some Windows compositing tests that no longer assert.

        * platform/win/Skipped:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67237 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Justin Schuh  <jschuh@chromium.org>

        Reviewed by Darin Adler.

        SVGGElement::rendererIsNeeded should return false when parent isn't SVG
        https://bugs.webkit.org/show_bug.cgi?id=45562

        Test: svg/custom/g-outside-svg.html

        * svg/SVGGElement.cpp:
        (WebCore::SVGGElement::rendererIsNeeded):
        * svg/SVGGElement.h:
2010-09-10  Justin Schuh  <jschuh@chromium.org>

        Reviewed by Darin Adler.

        SVGGElement::rendererIsNeeded should return false when parent isn't SVG
        https://bugs.webkit.org/show_bug.cgi?id=45562

        * svg/custom/g-outside-svg-expected.txt: Added.
        * svg/custom/g-outside-svg.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67236 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Skip file api tests for workers that could crash occasionally before
we figure out how to fix them.

* platform/mac/Skipped:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67235 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Adam Barth  <abarth@webkit.org>

        Chromium build fix (one hopes!)

        * src/WebFrameImpl.cpp:
        (WebKit::WebFrameImpl::commitDocumentData):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67234 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Adam Barth  <abarth@webkit.org>

        Reviewed by Darin Adler.

        REGRESSION (HTML Parser): CNN's Money site is not formatted correctly with AdBlock installed
        https://bugs.webkit.org/show_bug.cgi?id=41371

        Drop support for closing comments with "-- >".  This was an attempt to
        be more IE-like in comment parsing, but it turns out to cause problems
        for some high-profile sites.  Firefox 4 is dropping support as well.
        We expect the spec to change soon too.

        * html/parser/HTMLTokenizer.cpp:
        (WebCore::HTMLTokenizer::nextToken):
        * html/parser/HTMLTokenizer.h:
2010-09-10  Adam Barth  <abarth@webkit.org>

        Reviewed by Darin Adler.

        REGRESSION (HTML Parser): CNN's Money site is not formatted correctly with AdBlock installed
        https://bugs.webkit.org/show_bug.cgi?id=41371

        Update test results for new behavior.

        * fast/parser/comments-expected.txt:
        * fast/parser/comments.html:
          - Removed the redundant in-test expectations.
        * html5lib/runner-expected.txt:
          - We now "fail" these two tests.  We'll pass them again once the spec
            changes and HTML5lib updates its expectations.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67233 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Simon Fraser  <simon.fraser@apple.com>

        Reviewed by Adam Roben.

        compositing/iframes/iframe-src-change.html test fails on Windows
        https://bugs.webkit.org/show_bug.cgi?id=45554

        This test had the wrong behavior. It never actually entered compositing mode.
        Now give it a chance to do so (and test for it), and then test that the layers
        disappeared.

        * compositing/iframes/iframe-src-change-expected.txt:
        * compositing/iframes/iframe-src-change.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67232 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Restored a Changelog entry for a change that was reverted. I mistakenly thought reverted changes should be removed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67231 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Give expected results files the right names

* platform/win/fast/text/hyphenate-first-word-expected.txt: Renamed from LayoutTests/platform/win/fast/text/hyphenate-first-word-actual.txt.
* platform/win/fast/text/hyphenate-locale-expected.txt: Renamed from LayoutTests/platform/win/fast/text/hyphenate-locale-actual.txt.
* platform/win/html5lib/runner-expected.txt: Renamed from LayoutTests/platform/win/html5lib/runner-actual.txt.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67230 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Chris Marrin  <cmarrin@apple.com>

        Unreviewed.

        Adding file that should have been in http://trac.webkit.org/changeset/67226 but
        I forgot to svn add.

        * platform/graphics/opengl: Added.
        * platform/graphics/opengl/GraphicsContext3DOpenGL.cpp: Copied from platform/graphics/mac/GraphicsContext3DMac.mm.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67229 268f45cc-cd09-0410-ab3c-d52691b4dbfc

one more typo fix in Changelog

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67228 268f45cc-cd09-0410-ab3c-d52691b4dbfc

fixed typo in Changelog

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67227 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Paul Sawaya  <psawaya@apple.com>

        Reviewed by Chris Marr.

        Refactored non-Mac specific code into GraphicsContext3DOpenGL.

        Moving this file into platform/graphics/mac for now, and only building on Mac.

        https://bugs.webkit.org/show_bug.cgi?id=30625

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67226 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Qt build fix.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67225 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Adam Barth  <abarth@webkit.org>

        Reviewed by Darin Adler.

        Remove useless FrameLoader method
        https://bugs.webkit.org/show_bug.cgi?id=45492

        This method has one caller, and that caller already has a more direct
        way of accessing the correct DocumentLoader.

        * loader/FrameLoader.cpp:
        * loader/FrameLoader.h:
        * loader/MainResourceLoader.cpp:
        (WebCore::MainResourceLoader::addData):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67224 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        Main resource bytes shouldn't bounce through FrameLoader
        https://bugs.webkit.org/show_bug.cgi?id=45496

        After the MainResourceLoader hands the bytes for the main resource to
        the DocumentLoader, the DocumentLoader hands the bytes to the
        FrameLoaderClient.  The FrameLoaderClient, in turn, returns the bytes
        to WebCore.  Prior to this patch, the FrameLoaderClient returned the
        bytes to the FrameLoader.  However, the FrameLoader was actually adding
        any value.

        In this patch, we change the FrameLoaderClient to return the bytes to
        the DocumentLoader directly.  In a future patch, we'll cut out the
        FrameLoaderClient middleman and handle the bytes internally in the
        DocumentLoader.

        * WebCore.exp.in:
        * loader/DocumentLoader.cpp:
        (WebCore::DocumentLoader::addData):
        * loader/DocumentLoader.h:
        * loader/FrameLoader.cpp:
        (WebCore::FrameLoader::finishedLoadingDocument):
        * loader/FrameLoader.h:
2010-09-10  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        Main resource bytes shouldn't bounce through FrameLoader
        https://bugs.webkit.org/show_bug.cgi?id=45496

        Now return the bytes to the DocumentLoader.

        Also, removed comment about wrong code in the Mac port.

        * src/WebFrameImpl.cpp:
        (WebKit::WebFrameImpl::commitDocumentData):
2010-09-10  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        Main resource bytes shouldn't bounce through FrameLoader
        https://bugs.webkit.org/show_bug.cgi?id=45496

        Now return the bytes to the DocumentLoader.

        * WebCoreSupport/FrameLoaderClientEfl.cpp:
        (WebCore::FrameLoaderClientEfl::committedLoad):
2010-09-10  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        Main resource bytes shouldn't bounce through FrameLoader
        https://bugs.webkit.org/show_bug.cgi?id=45496

        Now return the bytes to the DocumentLoader.

        * WebCoreSupport/FrameLoaderClientGtk.cpp:
        (WebKit::FrameLoaderClient::committedLoad):
2010-09-10  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        Main resource bytes shouldn't bounce through FrameLoader
        https://bugs.webkit.org/show_bug.cgi?id=45496

        Now return the bytes to the DocumentLoader.

        * WebCoreSupport/FrameLoaderClientHaiku.cpp:
        (WebCore::FrameLoaderClientHaiku::committedLoad):
2010-09-10  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        Main resource bytes shouldn't bounce through FrameLoader
        https://bugs.webkit.org/show_bug.cgi?id=45496

        Now return the bytes to the DocumentLoader.

        Previously, we were checking the document for null.  However, Frame can
        never have a null document, so this check is no longer needed.

        * WebView/WebFrame.mm:
        (-[WebFrame _addData:]):
2010-09-10  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        Main resource bytes shouldn't bounce through FrameLoader
        https://bugs.webkit.org/show_bug.cgi?id=45496

        Now return the bytes to the DocumentLoader.

        * WebCoreSupport/FrameLoaderClientQt.cpp:
        (WebCore::FrameLoaderClientQt::committedLoad):
2010-09-10  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        Main resource bytes shouldn't bounce through FrameLoader
        https://bugs.webkit.org/show_bug.cgi?id=45496

        Now return the bytes to the DocumentLoader.

        * WebCoreSupport/WebFrameLoaderClient.cpp:
        (WebFrameLoaderClient::receivedData):
2010-09-10  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        Main resource bytes shouldn't bounce through FrameLoader
        https://bugs.webkit.org/show_bug.cgi?id=45496

        Now return the bytes to the DocumentLoader.

        * WebKitSupport/FrameLoaderClientWx.cpp:
        (WebCore::FrameLoaderClientWx::committedLoad):
2010-09-10  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        Main resource bytes shouldn't bounce through FrameLoader
        https://bugs.webkit.org/show_bug.cgi?id=45496

        Now return the bytes to the DocumentLoader.

        * WebProcess/WebCoreSupport/WebFrameLoaderClient.cpp:
        (WebKit::WebFrameLoaderClient::receivedData):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67223 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Oliver Hunt  <oliver@apple.com>

        Reviewed by Darin Adler.

        Speed up deserialisation of strings
        https://bugs.webkit.org/show_bug.cgi?id=45555

        Rather than building a list of Identifiers for the string pool
        we now build a list of a tuple of UString and JSString*.  This
        doesn't hurt the property name case as ustring->identifier conversion
        is essentially free if the ustring has already been converted to
        an Identifier, but saves an unnecessary Identifier creation for
        strings we only ever use to create JSStrings.  We also reduce
        GC pressure for duplicate strings by caching the JSStrings.

