2009-03-20  Darin Adler  <darin@apple.com>

        Reviewed by Timothy Hatcher.

        * WebView/WebTextIterator.h: Improved comments to point out some of the pitfalls
        of this SPI.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41867 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-20  Jan Michael Alonzo  <jmalonzo@gmail.com>

        Reviewed by Holger Freyther.

        [GTK] Misc patches for WebKitWebHistoryItem
        https://bugs.webkit.org/show_bug.cgi?id=24493

        Added build support for build the WebKitWebHistoryItem unit test.

        * GNUmakefile.am:

2009-03-20  Jan Michael Alonzo  <jmalonzo@gmail.com>

        Reviewed by Holger Freyther.

        [GTK] Misc patches for WebKitWebHistoryItem
        https://bugs.webkit.org/show_bug.cgi?id=24493

        Add unit test for WebKitWebHistoryItem

        * tests/testwebhistoryitem.c: Added.
        (web_history_item_fixture_setup):
        (web_history_item_fixture_teardown):
        (test_webkit_web_history_item_get_data):
        (test_webkit_web_history_item_alternate_title):
        (main):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41866 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WebCore:

2009-03-20  Beth Dakin  <bdakin@apple.com>

        Reviewed by Dave Hyatt.

        Fix for https://bugs.webkit.org/show_bug.cgi?id=20909 REGRESSION
        (r35318): A press release at pfizer.com does not display correctly
        - and corresponding -
        <rdar://problem/6680073>

        * rendering/RenderObject.cpp:
        (WebCore::RenderObject::invalidateContainerPrefWidths):
        * rendering/RenderObject.h:
        (WebCore::RenderObject::markContainingBlocksForLayout):

LayoutTests:

2009-03-20  Beth Dakin  <bdakin@apple.com>

        Reviewed by Dave Hyatt.

        Test for https://bugs.webkit.org/show_bug.cgi?id=20909 REGRESSION
        (r35318): A press release at pfizer.com does not display correctly
        - and corresponding -
        <rdar://problem/6680073>

        * http/tests/misc/generated-content-inside-table.html: Added.
        * platform/mac/http/tests/misc/generated-content-inside-table-expected.checksum: Added.
        * platform/mac/http/tests/misc/generated-content-inside-table-expected.png: Added.
        * platform/mac/http/tests/misc/generated-content-inside-table-expected.txt: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41865 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-20  Jan Michael Alonzo  <jmalonzo@gmail.com>

        Reviewed by Holger Freyther.

        Separate gtk unit tests
        https://bugs.webkit.org/show_bug.cgi?id=24039

        Build the unit tests accordingly.

        * GNUmakefile.am:

2009-03-20  Jan Michael Alonzo  <jmalonzo@gmail.com>

        Reviewed by Holger Freyther.

        Separate gtk unit tests
        https://bugs.webkit.org/show_bug.cgi?id=24039

        Split the current single-file unit test to make it more
        modularized and manageable in the future as more unit tests are
        written.

        * tests/main.c: Removed.
        * tests/testwebbackforwardlist.c: Copied from WebKit/gtk/tests/main.c.
        (main):
        * tests/testwebframe.c: Copied from WebKit/gtk/tests/main.c.
        (main):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41864 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by Justin Garcia.

        maxDeepOffset is confusing and should be removed
        https://bugs.webkit.org/show_bug.cgi?id=24586

        Abstract some hard-to-read (but shared) logic into a new renderedAsNonInlineTableOrHR function.
        Add first/lastDeepEditingPositionForNode Position creation functions
        and deploy them to places we used to call maxDeepOffset.

        Rename Position::atStart and atEnd to atStartOfTree atEndOfTree
        Add a new Position::atFirst/atLastEditingPositionForNode() and use these
        to replace a few more callers for maxDeepOffset()

        Rename maxDeepOffset to lastEditingOffsetForNode (so that we mere mortals have some clue what it does)

        "Editing positions" are confusing because they have one
        of two behaviors, depending on if the container node is ignored
        by editing (if editingIgnoresContent(node) returns true) or not.
        Positions referring to nodes ignored by editing are
        neighbor-relative (they are before or after the node) where as
        positions reffering to other nodes are container-relative
        (they are between two child nodes of the container, identified
        by the offset() member).  I will be fixing this confusion in
        future patches.  These renames hopefully make the current behavior clearer.