        * bindings/js/SerializedScriptValue.cpp:
        (WebCore::CloneDeserializer::CachedString::CachedString):
        (WebCore::CloneDeserializer::CachedString::jsString):
        (WebCore::CloneDeserializer::CachedString::ustring):
        (WebCore::CloneDeserializer::readStringData):
        (WebCore::CloneDeserializer::putProperty):
        (WebCore::CloneDeserializer::readFile):
        (WebCore::CloneDeserializer::readTerminal):
        (WebCore::CloneDeserializer::deserialize):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67222 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Ryosuke Niwa  <rniwa@webkit.org>

        Reviewed by Antonio Gomes.

        REGRESSION(r67176): editing/selection/doubleclick-inline-first-last-contenteditable.html crashes
        https://bugs.webkit.org/show_bug.cgi?id=45525

        The crash was caused by ancestorToRetainStructureAndAppearance not taking into consideration
        the case where there is no common block ancestor. Fixed this by adding a null check.

        Test: editing/pasteboard/copy-without-common-block-crash.html

        * editing/markup.cpp:
        (WebCore::ancestorToRetainStructureAndAppearance): Added a null check.
2010-09-10  Ryosuke Niwa  <rniwa@webkit.org>

        Reviewed by Antonio Gomes.

        REGRESSION(r67176): editing/selection/doubleclick-inline-first-last-contenteditable.html crashes
        https://bugs.webkit.org/show_bug.cgi?id=45525

        Added a test to ensure WebKit does not crash when copying a region that doesn't have any common block ancestor.
        Also removed doubleclick-inline-first-last-contenteditable.html from the qt's skipped list since
        this test should not crash anymore.

        * editing/pasteboard/copy-without-common-block-crash-expected.txt: Added.
        * editing/pasteboard/copy-without-common-block-crash.html: Added.
        * platform/qt/Skipped: Removed doubleclick-inline-first-last-contenteditable.html from the list.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67221 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  Kinuko Yasuda  <kinuko@chromium.org>

        Reviewed by Dumitru Daniliuc.

        [FileSystem] Do not call EntriesCallback more than once if there're no entries.
        https://bugs.webkit.org/show_bug.cgi?id=45498

        No new tests, layout-tests that confirm the behavior will be added in later patches.

        * fileapi/FileSystemCallbacks.cpp:
        (WebCore::EntriesCallbacks::EntriesCallbacks):
        (WebCore::EntriesCallbacks::didReadDirectoryEntry):
        (WebCore::EntriesCallbacks::didReadDirectoryEntries):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67220 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WebCore: Use String::fromUTF8 instead of UTF8Encoding().decode
in the libxml XMLDocumentParser
https://bugs.webkit.org/show_bug.cgi?id=45488

Patch by Patrick Gansterer <paroga@paroga.com> on 2010-09-10
Reviewed by Darin Adler.

* dom/XMLDocumentParserLibxml2.cpp:
(WebCore::toString):

LayoutTests: Land new expected results for some tests on Windows

Patch by Adam Roben <aroben@apple.com> on 2010-09-10
Rubber-stamped in advance by Simon Fraser.

* platform/win/fast/text/hyphenate-first-word-actual.txt: Added.
* platform/win/fast/text/hyphenate-locale-actual.txt: Added.
Windows doesn't support hyphenation.

* platform/win/html5lib/runner-actual.txt: Added.
We don't implement <keygen> on Windows, so we pass some tests that
expect <keygen> to have no child elements (on Mac, the child elements
are real DOM elements, rather than living in the shadow DOM tree like
HTML5 mandates, so the tests fail).

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67219 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Tony Chang  <tony@chromium.org>

        Unreviewed, rolling out r67216.
        http://trac.webkit.org/changeset/67216
        https://bugs.webkit.org/show_bug.cgi?id=44709

        Broke

        * Scripts/webkitpy/layout_tests/deduplicate_tests.py:
        * Scripts/webkitpy/layout_tests/deduplicate_tests_unittest.py:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67218 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Land new expected results for some tests on Windows

Rubber-stamped in advance by Simon Fraser.

* platform/win/fast/text/hyphenate-first-word-actual.txt: Added.
* platform/win/fast/text/hyphenate-locale-actual.txt: Added.
Windows doesn't support hyphenation.

* platform/win/html5lib/runner-actual.txt: Added.
We don't implement <keygen> on Windows, so we pass some tests that
expect <keygen> to have no child elements (on Mac, the child elements
are real DOM elements, rather than living in the shadow DOM tree like
HTML5 mandates, so the tests fail).

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67217 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Tony Chang  <tony@chromium.org>

        Reviewed by Ojan Vafai.

        deduplicate-tests should be runnable from any WebKit directory
        https://bugs.webkit.org/show_bug.cgi?id=44709

        * Scripts/webkitpy/layout_tests/deduplicate_tests.py:
        * Scripts/webkitpy/layout_tests/deduplicate_tests_unittest.py:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67216 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Martin Robinson  <mrobinson@igalia.com>

        Reviewed by Xan Lopez.

        [GTK] Placement new / manual destructor invocation should be used on private GObject memory
        https://bugs.webkit.org/show_bug.cgi?id=45550

        GLib allocates and deallocates GObject private data structs itself. When
        those structs contain C++ members, their constructors and destructors are not
        called. This is not only dangerous, it makes RefPtr-type smart pointers much
        less useful. We can fix this problem by calling placement new on the private
        data struct during instance initialization and calling the destructor during
        finalization.

        This patch takes that approach and switches plain char* members of
        WebKitWebView (with manual memory allocation) to use CString.

        * webkit/webkitprivate.h: Switch char* members to CString.
        * webkit/webkitwebview.cpp:
        (webkit_web_view_finalize): Manually call the destructor on the private
        data. Remove manual deallocation of members which are now CString.
        (webkit_web_view_query_tooltip): Update to reflect CString change.
        (webkit_web_view_init): Use placement new to initialize C++ members of
        the private data section.
        (webkit_web_view_get_encoding): Update to reflect CString change.
        (webkit_web_view_get_custom_encoding): Ditto.
        (webkit_web_view_add_resource): Ditto.
        (webkit_web_view_get_resource): Ditto.
        (webkit_web_view_clear_resources): Ditto.
        (webkit_web_view_set_tooltip_text): Ditto.
        (webkit_web_view_get_icon_uri): Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67215 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Patrick Gansterer  <paroga@paroga.com>

        Reviewed by Darin Adler.

        Use String::fromUTF8 instead of UTF8Encoding().decode
        in the libxml XMLDocumentParser
        https://bugs.webkit.org/show_bug.cgi?id=45488

        * dom/XMLDocumentParserLibxml2.cpp:
        (WebCore::toString):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67214 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  Joseph Pecoraro  <joepeck@webkit.org>

        Reviewed by Andreas Kling.

        Clean Up Some Expected Results for Missing Tests
        https://bugs.webkit.org/show_bug.cgi?id=45489

          Test was removed in r35456 but expected results were missed.

        * http/tests/xmlhttprequest/access-control-basic-exclude-expected.txt: Removed.

          Results were added in r19036 but the test was never added. No record of the test.

        * editing/pasteboard/4944770-expected.txt: Removed.
        * platform/mac/editing/pasteboard/4944770-expected.checksum: Removed.
        * platform/mac/editing/pasteboard/4944770-expected.png: Removed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67213 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Simon Fraser  <simon.fraser@apple.com>

        Reviewed by Chris Marrin.

        Update Windows LayoutTest/compositing results
        https://bugs.webkit.org/show_bug.cgi?id=45323

        Fix up some windows compositing test results.

        * platform/win/compositing/iframes/composited-parent-iframe-expected.txt: Removed.
        * platform/win/compositing/iframes/iframe-resize-expected.txt: Removed.
        * platform/win/compositing/iframes/leave-compositing-iframe-expected.txt: Removed.
        * platform/win/compositing/iframes/overlapped-iframe-expected.txt: Removed.
        * platform/win/compositing/plugins/composited-plugin-expected.txt: Added.
        * platform/win/compositing/plugins/small-to-large-composited-plugin-expected.txt: Copied from LayoutTests/compositing/iframes/leave-compositing-iframe-expected.txt.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67212 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Gustavo Noronha Silva  <gns@gnome.org>

        Reviewed by Martin Robinson.

        [GTK] Google sites do not like WebKitGTK+
        https://bugs.webkit.org/show_bug.cgi?id=39617

        Special-case Google domains, and spoof User-Agent when talking to
        them, to stop being treated as a second-class citizen.

        * WebCoreSupport/FrameLoaderClientGtk.cpp:
        (WebKit::initializeDomainsList):
        (WebKit::isGoogleDomain):
        (WebKit::FrameLoaderClient::userAgent): If enable-site-specific-quirks
        is enabled, send the standard WebKit User-Agent string, disregarding the
        custom one set by the browser for Google domains.
        * webkit/webkitprivate.h:
        * webkit/webkitwebsettings.cpp:
        (webkitPlatform): Fix style, and simplify.
        (webkitOSVersion): Ditto.
        (webkitUserAgent): Add the Version/x.y string Safari has been
        using since Safari 2.2.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67211 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Pavel Podivilov  <podivilov@chromium.org>

        Reviewed by Yury Semikhatsky.

        [Chromium] rebaseline SVGFEDiffuseLightingElement-lighting-color-css-prop.html
        https://bugs.webkit.org/show_bug.cgi?id=45540

        * platform/chromium-linux/svg/dynamic-updates/SVGFEDiffuseLightingElement-lighting-color-css-prop-expected.checksum:
        * platform/chromium-linux/svg/dynamic-updates/SVGFEDiffuseLightingElement-lighting-color-css-prop-expected.png:
        * platform/chromium-win/svg/dynamic-updates/SVGFEDiffuseLightingElement-lighting-color-css-prop-expected.checksum:
        * platform/chromium-win/svg/dynamic-updates/SVGFEDiffuseLightingElement-lighting-color-css-prop-expected.png:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67210 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Mail crashes with searching for next misspelled word with VoiceOver
https://bugs.webkit.org/show_bug.cgi?id=45501

Reviewed by David Kilzer.

WebCore:

Test: platform/mac/accessibility/crash-in-element-for-text-marker.html

* accessibility/mac/AccessibilityObjectWrapper.mm:
(-[AccessibilityObjectWrapper accessibilityAttributeValue:forParameter:]):
    Check that the AX object is not nil before asking for the wrapper().

LayoutTests:

* platform/mac/accessibility/crash-in-element-for-text-marker-expected.txt: Added.
* platform/mac/accessibility/crash-in-element-for-text-marker.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67209 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Michael Nordman  <michaeln@google.com>

        Reviewed by Dumitru Daniliuc.

        Decouple Blob from ScriptExecutionContext.
        https://bugs.webkit.org/show_bug.cgi?id=45410

        - Removed ScriptExecutionContext pointers from Blob and File constructors.
        - Removed ScriptExecutionContext's collection of Blobs instance in that context.
        - Refactored BlobURL to make a distinction between 'public' blob urls and 'internal'
          urls that are used solely as identifiers. The former requires a SecurityOrigin to
          to manufacture, the latter does not.
        - Added a constant and accessor for the blob protocol scheme on the BlobURL class.

        No new tests are added since this is a cleanup.

        * bindings/js/SerializedScriptValue.cpp:
        (WebCore::CloneDeserializer::readFile):
        (WebCore::CloneDeserializer::readTerminal):
        * bindings/v8/SerializedScriptValue.cpp:
        (WebCore::ZigZag::Reader::readBlob):
        (WebCore::ZigZag::Reader::readFile):
        (WebCore::ZigZag::Reader::readFileList):
        * dom/ScriptExecutionContext.cpp:
        (WebCore::ScriptExecutionContext::~ScriptExecutionContext):
        (WebCore::ScriptExecutionContext::createPublicBlobURL):
        (WebCore::ScriptExecutionContext::revokePublicBlobURL):
        * dom/ScriptExecutionContext.h:
        * fileapi/Blob.cpp:
        (WebCore::Blob::Blob):
        (WebCore::Blob::~Blob):
        (WebCore::Blob::slice):
        * fileapi/Blob.h:
        (WebCore::Blob::create):
        (WebCore::Blob::url):
        * fileapi/Blob.idl:
        * fileapi/BlobBuilder.cpp:
        (WebCore::BlobBuilder::getBlob):
        * fileapi/BlobBuilder.h:
        * fileapi/BlobBuilder.idl:
        * fileapi/BlobURL.cpp:
        (WebCore::BlobURL::createPublicURL):
        (WebCore::BlobURL::createInternalURL):
        (WebCore::BlobURL::getOrigin):
        (WebCore::BlobURL::getIdentifier):
        (WebCore::BlobURL::createBlobURL):
        * fileapi/BlobURL.h:
        (WebCore::BlobURL::blobProtocol):
        (WebCore::BlobURL::BlobURL):
        * fileapi/File.cpp:
        (WebCore::File::File):
        * fileapi/File.h:
        (WebCore::File::create):
        * fileapi/FileReader.cpp: Register a public url to perform the read.
        (WebCore::FileReader::cleanup):
        (WebCore::FileReader::start):
        (WebCore::FileReader::didFinishLoading):
        (WebCore::FileReader::failed):
        * fileapi/FileReader.h: Added m_urlForReading data member.
        * fileapi/FileReaderSync.cpp:
        (WebCore::FileReaderSync::read): Register a public url to perform the read.
        * fileapi/ThreadableBlobRegistry.cpp:
        (WebCore::ThreadableBlobRegistry::registerBlobURL):
        (WebCore::ThreadableBlobRegistry::unregisterBlobURL):
        * fileapi/ThreadableBlobRegistry.h:
        * html/HTMLInputElement.cpp:
        (WebCore::HTMLInputElement::appendFormData):
        (WebCore::HTMLInputElement::setFileListFromRenderer):
        * page/SecurityOrigin.cpp: Use BlobURL.blobProtocol().
        (WebCore::SecurityOrigin::create):
        (WebCore::SecurityOrigin::canRequest):
        (WebCore::SecurityOrigin::canDisplay):
        * platform/chromium/ClipboardChromium.cpp:
        (WebCore::ClipboardChromium::files):
        * platform/gtk/ClipboardGtk.cpp:
        (WebCore::ClipboardGtk::files):
        * platform/mac/ClipboardMac.mm:
        (WebCore::ClipboardMac::files):
        * platform/win/ClipboardWin.cpp:
        (WebCore::ClipboardWin::files):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67208 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Xan Lopez  <xlopez@igalia.com>

        Reviewed by Martin Robinson.

        GDK_DISPLAY was removed in GTK+ 3.x, re-declare it when built
        against that.

        * platform/gtk/GtkVersioning.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67207 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WebCore:

2010-09-10  Xan Lopez  <xlopez@igalia.com>

        Reviewed by Martin Robinson.

        Fix compilation with GTK+ 3.x.

        GDK key macros were renamed from GDK_FOO to GDK_KEY_FOO, we need
        to include the compat header provided if we want to keep using the
        old names.

        * platform/gtk/GtkVersioning.h: add the compatibility header.
        * platform/gtk/KeyEventGtk.cpp: include GtkVersioning.h

WebKit/gtk:

2010-09-10  Xan Lopez  <xlopez@igalia.com>

        Reviewed by Martin Robinson.

        Fix compilation with GTK+ 3.x.

        GDK key macros were renamed from GDK_FOO to GDK_KEY_FOO, we need
        to include the compat header provided if we want to keep using the
        old names.

        * WebCoreSupport/FullscreenVideoController.cpp: include GtkVersioning.h

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67206 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Skip a few media tests that started randomly failing after r66961

Reviewed by Eric Carlson.

* platform/mac-leopard/Skipped:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67205 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  Tony Gentilcore  <tonyg@chromium.org>

        Reviewed by Eric Seidel.

        Factor out the NestingLevelIncrementer
        https://bugs.webkit.org/show_bug.cgi?id=45471

        No new tests because no new functionality.

        * GNUmakefile.am:
        * WebCore.gypi:
        * WebCore.vcproj/WebCore.vcproj:
        * WebCore.xcodeproj/project.pbxproj:
        * html/parser/HTMLDocumentParser.cpp:
        * html/parser/HTMLDocumentParser.h:
        * html/parser/HTMLScriptRunner.cpp:
        * html/parser/NestingLevelIncrementer.h: Added.
        (WebCore::NestingLevelIncrementer::NestingLevelIncrementer):
        (WebCore::NestingLevelIncrementer::~NestingLevelIncrementer):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67204 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Nikolas Zimmermann  <nzimmermann@rim.com>

        Not reviewed. Fix wrong gtk baseline for select-x-list-1.svg.

        * platform/gtk/svg/text/select-x-list-1-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67203 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Don't crash when a frame is destroyed after the UI process has disconnected

Fixes <http://webkit.org/b/45535> <rdar://problem/8412928> Crash in
WebProcess::removeWebFrame on exit with WebKit2 (Release builds only)

Reviewed by Darin Adler.

* WebProcess/WebProcess.cpp:
(WebKit::WebProcess::removeWebFrame): Null-check m_connection before
dereferencing it.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67202 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Martin Robinson  <mrobinson@igalia.com>

        Slight skipped list reorganization.

        * platform/gtk/Skipped: Group one more test in the beginDragWithFiles section.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67201 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Nikolas Zimmermann  <nzimmermann@rim.com>

        Reviewed by Dirk Schulze.

        SVG text chunk concept needs to be integrated in the InlineBox structure
        https://bugs.webkit.org/show_bug.cgi?id=45533

        Begin rewriting the SVG text layout code. As first step move the text chunk concept right into the inline box structure.
        This obsoletes the whole text chunk part concept (see SVGTextChunkLayoutInfo.h for details), and will soon simplify the
        whole layout code a lot. Previously we hacked around the fact that a single InlineTextBox could contain multiple text chunks
        For example: <text x="20 30 40">ABC</text> only led to the creation of one InlineTextBox containing three chunks. When
        painting such an InlineTextBox, we had to do nasty hacks (same for text selection etc.)

        This is finally fixed by incorporating the knownledge about text chunks into the line layout algorithm.
        Note, this has no impact on HTML at all, no worries.

        It's even required to do it this way as BiDi reordering shouldn't happen across text chunks.
        These are only the first bits, and currently slows down text layouting. Follow-up patches will remove the old layout
        algorithm, and will rely on the stored SVGTextLayoutAttributes in the RenerSVGInlineText renderers (instead of computing again).