        * dom/Position.cpp:
        (WebCore::Position::previous):
        (WebCore::Position::next):
        (WebCore::Position::atFirstEditingPositionForNode):
        (WebCore::Position::atLastEditingPositionForNode):
        (WebCore::Position::atStartOfTree):
        (WebCore::Position::atEndOfTree):
        (WebCore::Position::previousCharacterPosition):
        (WebCore::Position::nextCharacterPosition):
        (WebCore::Position::upstream):
        (WebCore::Position::isCandidate):
        (WebCore::firstDeepEditingPositionForNode):
        (WebCore::lastDeepEditingPositionForNode):
        * dom/Position.h:
        * dom/PositionIterator.cpp:
        (WebCore::PositionIterator::operator Position):
        (WebCore::PositionIterator::increment):
        (WebCore::PositionIterator::decrement):
        (WebCore::PositionIterator::atEnd):
        (WebCore::PositionIterator::atEndOfNode):
        * editing/CompositeEditCommand.cpp:
        (WebCore::CompositeEditCommand::positionAvoidingSpecialElementBoundary):
        * editing/DeleteSelectionCommand.cpp:
        (WebCore::isTableCellEmpty):
        (WebCore::DeleteSelectionCommand::removeNode):
        (WebCore::DeleteSelectionCommand::handleGeneralDelete):
        * editing/Editor.cpp:
        (WebCore::Editor::advanceToNextMisspelling):
        * editing/InsertLineBreakCommand.cpp:
        (WebCore::InsertLineBreakCommand::doApply):
        * editing/InsertListCommand.cpp:
        (WebCore::InsertListCommand::doApply):
        * editing/ReplaceSelectionCommand.cpp:
        (WebCore::ReplaceSelectionCommand::positionAtEndOfInsertedContent):
        * editing/TypingCommand.cpp:
        (WebCore::TypingCommand::forwardDeleteKeyPressed):
        * editing/VisiblePosition.cpp:
        (WebCore::VisiblePosition::previous):
        (WebCore::VisiblePosition::leftVisuallyDistinctCandidate):
        (WebCore::VisiblePosition::left):
        (WebCore::VisiblePosition::rightVisuallyDistinctCandidate):
        (WebCore::VisiblePosition::right):
        * editing/VisibleSelection.cpp:
        (WebCore::VisibleSelection::selectionFromContentsOfNode):
        (WebCore::VisibleSelection::adjustSelectionToAvoidCrossingEditingBoundaries):
        * editing/htmlediting.cpp:
        (WebCore::nextVisuallyDistinctCandidate):
        (WebCore::previousVisuallyDistinctCandidate):
        (WebCore::firstEditablePositionAfterPositionInRoot):
        (WebCore::lastEditablePositionBeforePositionInRoot):
        (WebCore::lastOffsetForEditing):
        (WebCore::isFirstPositionAfterTable):
        (WebCore::isLastPositionBeforeTable):
        (WebCore::positionBeforeNode):
        (WebCore::positionAfterNode):
        (WebCore::enclosingEmptyListItem):
        (WebCore::caretMaxOffset):
        * editing/htmlediting.h:
        * editing/visible_units.cpp:
        (WebCore::renderedAsNonInlineTableOrHR):
        (WebCore::startOfParagraph):
        (WebCore::endOfParagraph):
        (WebCore::startOfEditableContent):
        (WebCore::endOfEditableContent):
        * page/AccessibilityObject.cpp:
        (WebCore::endOfStyleRange):
        * page/AccessibilityRenderObject.cpp:
        (WebCore::AccessibilityRenderObject::visiblePositionRange):
        * rendering/RenderBox.cpp:
        (WebCore::RenderBox::positionForPoint):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41863 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-20  Dmitry Titov  <dimich@chromium.org>

        Reviewed by Dimitri Glazkov.

        https://bugs.webkit.org/show_bug.cgi?id=24689
        Add (upstream) V8 bindings for Workers. Mostly style cleaning.

        * bindings/v8/WorkerContextExecutionProxy.cpp: Added.
        * bindings/v8/WorkerContextExecutionProxy.h: Added.
        * bindings/v8/WorkerScriptController.cpp: Added.
        * bindings/v8/WorkerScriptController.h: Added.
        * bindings/v8/V8Index.h: Added. This is just a wrapper for v8_index.h, like V8Proxy.h
        * bindings/v8/V8Proxy.h: Added domObjectMap() function that wraps GetDOMObjectMap().
        * bindings/v8/V8WorkerContextEventListener.cpp:
        (WebCore::V8WorkerContextEventListener::callListenerFunction): TrackEvent() renamed trackEvent()

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41862 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-20  Erik L. Bunce  <elbunce@xendom.com>

        Reviewed by Tor Arne Vestbø.

        Fix QWebPage::WebActions action states to more closely match when they are
        actually applicable and remove erroneous documentation.

        * Most WebActions implemented using editor commands now use the
          Editor::Command::isEnabled() to control their availability.
        * SelectAll is always enabled (since it's editor command is).
        * SetTextDirection{} family of WebActions are available when canEdit() is true
          and not just canEditRichly().

        Fix and clarify documentation about the availability of various web actions.