        * Android.mk: Add rendering/svg/SVGTextLayoutAttributes.* & rendering/svg/SVGTextLayoutBuilder.* to build.
        * CMakeLists.txt: Ditto.
        * GNUmakefile.am: Ditto.
        * WebCore.gyp/WebCore.gyp: Ditto.
        * WebCore.gypi: Ditto.
        * WebCore.pro: Ditto.
        * WebCore.vcproj/WebCore.vcproj: Ditto.
        * WebCore.vcproj/WebCoreCommon.vsprops: Ditto. (Recognize new rendering/svg directory.)
        * WebCore.vcproj/copyForwardingHeaders.cmd: Ditto. (Copy forwarding headers from rendering/svg directory.)
        * WebCore.xcodeproj/project.pbxproj: Ditto.
        * rendering/RenderBlockLineLayout.cpp: This is the key of the new concept, take text chunks into account when building the InlineBox structure.
        (WebCore::RenderBlock::findNextLineBreak): Split midpointState, if we encounter the beginning of a new text chunk.
        * rendering/RenderObject.h:
        (WebCore::RenderObject::isSVGInline): New method, which will be used to identify RenderSVGInline objects.
        * rendering/RenderSVGInline.h:
        (WebCore::RenderSVGInline::isSVGInline): Return true.
        * rendering/RenderSVGInlineText.cpp:
        (WebCore::RenderSVGInlineText::linesBoundingBox): Overriden, to simplify the calculations for SVG.
        (WebCore::RenderSVGInlineText::characterStartsNewTextChunk): Determines wheter the given position starts a new text chunk.
        * rendering/RenderSVGInlineText.h:
        (WebCore::RenderSVGInlineText::storeLayoutAttributes): Stores the layout attributes calculated in the text layout process (will be used everywhere in follow-up patches).
        * rendering/RenderSVGText.cpp:
        (WebCore::RenderSVGText::layout): Invoke SVGTextLayoutBuilder before creating the inline box tree, so we can decide which characters start new text chunks.
        * rendering/RenderText.h: Made linesBoundingBox() virtual so RenderSVGInlineText can override it.
        * rendering/svg: Added.
        * rendering/svg/SVGTextLayoutAttributes.cpp: Added. Holds x/y/dx/dy/rotate values and all kerning/ligature information used during layout.
        * rendering/svg/SVGTextLayoutAttributes.h: Added.
        * rendering/svg/SVGTextLayoutBuilder.cpp: Added.
        * rendering/svg/SVGTextLayoutBuilder.h: Added.
2010-09-10  Nikolas Zimmermann  <nzimmermann@rim.com>

        Reviewed by Dirk Schulze.

        SVG text chunk concept needs to be integrated in the InlineBox structure
        https://bugs.webkit.org/show_bug.cgi?id=45533

        Updated all tests containing text chunks, to reflect the fact that more SVGInlineTextBoxes are created now.
        All 'startOffset' dumps now begin at 0, indicating that there's only one chunk contained in a single SVGInlineTextBox.
        These dumps will be removed in follow-up patches, as they don't make any sense anymore in the new concept.

        * platform/gtk/svg/text/select-x-list-1-expected.txt:
        * platform/gtk/svg/text/select-x-list-2-expected.txt:
        * platform/gtk/svg/text/select-x-list-3-expected.txt:
        * platform/gtk/svg/text/select-x-list-4-expected.txt:
        * platform/gtk/svg/text/select-x-list-with-tspans-1-expected.txt:
        * platform/gtk/svg/text/select-x-list-with-tspans-2-expected.txt:
        * platform/gtk/svg/text/select-x-list-with-tspans-3-expected.txt:
        * platform/gtk/svg/text/select-x-list-with-tspans-4-expected.txt:
        * platform/mac/svg/W3C-SVG-1.1/animate-elem-40-t-expected.txt:
        * platform/mac/svg/W3C-SVG-1.1/text-text-04-t-expected.txt:
        * platform/mac/svg/W3C-SVG-1.1/text-text-05-t-expected.txt:
        * platform/mac/svg/W3C-SVG-1.1/text-text-06-t-expected.txt:
        * platform/mac/svg/W3C-SVG-1.1/text-text-07-t-expected.txt:
        * platform/mac/svg/W3C-SVG-1.1/text-tspan-01-b-expected.txt:
        * platform/mac/svg/batik/text/textAnchor2-expected.txt:
        * platform/mac/svg/batik/text/textAnchor3-expected.txt:
        * platform/mac/svg/batik/text/textLayout-expected.txt:
        * platform/mac/svg/batik/text/textLayout2-expected.txt:
        * platform/mac/svg/batik/text/textOnPathSpaces-expected.txt:
        * platform/mac/svg/batik/text/textPosition-expected.txt:
        * platform/mac/svg/batik/text/textPosition2-expected.txt:
        * platform/mac/svg/custom/text-letter-spacing-expected.txt:
        * platform/mac/svg/custom/text-x-dx-lists-expected.txt:
        * platform/mac/svg/custom/text-x-override-in-tspan-child-expected.txt:
        * platform/mac/svg/text/select-x-list-1-expected.txt:
        * platform/mac/svg/text/select-x-list-2-expected.txt:
        * platform/mac/svg/text/select-x-list-3-expected.txt:
        * platform/mac/svg/text/select-x-list-4-expected.txt:
        * platform/mac/svg/text/select-x-list-with-tspans-1-expected.txt:
        * platform/mac/svg/text/select-x-list-with-tspans-2-expected.txt:
        * platform/mac/svg/text/select-x-list-with-tspans-3-expected.txt:
        * platform/mac/svg/text/select-x-list-with-tspans-4-expected.txt:
        * platform/mac/svg/text/text-text-04-t-expected.txt:
        * platform/mac/svg/text/text-text-05-t-expected.txt:
        * platform/mac/svg/text/text-text-06-t-expected.txt:
        * platform/mac/svg/text/text-text-07-t-expected.txt:
        * platform/mac/svg/text/text-tspan-01-b-expected.txt:
        * svg/custom/glyph-transformation-with-hkern-expected.txt:
        * platform/qt/svg/text/select-x-list-1-expected.txt:
        * platform/qt/svg/text/select-x-list-2-expected.txt:
        * platform/qt/svg/text/select-x-list-3-expected.txt:
        * platform/qt/svg/text/select-x-list-4-expected.txt:
        * platform/qt/svg/text/select-x-list-with-tspans-1-expected.txt:
        * platform/qt/svg/text/select-x-list-with-tspans-2-expected.txt:
        * platform/qt/svg/text/select-x-list-with-tspans-3-expected.txt:
        * platform/qt/svg/text/select-x-list-with-tspans-4-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67200 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Unreviewed fix after r67085.

* WebCore.pro: Fix previously moved filenames.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67199 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Adam Barth  <abarth@webkit.org>

        Reviewed by Eric Seidel.

        Inline FrameLoader::committedLoad into its one caller
        https://bugs.webkit.org/show_bug.cgi?id=45481

        This saves one round-trip through the FrameLoader when loading
        documents.

        * loader/DocumentLoader.cpp:
        (WebCore::DocumentLoader::commitLoad):
        * loader/FrameLoader.cpp:
        * loader/FrameLoader.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67198 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Patrick Gansterer  <paroga@paroga.com>

        Reviewed by Adam Roben.

        [WINCE] Remove ImageBuffer::toDataURL implementation
        https://bugs.webkit.org/show_bug.cgi?id=45478

        At the moment ImageBuffer::toDataURL uses JPEGEncoder.h and PNGEncoder.h.
        Both files are not in the repository. So remove the implementation for now.

        * platform/graphics/wince/ImageBufferWince.cpp:
        (WebCore::ImageBuffer::toDataURL):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67197 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Anton Muhin <antonm@chromium.org>

        Unreview fixed.

        Remove duplicate test expectations for media/media-can-play-mpeg4-video.html.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67196 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Alexander Pavlov  <apavlov@chromium.org>

        Unreviewed, build fix.

        * inspector/elements-panel-rewrite-href.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67195 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Andreas Kling  <andreas.kling@nokia.com>

        Reviewed by Antonio Gomes.

        WebGL: Fix GCC warnings about suggested parentheses around && within ||
        https://bugs.webkit.org/show_bug.cgi?id=45534

        * html/canvas/WebGLRenderingContext.cpp:
        (WebCore::WebGLRenderingContext::getVertexAttrib):
        (WebCore::WebGLRenderingContext::readPixels):
        (WebCore::WebGLRenderingContext::texParameter):
        (WebCore::WebGLRenderingContext::handleNPOTTextures):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67194 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Hans Wennborg  <hans@chromium.org>

        Reviewed by Jeremy Orlow.

        Un-skip DeviceOrientation layout tests for Chromium.
        https://bugs.webkit.org/show_bug.cgi?id=45531

        fast/dom/DeviceOrientation/window-property.html and
        fast/dom/Window/window-properties-device-orientation.html need to be
        rebaselined, which will be done in a subsequent patch.

        * platform/chromium/test_expectations.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67193 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Anton Muhin  <antonm@chromium.org>

        Reviewed by Kent Tamura.

        [Chromium] media/media-can-play-mpeg4-video.html fails under Chromium
        https://bugs.webkit.org/show_bug.cgi?id=45102

        Chromium is not going to support mpeg, so media/media-can-play-mpeg4-video.html
        is expected to fail.

        * platform/chromium/test_expectations.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67192 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  Alexander Pavlov  <apavlov@chromium.org>

        Reviewed by Yury Semikhatsky.

        Web Inspector: Add navigation items into the Elements panel context menu.

        For all anchors that have the class of "webkit-html-resource-link" or "webkit-html-external-link",
        a custom context menu with the items "Open Link in New Window" (invokes InspectorBackend.openInInspectedWindow() with
        the related URL) and "Open Link in Resources Panel" (opens the related resource in the Resources panel) is shown
        (the latter item is displayed only if the corresponding resource is known to the Web Inspector).
        https://bugs.webkit.org/show_bug.cgi?id=34250

        * English.lproj/localizedStrings.js:
        * inspector/front-end/ElementsPanel.js:
        (WebInspector.ElementsPanel.prototype.populateHrefContextMenu):
        * inspector/front-end/ElementsTreeOutline.js:
        (WebInspector.ElementsTreeOutline.prototype._contextMenuEventFired):
        (WebInspector.ElementsTreeElement.prototype._attributeHTML):
        * inspector/front-end/StylesSidebarPane.js:
        (WebInspector.StylesSidebarPane.prototype._contextMenuEventFired):
        * inspector/front-end/inspector.js:
        (WebInspector.openResource):
        (WebInspector.resourceURLForRelatedNode):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67191 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Andreas Kling  <andreas.kling@nokia.com>

        Reviewed by Tor Arne Vestbø.

        [Qt] Unskip canvas/philip/tests/2d.text.draw.fontface.repeat.html

        This test was fixed by <http://trac.webkit.org/changeset/62016>

        * platform/qt/Skipped:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67190 268f45cc-cd09-0410-ab3c-d52691b4dbfc

An individual renderer should be assigned to each SVGFE*Element class
https://bugs.webkit.org/show_bug.cgi?id=43954

Reviewed by Nikolas Zimmermann.

Updating the description of the following test (it does not fail anymore):

* platform/mac/svg/dynamic-updates/SVGFEDiffuseLightingElement-lighting-color-css-prop-expected.checksum:
* platform/mac/svg/dynamic-updates/SVGFEDiffuseLightingElement-lighting-color-css-prop-expected.png:
* svg/dynamic-updates/SVGFEDiffuseLightingElement-lighting-color-css-prop-expected.txt:
* svg/dynamic-updates/script-tests/SVGFEDiffuseLightingElement-lighting-color-css-prop.js:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67189 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-08  Darin Adler  <darin@apple.com>

        Reviewed by Adam Barth.

        Move functions from Frame to Editor as planned
        https://bugs.webkit.org/show_bug.cgi?id=45218

        * src/ContextMenuClientImpl.cpp:
        (WebKit::selectMisspelledWord):
        (WebKit::ContextMenuClientImpl::getCustomMenuFromDefaultItems):
        * src/WebFrameImpl.cpp:
        (WebKit::WebFrameImpl::find):
        (WebKit::WebFrameImpl::stopFinding):
        (WebKit::WebFrameImpl::scopeStringMatches):
        * src/WebViewImpl.cpp:
        (WebKit::WebViewImpl::caretOrSelectionBounds):
        Changed call sites to use editor().

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67188 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Kwang Yul Seo  <skyul@company100.net>

        Reviewed by James Robinson.

        Make sure skia is not Chromium specific
        https://bugs.webkit.org/show_bug.cgi?id=39672

        Move Image::loadPlatformResource to platform/graphics/chromium/ImageChromium.cpp.
        Other ports can use ImageSkia.cpp.

        No new tests because this is pure refacotring.

        * WebCore.gyp/WebCore.gyp:
        * WebCore.gypi:
        * platform/graphics/chromium/ImageChromium.cpp: Added.
        (WebCore::Image::loadPlatformResource):
        * platform/graphics/skia/ImageSkia.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67187 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Unreviewed. expectations update.

fast/dom/replaceChild.html image failure.

* platform/chromium/test_expectations.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67186 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  Ilya Tikhonovsky  <loislo@chromium.org>

        Reviewed by Joseph Pecoraro.

        WebInspector: Selecting 'inspect element' always shows body in inspector instead of clicked element.

        1. Open any page
        2. Right-click on any element, select "Inspect Element"
        3. Web Inspector is opened but the <body> element (for HTML) is focused rather than the selected element
        4. Inspect the Web Inspector. An error message is logged in the console.

        The regression was introduced in http://trac.webkit.org/changeset/66012 - focusNode() is invoked in
        InspectorController::populateScriptObjects() before the main document is set for the InspectorDOMAgent
        instance (which makes focusNode() bail out prematurely).

        https://bugs.webkit.org/show_bug.cgi?id=45374

        * inspector/InspectorController.cpp:
        (WebCore::InspectorController::populateScriptObjects):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67185 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Cris Neckar  <cdn@chromium.org>

        Reviewed by Chris Fleizach.

        Test for null deref in RenderBlock::updateFirstLetter().
        https://bugs.webkit.org/show_bug.cgi?id=45404

        * fast/dom/firstletter-tablecell-crash-expected.txt: Added.
        * fast/dom/firstletter-tablecell-crash.html: Added.
2010-09-10  Cris Neckar  <cdn@chromium.org>

        Reviewed by Chris Fleizach.

        Null deref in RenderBlock::updateFirstLetter(). Check for null psuedo style.
        https://bugs.webkit.org/show_bug.cgi?id=45404

        Test: fast/dom/firstletter-tablecell-crash.html

        * rendering/RenderBlock.cpp:
        (WebCore::RenderBlock::updateFirstLetter):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67184 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Cris Neckar  <cdn@chromium.org>

        Reviewed by Chris Fleizach.

        Test for null deref in RenderBlock::findNextLineBreak().
        https://bugs.webkit.org/show_bug.cgi?id=45350

        * fast/dom/firstline-fixed-crash-expected.txt: Added.
        * fast/dom/firstline-fixed-crash.html: Added.
2010-09-10  Cris Neckar  <cdn@chromium.org>

        Reviewed by Chris Fleizach.

        Null deref in RenderBlock::findNextLineBreak(). Check for null style with
        first-line style and fixed position.
        https://bugs.webkit.org/show_bug.cgi?id=45350

        Test: fast/dom/firstline-fixed-crash.html

        * rendering/RenderBlockLineLayout.cpp:
        (WebCore::RenderBlock::findNextLineBreak):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67183 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Eric Seidel  <eric@webkit.org>

        Reviewed by Dimitri Glazkov.

        Remove support for non-lazy attach and fix frames to load from insertedIntoDocument instead of attach
        https://bugs.webkit.org/show_bug.cgi?id=45365

        * fast/dom/beforeload/remove-frame-in-beforeload-listener.html:
          - Add a comment to make clear that this test intentionally
            does not cancel the load in the beforeload event, but rather
            relies on WebCore's own Node::willRemove() implementation to
            cancel the load when the frame is removed.
            (The load is not actually directly canceled per say, but the
            loading logic checks to see if the frame is still in the frame
            tree right after creating the frame and aborts if its not.
            HTMLFrameOwnerElement::willRemove clears the frame owner pointer
            causing the frame to be removed from the tree.)
2010-09-10  Eric Seidel  <eric@webkit.org>

        Reviewed by Dimitri Glazkov.

        Remove support for non-lazy attach and fix frames to load from insertedIntoDocument instead of attach
        https://bugs.webkit.org/show_bug.cgi?id=45365

        This change is the last piece of the puzzle which was preventing us from
        removing all eager-attach logic and moving WebCore to using an entirely
        recalcStyle-driven, on-demand renderer creation system, instead of every
        element being synchronously attached during parsing, regardless of whether
        it was ever going to be displayed or not.

        This does not change the places we call lazyAttach vs. attach.  This only
        changes the behavior of frame/plugin elements when lazyAttach is called.
        Previously lazyAttach would eager-attach those elements (and their ancestors)
        because they were marked as canLazyAttach() { return false; }.

        This is a very tricky change, please review carefully.

        Mostly I'm moving logic which used to be in attach() into
        insertedIntoDocument.  Once it was there, there was no longer any reason
        why frame elements couldn't lazyAttach, thus removing the need
        for the non-lazy attach code path entirely.
        We've not yet converted all callsites over to using lazyAttach() instead
        of attach() however.

        In order to move frame loading logic into insertedIntoDocument
        instead of attach, I needed to make sure that any javascript calls
        during the load would see an attached element.  Thus I needed to mark
        the element as needing style resolve so that it would attach itself
        if needed.

        I was not able to just call lazyAttach() from insertedIntoDocument directly
        due to two conflicting assumptions in the rendering tree:
         1. ContainerNode::attach() assumes its "safe" to call attach on all children
            without checking first if the child is already attached.  This seems sane
            since its strange to think of a subtree as being attached w/o ancestors
            already being attached.  Although it is a hard rule that subtrees may not
            have renderers w/o ancestors also having renderers, I do not believe it's
            a hard rule that subtrees may not be attached.  Remember, attached() does
            not imply renderer()!  It's possible ContainerNode::attach()'s assumption is wrong here.
         2. Node::attach() asserts !attached().  This makes sense and I would not
            want to remove this assert, however it means that if insertedIntoDocument
            were to call lazyAttach() (thus marking the element as attached()) then
            all callers would need to check if the element was already attached after
            insertedIntoDocument (called by appendChild, parserAppendChild, etc.)
            before calling attach or lazyAttach().  The following example:
            element.innerHTML = "<span><iframe></span>" is one case where this
            ASSERT would be hit if insertedIntoDocument called lazyAttach, since
            ContainerNode::attach() (called on the span by in appendChild(DocumentFragment) code)
            does not check if iframe is already attached.

        Note: One subtle change here is that synchronous javascript which results
        from javascript: or beforeload is now run as part of insertedIntoDocument
        (thus any insert/append call *even* parserAddChild) instead of being
        run during attach (technically in the post-attach callbacks).

        This is covered by numerous layout tests.

        * dom/ContainerNode.cpp:
        (WebCore::willRemoveChild):
        (WebCore::willRemoveChildren):
         - Since insertedIntoDocument starts a load and yet does not mark the
           element as attached, we need to always call willRemove().
           See note above as to why we don't just mark attached() in insertedIntoDocument.
        * dom/Node.cpp:
        (WebCore::Node::markAncestorsWithChildNeedsStyleRecalc):
         - Share some code between lazyAttach and setNeedsStyleRecalc.
        (WebCore::Node::setNeedsStyleRecalc):
         - Use the new markAncestorsWithChildNeedsStyleRecalc
        (WebCore::Node::lazyAttach):
         - Remove the non-lazy code path, and use markAncestorsWithChildNeedsStyleRecalc.
         - Add an option to lazyAttach without marking attached(), used by HTMLFrameElementBase::insertedIntoDocument.
        * dom/Node.h:
        * html/HTMLFrameElementBase.cpp:
         - m_shouldOpenURLAfterAttach is no longer needed, yay!
         - m_checkAttachedTimer no longer has anything to do with attached(), so renamed it.
           I also documented that the newly named m_checkInDocumentTimer is all about the
           "magic iframe" performance quirk.  (Which is actually speced in HTML5).
           I was initially baffled by this code, so I documented it.
        (WebCore::HTMLFrameElementBase::HTMLFrameElementBase)
        (WebCore::HTMLFrameElementBase::insertedIntoDocument):
         - This is the meat of this change, see discussion above.
        (WebCore::HTMLFrameElementBase::attach):
         - Code deleted or moved to insertedIntoDocument.
        (WebCore::HTMLFrameElementBase::width):
         - Fixed a bug in height()/width() which was probably causing crashes
           and was causing incorrect behavior after this change.
           renderBox() is not necessarily valid unless layout is up to date.
           Updating layout, can cause renderBox() to go invalid, thus this
           could have been a null-pointer crash.
        (WebCore::HTMLFrameElementBase::height): see width()
        (WebCore::HTMLFrameElementBase::setRemainsAliveOnRemovalFromTree): Timer rename.
        (WebCore::HTMLFrameElementBase::checkInDocumentTimerFired): Timer rename.
        * html/HTMLFrameElementBase.h:
        * html/HTMLFrameOwnerElement.cpp:
        (WebCore::HTMLFrameOwnerElement::willRemove):
         - Disconnecting the owner element removes the frame from the frame tree.
           frameDetached() calls Page::frameCount which expects that the frame is
           already gone at this point and asserts when it's not.  It's unclear how
           this worked before, except that the frame removal was likely done in the
           post-attach callback, so the frameCount was wrong (too high) during
           frameDetached(), but was fixed up in the post-detach callback.
        * html/parser/HTMLConstructionSite.cpp:
        (WebCore::HTMLConstructionSite::attachAtSite):
          - Simplified this code, and added a check for the case when the node was already removed.
            Since the load logic is now run during insertedIntoDocument instead of attach(),
            synchronous javascript is now running during insertedIntoDocument and we need to
            make sure that the child is still in the tree.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67182 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Kent Tamura  <tkent@chromium.org>

        Unreviewed, test expectation update.

        * platform/chromium/drt_expectations.txt: Add some comments and sort.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67181 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Ryosuke Niwa  <rniwa@webkit.org>

        Unreviewed.

        Added editing/selection/doubleclick-inline-first-last-contenteditable.html to the qt's skipped list.
        See https://bugs.webkit.org/show_bug.cgi?id=45525 for the detail.

        * platform/qt/Skipped:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67180 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Dirk Pranke  <dpranke@chromium.org>

        Unreviewed, rolling out r67178.
        http://trac.webkit.org/changeset/67178
        https://bugs.webkit.org/show_bug.cgi?id=45524

        re-roll-out patch to put original changes back in

        * fast/replaced/object-param-url-control-char.html:
        * fast/replaced/object-with-embed-url-param-expected.txt:
        * fast/replaced/object-with-embed-url-param.html:
2010-09-10  Dirk Pranke  <dpranke@chromium.org>

        Unreviewed, rolling out r67178.
        http://trac.webkit.org/changeset/67178
        https://bugs.webkit.org/show_bug.cgi?id=45524

        re-roll-out patch to put original changes back in

        * html/HTMLObjectElement.cpp:
        (WebCore::HTMLObjectElement::parametersForPlugin):
        (WebCore::HTMLObjectElement::updateWidget):
        * loader/SubframeLoader.cpp:
        (WebCore::SubframeLoader::resourceWillUsePlugin):
        * loader/SubframeLoader.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67179 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Dirk Pranke  <dpranke@chromium.org>

        Reviewed by Kent Tamura.

        roll out r66992 and r66997 - possibly introduced a perf slowdown in chromium tests
        https://bugs.webkit.org/show_bug.cgi?id=45524

        * fast/replaced/object-param-url-control-char.html:
        * fast/replaced/object-with-embed-url-param-expected.txt:
        * fast/replaced/object-with-embed-url-param.html:
2010-09-10  Dirk Pranke  <dpranke@chromium.org>

        Reviewed by Kent Tamura.

        roll out r66992 and r66997 - possibly introduced a perf slowdown in chromium tests
        https://bugs.webkit.org/show_bug.cgi?id=45524

        * html/HTMLObjectElement.cpp:
        (WebCore::HTMLObjectElement::parametersForPlugin):
        (WebCore::HTMLObjectElement::updateWidget):
        * loader/SubframeLoader.cpp:
        (WebCore::SubframeLoader::requestFrame):
        * loader/SubframeLoader.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67178 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  Hans Wennborg  <hans@chromium.org>

        Reviewed by Jeremy Orlow.

        Hook up LayoutTestController.setMockDeviceOrientation() in Chromium DumpRenderTree.
        https://bugs.webkit.org/show_bug.cgi?id=45460

        This enables DumpRenderTree to run layout tests for DeviceOrientation.

        Also declare the LayoutTestController destructor out-of-line.
        Otherwise the implicit destructor would cause compiler errors because
        of the OwnPtr<WebKit::WebDeviceOrientationClientMock> member.

        * DumpRenderTree/chromium/LayoutTestController.cpp:
        (LayoutTestController::~LayoutTestController):
        (LayoutTestController::setMockDeviceOrientation):
        (LayoutTestController::deviceOrientationClient):
        * DumpRenderTree/chromium/LayoutTestController.h:
        * DumpRenderTree/chromium/WebViewHost.cpp:
        (WebViewHost::deviceOrientationClient):
        * DumpRenderTree/chromium/WebViewHost.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67177 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Ryosuke Niwa  <rniwa@webkit.org>

        Reviewed by Tony Chang.

        Extract the code to find special ancestors in createMarkup into a function
        https://bugs.webkit.org/show_bug.cgi?id=45449

        Extracted the code to find special common ancestors in createMarkup as highestAncestorToWrapMarkup.
        Also isolated the code to find ancestors required to retain structure and appearance into
        ancestorToRetainStructureAndAppearance.

        No new tests are added since this is a cleanup.

        * editing/markup.cpp:
        (WebCore::ancestorToRetainStructureAndAppearance): Added.
        (WebCore::propertyMissingOrEqualToNone): Moved.
        (WebCore::isElementPresentational): Moved.
        (WebCore::shouldIncludeWrapperForFullySelectedRoot): Added. isSpecialAncestorBlock is merged into this function.
        (WebCore::highestAncestorToWrapMarkup): Extracted from createMarkup.
        (WebCore::createMarkup): Calls highestAncestorToWrapMarkup.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67176 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-10  Ryosuke Niwa  <rniwa@webkit.org>

        Unreviewed.

        Added editing/spelling/spelling-backspace-between-lines.html to the skipped list for WebKit2
        since the test uses textInputController, which hasn't been implemented yet in WebKitTestRunner.

        * platform/mac-wk2/Skipped:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67175 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  Adam Barth  <abarth@webkit.org>

        Reviewed by Dimitri Glazkov.

        Spurious null check in MediaDocumentParser
        https://bugs.webkit.org/show_bug.cgi?id=45461

        Trivial change.  FrameLoader can never be null.

        * html/MediaDocument.cpp:
        (WebCore::MediaDocumentParser::createDocumentStructure):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67174 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  Dirk Pranke  <dpranke@chromium.org>

        Unreviewed, expectations update.

        * platform/chromium/test_expectations.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67173 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  Kent Tamura  <tkent@chromium.org>

        Unreviewed, second attempt to fix a Qt test failure.

        * fast/forms/input-number-unacceptable-style-expected.txt:
        * fast/forms/input-number-unacceptable-style.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67172 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  Kent Tamura  <tkent@chromium.org>

        Unreviewed, attempt to fix a Qt test failure.

        * fast/forms/input-number-unacceptable-style.html: Set background-color explicitly.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67171 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  Ryosuke Niwa  <rniwa@webkit.org>

        Reviewed by Eric Seidel.

        REGRESSION: applying new font size causes font-size outside selection to change
        https://bugs.webkit.org/show_bug.cgi?id=45026

        The regression was caused by removeImplicitlyStyledElement not converting font size to CSS value properly.
        Namely, it was treating font size as a pixel value for font-size property.  Fixed this by adding fontSizeToCSSValue
        which uses HTMLFontElement::cssValueFromFontSizeNumber and deployed it in removeImplicitlyStyledElement

        Also fixed a bug in StyleChange::extractTextStyles that it ignores font-size property when the values were
        relative terms such as x-small and -webkit-xxx-large. And replaced the logic to convert from pixel font size
        to legacy font size by CSSStyleSelector::legacyFontSize.

        Test: editing/style/push-down-font-styles.html

        * editing/ApplyStyleCommand.cpp:
        (WebCore::StyleChange::extractTextStyles): Fixed a bug. See above.
        (WebCore::stringToCSSValue): Converts string to CSSValue.
        (WebCore::fontSizeToCSSValue): Converts font size number (String) to CSSPrimitiveValue.
        (WebCore::ApplyStyleCommand::removeImplicitlyStyledElement): Calls stringToCSSValue and fontSizeToCSSValue.
        * css/CSSStyleSelector.h: Made fontSizeForKeyword public.
2010-09-09  Ryosuke Niwa  <rniwa@webkit.org>

        Reviewed by Eric Seidel.

        REGRESSION: applying new font size causes font-size outside selection to change
        https://bugs.webkit.org/show_bug.cgi?id=45026

        Added a test to apply font-size and font-family on various markups.

        * editing/deleting/delete-br-012-expected.txt: Conflicting font and span elements are merged into one font element.
        * editing/inserting/5994480-2-expected.txt: Removed redundant size attribute from font element.
        * editing/deleting/delete-select-all-001-expected.txt: Style spans are replaced by font elements.
        * editing/style/block-style-004-expected.txt: Ditto.
        * editing/style/block-style-005-expected.txt: Ditto.
        * editing/style/block-style-006-expected.txt: Ditto.
        * editing/style/fontsize-1-expected.txt: Ditto.
        * editing/execCommand/query-font-size-expected.txt: Ditto; also fixed the test not to add extraneous spans.
        * editing/style/push-down-font-styles-expected.txt: Added.
        * editing/style/push-down-font-styles.html: Added.
        * editing/style/script-tests/push-down-font-styles.js: Added.
        (testSingleToggle):
        (selectAll):
        (selectTest):
        (selectFirstWord):
        (selectSecondWord):
        (selectLastTwoWords):
        * platform/chromium-win/editing/inserting/5994480-2-expected.txt: Removed extraneous span.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67170 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  James Kozianski  <koz@chromium.org>

        Reviewed by Ojan Vafai.

        Test expectation update.

        * platform/chromium/test_expectations.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67169 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  Kent Tamura  <tkent@chromium.org>

        Unreviewed, build fix for Qt.

        * css/CSSStyleSelector.cpp:
        (WebCore::CSSStyleSelector::SelectorChecker::checkOneSelector):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67168 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  Dominic Cooney  <dominicc@google.com>

        Reviewed by Kent Tamura.

        Moves location.replace bindings logic into bindings/generic and
        instantiates it for JSC and V8.

        https://bugs.webkit.org/show_bug.cgi?id=44891

        Covered by existing location.replace tests.

        * GNUmakefile.am:
        * WebCore.gypi:
        * WebCore.pro:
        * WebCore.vcproj/WebCore.vcproj:
        * WebCore.vcproj/WebCoreCommon.vsprops:
        * WebCore.xcodeproj/project.pbxproj:
        * bindings/generic/BindingDOMWindow.h:
        * bindings/generic/BindingFrame.h: Added.
        (WebCore::::navigateIfAllowed):
        * bindings/generic/BindingLocation.h: Added.
        (WebCore::::replace):
        * bindings/generic/GenericBinding.h:
        (WebCore::completeURL):
        * bindings/js/JSBinding.h: Added.
        * bindings/js/JSBindingsAllInOne.cpp:
        * bindings/js/JSDOMBinding.cpp:
        (WebCore::shouldAllowNavigation):
        (WebCore::toLexicalFrame):
        (WebCore::toDynamicFrame):
        (WebCore::processingUserGesture):
        (WebCore::completeURL):
        * bindings/js/JSLocationCustom.cpp:
        (WebCore::navigateIfAllowed):
        (WebCore::JSLocation::replace):
        * bindings/js/specialization/JSBindingState.cpp: Added.
        (WebCore::::getActiveFrame):
        (WebCore::::getFirstFrame):
        (WebCore::::processingUserGesture):
        (WebCore::::allowsAccessFromFrame):
        * bindings/js/specialization/JSBindingState.h: Added.
        * bindings/v8/V8Binding.h:
        * bindings/v8/V8Utilities.cpp:
        (WebCore::completeURL):
        (WebCore::navigateIfAllowed):
        * bindings/v8/custom/V8LocationCustom.cpp:
        (WebCore::V8Location::replaceCallback):
        * bindings/v8/specialization/V8BindingState.cpp:
        (WebCore::::allowsAccessFromFrame):
        * bindings/v8/specialization/V8BindingState.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67167 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  Kent Tamura  <tkent@chromium.org>

        Reviewed by Dimitri Glazkov.

        Apply :invalid CSS class to <input type=number> with an unacceptable value
        https://bugs.webkit.org/show_bug.cgi?id=45376

        * fast/forms/input-number-unacceptable-style-expected.txt: Added.
        * fast/forms/input-number-unacceptable-style.html: Added.
2010-09-09  Kent Tamura  <tkent@chromium.org>

        Reviewed by Dimitri Glazkov.

        Apply :invalid CSS class to <input type=number> with an unacceptable value
        https://bugs.webkit.org/show_bug.cgi?id=45376

        Apply :invalid CSS class to <input type=number> elements with an
        unacceptable value in order to tell users that a value is not
        valid.

        Introducing Element::hasUnaccceptableValue(), and CSSStyleSelector
        applies :invalid to not only elements with !isValidFormControlElement()
        but also elements with hasUnaccceptableValue().

        HTMLInputElement and RenderTextControlSingleLine need some changes
        to update style and to avoid updating renderer value during style
        update.

        Test: fast/forms/input-number-unacceptable-style.html

        * css/CSSStyleSelector.cpp:
        (WebCore::CSSStyleSelector::SelectorChecker::checkOneSelector):
        * dom/Element.h:
        (WebCore::Element::hasUnacceptableValue):
        * html/HTMLInputElement.cpp:
        (WebCore::HTMLInputElement::handleBlurEvent):
        (WebCore::HTMLInputElement::hasUnacceptableValue):
        * html/HTMLInputElement.h:
        * rendering/RenderTextControlSingleLine.cpp:
        (WebCore::RenderTextControlSingleLine::subtreeHasChanged):
        (WebCore::RenderTextControlSingleLine::updateFromElement):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67166 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Unreviewed. Update the ChangeLog entry of r67164.
The ChangeLog entry had an old method named which was renamed in the code review.
https://bugs.webkit.org/show_bug.cgi?id=43973

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67165 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  Kent Tamura  <tkent@chromium.org>

        Reviewed by Dimitri Glazkov.

        <input type=number> should not have an invalid number as its value
        https://bugs.webkit.org/show_bug.cgi?id=43973

        * fast/forms/ValidityState-rangeOverflow-expected.txt:
        * fast/forms/ValidityState-rangeUnderflow-expected.txt:
        * fast/forms/ValidityState-typeMismatch-number-expected.txt:
        * fast/forms/input-number-commit-valid-only-expected.txt: Added.
        * fast/forms/input-number-commit-valid-only.html: Added.
        * fast/forms/script-tests/ValidityState-typeMismatch-number.js:
        * fast/forms/script-tests/input-number-commit-valid-only.js: Added.
2010-09-09  Kent Tamura  <tkent@chromium.org>

        Reviewed by Dimitri Glazkov.

        <input type=number> should not have an invalid number as its value
        https://bugs.webkit.org/show_bug.cgi?id=43973

        This change implements the value sanitization algorithm for
        type=number, and remove validity.typeMismatch support for
        type=number in order to improve HTML5 conformance.

        HTMLInputElement::value for type=number always has a valid number
        string or an empty string. However, the input field for it, a
        RenderTextControlSingleLine, can have a non-number string. For
        example, '-' is a non-number string, but a user needs to type
        '-'. So, the string in the input field has never been committed to
        HTMLInputElement until the string becomes a valid number string.

        This change is also a preparation of supporting localized
        numbers. A localized string in the input field would not be
        matched with HTMLInputElement::value.

        Test: fast/forms/input-number-commit-valid-only.html

        * dom/InputElement.h: Add isCommittableValue()
        * html/HTMLInputElement.cpp:
        (WebCore::HTMLInputElement::handleBlurEvent):
         Set formControlValueMatchesRenderer false to clean an invalid value in the renderer.
        (WebCore::HTMLInputElement::isCommittableValue):
        (WebCore::HTMLInputElement::sanitizeValue):
        * html/HTMLInputElement.h:
        * html/ValidityState.cpp:
        (WebCore::ValidityState::typeMismatch): Always returns false for type=number.
        * rendering/RenderTextControlSingleLine.cpp:
        (WebCore::RenderTextControlSingleLine::subtreeHasChanged):
        * wml/WMLInputElement.h:
        (WebCore::WMLInputElement::isCommittableValue):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67164 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  Tony Gentilcore  <tonyg@chromium.org>

        Reviewed by Adam Barth.

        Support <script async> as specified by HTML5
        https://bugs.webkit.org/show_bug.cgi?id=45310

        * fast/dom/Document/readystate.html: Dynamically inserted scripts now block the load event like in FF4. So move the finishJSTest call to load.
        * fast/dom/Document/readystate-expected.txt: Dynamically inserted script now runs before load during "interactive" phase.
        * fast/dom/HTMLScriptElement/async-inline-script-expected.txt: Added.
        * fast/dom/HTMLScriptElement/async-inline-script.html: Added. The async attribute doesn't apply to inline scripts, so check that they execute in order.
        * fast/dom/HTMLScriptElement/async-onbeforeload-expected.txt: Added.
        * fast/dom/HTMLScriptElement/async-onbeforeload.html: Added. Check that onbeforeload events fire synchronously and that they may be cancelled.
        * fast/dom/HTMLScriptElement/async-write-expected.txt: Added.
        * fast/dom/HTMLScriptElement/async-write.html: Added. Check that writes from async scripts are neutralized.
        * fast/dom/HTMLScriptElement/resources/async.js: Added.
        * http/tests/misc/async-and-defer-script-expected.txt: Added.
        * http/tests/misc/async-and-defer-script.html: Added. The async attributes takes precendence over the defer attribute.
        Check that a script which has both async and defer acts as async by ensuring it loads after DOMContentLoaded.
        A deferred scripts would block DCL.
        * http/tests/misc/resources/async-script.js: Added.
        * http/tests/misc/resources/slow-async-script.cgi: Added.
        * http/tests/misc/script-async-expected.txt: Added.
        * http/tests/misc/script-async.html: Added. Test general execution order of deferred scripts.
        The test checks for multiple valid orders since it is indeterminate.
2010-09-09  Tony Gentilcore  <tonyg@chromium.org>

        Reviewed by Adam Barth.

        Support <script async> as specified by HTML5
        https://bugs.webkit.org/show_bug.cgi?id=45310

        Tests: fast/dom/HTMLScriptElement/async-inline-script.html
               fast/dom/HTMLScriptElement/async-onbeforeload.html
               fast/dom/HTMLScriptElement/async-write.html
               http/tests/misc/async-and-defer-script.html
               http/tests/misc/script-async.html

        * dom/AsyncScriptRunner.cpp:
        (WebCore::AsyncScriptRunner::AsyncScriptRunner):
        (WebCore::AsyncScriptRunner::~AsyncScriptRunner):
        (WebCore::AsyncScriptRunner::executeScriptSoon): Increment the delay count when a task to execute scripts is queued up.
        (WebCore::AsyncScriptRunner::timerFired): decrementLoadEventDelayCount handles calling checkCompleted().
        * dom/AsyncScriptRunner.h:
        (WebCore::AsyncScriptRunner::create):
        * dom/Document.cpp:
        (WebCore::Document::Document):
        * dom/ScriptElement.cpp:
        (WebCore::ScriptElement::insertedIntoDocument): Treats async scripts just like a dynamically inserted script element rather than a parser inserted one.
        * html/parser/HTMLScriptRunner.cpp:
        (WebCore::HTMLScriptRunner::runScript):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67163 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  Dirk Pranke  <dpranke@chromium.org>

        Unreviewed, expectations update.

        inspector/extensions-resources.html isn't SLOW, it's timing out.

        * platform/chromium/test_expectations.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67162 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  Alexey Marinichev  <amarinichev@chromium.org>

        Reviewed by James Robinson.

        [chromium] Disable subpixel rendering in Linux when GPU compositor is active
        https://bugs.webkit.org/show_bug.cgi?id=45087

        This extends windows-only logic to chromium linux as well.

        * platform/graphics/chromium/ContentLayerChromium.cpp:
        (WebCore::ContentLayerChromium::updateContents):
        * platform/graphics/chromium/FontLinux.cpp:
        (WebCore::adjustTextRenderMode): Added a check to see if the compositor is active.
        (WebCore::Font::drawGlyphs):
        (WebCore::Font::drawComplexText):
        * platform/graphics/chromium/LayerRendererChromium.cpp:
        (WebCore::LayerRendererChromium::setRootLayerCanvasSize):
        * platform/graphics/chromium/VideoLayerChromium.cpp:
        (WebCore::VideoLayerChromium::updateContents):
        * platform/graphics/skia/ImageBufferSkia.cpp:
        (WebCore::ImageBuffer::ImageBuffer):
        (WebCore::ImageBuffer::clip):
        * platform/graphics/skia/PlatformContextSkia.cpp:
        Made setDrawingToImageBuffer and isDrawingToImageBuffer available to
        Linux; removed all #if OS(WINDOWS) || OS(LINUX).
        (WebCore::PlatformContextSkia::State::State):
        (WebCore::PlatformContextSkia::PlatformContextSkia):
        (WebCore::PlatformContextSkia::isDrawingToImageBuffer):
        (WebCore::PlatformContextSkia::save):
        (WebCore::PlatformContextSkia::beginLayerClippedToImage):
        (WebCore::PlatformContextSkia::restore):
        (WebCore::PlatformContextSkia::applyClipFromImage):
        * platform/graphics/skia/PlatformContextSkia.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67161 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  Kinuko Yasuda  <kinuko@chromium.org>

        Unreviewed, Windows build fix.

        * WebCore.vcproj/WebCore.vcproj:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67160 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  Kinuko Yasuda  <kinuko@chromium.org>

        Reviewed by Jian Li.

        [FileSystem] Add File and FileWriter accessor methods in FileEntry
        https://bugs.webkit.org/show_bug.cgi?id=45440

        Adding file() and createWrite() methods in FileEntry.
        http://dev.w3.org/2009/dap/file-system/file-dir-sys.html#the-fileentry-interface

        No new tests; this patch doesn't have implementation yet.

        * CMakeLists.txt:
        * DerivedSources.cpp:
        * DerivedSources.make:
        * GNUmakefile.am:
        * WebCore.gypi:
        * WebCore.pri:
        * WebCore.pro:
        * WebCore.vcproj/WebCore.vcproj:
        * WebCore.xcodeproj/project.pbxproj:
        * fileapi/FileEntry.cpp:
        (WebCore::FileEntry::createWriter): Added.
        (WebCore::FileEntry::file): Added.
        * fileapi/FileEntry.h:
        * fileapi/FileEntry.idl:
        * fileapi/FileCallback.h: Added.
        * fileapi/FileCallback.idl: Added.
        * fileapi/FileWriterCallback.h: Added.
        * fileapi/FileWriterCallback.idl: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67159 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  Yuzo Fujishima  <yuzo@google.com>

        Reviewed by Darin Adler.

        Fix for Bug 24742 - CSS tokenizer allows color in 6 hex digit notation to be followed by the next token without separation
        https://bugs.webkit.org/show_bug.cgi?id=24742

        * fast/css/invalid-hex-color-expected.txt: Added.
        * fast/css/invalid-hex-color.html: Added.
        * svg/custom/invalid-fill-hex.svg: Changed to avoid a questionable behavior that this bug fix revealed.
2010-09-09  Yuzo Fujishima  <yuzo@google.com>

        Reviewed by Darin Adler.

        Fix for Bug 24742 - CSS tokenizer allows color in 6 hex digit notation to be followed by the next token without separation
        Stop checking in the flex definition the number of hexadecimal digits for a hex color.
        Instead, rely on the checking in Color::parseHexColor().
        The resultant grammar is actually more close to the one in http://www.w3.org/TR/CSS21/grammar.html .

        https://bugs.webkit.org/show_bug.cgi?id=24742

        Test: fast/css/invalid-hex-color.html

        * css/tokenizer.flex:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67158 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  Chris Rogers  <crogers@google.com>

        Reviewed by Kenneth Russell.

        Add AudioSourceNode files
        https://bugs.webkit.org/show_bug.cgi?id=45010

        No new tests since audio API is not yet implemented.

        * webaudio/AudioSourceNode.h: Added.
        (WebCore::AudioSourceNode::AudioSourceNode):
        * webaudio/AudioSourceNode.idl: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67157 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  James Robinson  <jamesr@chromium.org>

        Reviewed by Kenneth Russell.

        [chromium] Don't call deleteTexture on a texture attached to an FBO that is not currently bound
        https://bugs.webkit.org/show_bug.cgi?id=45504

        The OpenGL semantics for glDeleteTextures() are that if one of the specified texture is attached
        to the currently bound FBO, that FBO's color attachment is removed.  However, if the specified
        texture is attached to a non-current FBO then OpenGL states explicitly that that FBO's color
        attachment is not cleared and that the texture stays alive even though its ID is no longer
        valid to use.  This is not the behavior we want in DrawingBuffer - we actually want the
        texture to be immediately deleted.  This patch makes sure the DrawingBuffer's FBO is bound
        before deleting its color attachment texture.

        * platform/graphics/chromium/DrawingBufferChromium.cpp:
        (WebCore::DrawingBuffer::~DrawingBuffer):
        (WebCore::DrawingBuffer::reset):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67156 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  James Robinson  <jamesr@chromium.org>

        Reviewed by Kenneth Russell.

        [chromium] Chromium mac build should exclude all Skia files in platform/graphics/skia/, not just some
        https://bugs.webkit.org/show_bug.cgi?id=45516

        This excludes platform/graphics/skia/*Skia.(h|cpp) by pattern instead of individual files.
        None of these files should be compiled into the Chromium mac build and the old file list had
        both omissions and stale entries.

        * WebCore.gyp/WebCore.gyp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67155 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  Dirk Pranke  <dpranke@chromium.org>

        Unreviewed, rolling out r67145.
        http://trac.webkit.org/changeset/67145
        https://bugs.webkit.org/show_bug.cgi?id=45431

        broke editing/inserting/5994480-2.html on Chromium Linux

        * editing/deleting/delete-br-012-expected.txt:
        * editing/deleting/delete-select-all-001-expected.txt:
        * editing/execCommand/query-font-size-expected.txt:
        * editing/execCommand/query-font-size.html:
        * editing/inserting/5994480-2-expected.txt:
        * editing/style/block-style-004-expected.txt:
        * editing/style/block-style-005-expected.txt:
        * editing/style/block-style-006-expected.txt:
        * editing/style/fontsize-1-expected.txt:
        * editing/style/push-down-font-styles-expected.txt: Removed.
        * editing/style/push-down-font-styles.html: Removed.
        * editing/style/script-tests/push-down-font-styles.js: Removed.
2010-09-09  Dirk Pranke  <dpranke@chromium.org>

        Unreviewed, rolling out r67145.
        http://trac.webkit.org/changeset/67145
        https://bugs.webkit.org/show_bug.cgi?id=45431

        broke editing/inserting/5994480-2.html on Chromium Linux

        * css/CSSStyleSelector.h:
        * editing/ApplyStyleCommand.cpp:
        (WebCore::StyleChange::init):
        (WebCore::StyleChange::extractTextStyles):
        (WebCore::ApplyStyleCommand::removeImplicitlyStyledElement):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67154 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Removing an obsolete comment.
https://bugs.webkit.org/show_bug.cgi?id=40112

Reviewed by Alexey Proskuryakov.

* bindings/generic/ActiveDOMCallback.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67153 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2010-09-09  James Robinson  <jamesr@chromium.org>

        Reviewed by Kenneth Russell.

        [chromium] Clear only the uploaded portion of the software backing store when in mixed mode
        https://bugs.webkit.org/show_bug.cgi?id=45503

        This clears out only the dirty region of the software backing store when uploading results
        to hardware rather than clearing out the entire software backing store.  This is a significant
        performance improvement when the dirty rects are small relative to the whole canvas.

        This also implements the non-swizzle path of copySubRect() using memcpy() to copy the rows, which
        is another respectable performance increase.

        * html/canvas/CanvasRenderingContext2D.cpp:
        * platform/graphics/gpu/Texture.cpp:
        (WebCore::copySubRect):
        * platform/graphics/skia/PlatformContextSkia.cpp:
        (WebCore::PlatformContextSkia::prepareForSoftwareDraw):
        (WebCore::PlatformContextSkia::uploadSoftwareToHardware):
        (WebCore::PlatformContextSkia::readbackHardwareToSoftware):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@67152 268f45cc-cd09-0410-ab3c-d52691b4dbfc