        * Api/qwebpage.cpp:
        (QWebPagePrivate::updateAction):
        (QWebPagePrivate::updateEditorActions):
        * tests/qwebpage/tst_qwebpage.cpp:
        (tst_QWebPage::textSelection):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41861 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-20  Stephen White  <senorblanco@chromium.org>

        Reviewed by Eric Seidel.

        Fix for LayoutTests/fast/canvas/canvas-text-alignment.html
        on chromium/skia.  The problem was that the gradient matrix
        for text was being applied twice.  Fixed by reverting some of
        https://bugs.webkit.org/show_bug.cgi?id=23957, so that skiaDrawText
        is no longer responsible for measuring the text and scaling up
        the gradient matrix.  Instead, the text bounding box is passed
        in from SVGPaintServerGradient.  I didn't make this change for CG,
        since it uses a different method (the gradient is drawn using the
        text as a pre-rendered mask).
        https://bugs.webkit.org/show_bug.cgi?id=24687

        * platform/graphics/skia/SkiaFontWin.cpp:
        (WebCore::skiaDrawText):
        * svg/graphics/SVGPaintServerGradient.cpp:
        (WebCore::SVGPaintServerGradient::setup):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41860 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-20  Darin Adler  <darin@apple.com>

        Reviewed by Adele Peterson.

        Use a better technique to handle finding out if something responds to a selector
        in WebHTMLView's doCommandBySelector method.

        * WebView/WebHTMLView.mm:
        (-[WebHTMLView doCommandBySelector:]): Removed unneeded check for 0 coreFrame;
        this is already handled by coreCommandBySelector: so doesn't need to be checked
        twice. Got rid of initial value for eventWasHandled boolean to make it more clear.
        Use WebResponderChainSink to find out if a command is handled rather than walking
        the responder chain explicitly.
        (-[WebResponderChainSink initWithResponderChain:]): Added.
        (-[WebResponderChainSink detach]): Added.
        (-[WebResponderChainSink receivedUnhandledCommand]): Added.
        (-[WebResponderChainSink noResponderFor:]): Added.
        (-[WebResponderChainSink doCommandBySelector:]): Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41859 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Fix the Tiger build, round two.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41858 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Fix the Tiger build.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41857 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-20  Xan Lopez  <xlopez@igalia.com>

        Rubber-stamped by Holger Freyther.

        There seems to be some rounding error in cairo (or in how we use
        cairo) with some fonts, like DejaVu Sans Mono, which makes cairo
        report a height smaller than ascent + descent, which is wrong and
        confuses WebCore's layout system. Workaround this while we figure
        out what's going on.

        * platform/graphics/gtk/SimpleFontDataGtk.cpp:
        (WebCore::SimpleFontData::platformInit):
        * platform/graphics/gtk/SimpleFontDataPango.cpp:
        (WebCore::SimpleFontData::platformInit):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41856 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by Darin Adler.

        https://bugs.webkit.org/show_bug.cgi?id=24122
        <rdar://problem/6674179>
        REGRESSION: DOM Range extractContents/deleteContents failures seen on Moxiecode tests

        Test: fast/dom/Range/deleted-range-endpoints.html

        * dom/Range.cpp: (WebCore::Range::processContents): Set the final range in accordance
        to the specification.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41855 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-19  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Oliver Hunt.

        Fixed <rdar://problem/6033712> -- a little bit of hardening in the Collector.

        SunSpider reports no change. I also verified in the disassembly that
        we end up with a single compare to constant.

        * runtime/Collector.cpp:
        (JSC::Heap::heapAllocate):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41854 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Remove #ifndef BUILDING_ON_TIGER around code that schedules runloop modes
        for Page, so the new RunLoopTimer in WebCore always gets a default mode.
        Fixes the layout test failures on the Tiger build bots.

        Reviewed by Mark Rowe.

        * WebView/WebView.mm:
        (-[WebView _commonInitializationWithFrameName:groupName:usesDocumentViews:]):
        (-[WebView scheduleInRunLoop:forMode:]):
        (-[WebView unscheduleFromRunLoop:forMode:]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41853 268f45cc-cd09-0410-ab3c-d52691b4dbfc

JavaScriptCore:

2009-03-19  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Cameron Zwarich and Oliver Hunt.

        Fixed <rdar://problem/6406045> REGRESSION: Stack overflow on PowerPC on
        fast/workers/use-machine-stack.html (22531)

        Dialed down the re-entry allowance to 64 (from 128).

        On a 512K stack, this leaves about 64K for other code on the stack while
        JavaScript is running. Not perfect, but it solves our crash on PPC.

        Different platforms may want to dial this down even more.

        Also, substantially shrunk BytecodeGenerator. Since we allocate one on
        the stack in order to throw a stack overflow exception -- well, let's
        just say the old code had an appreciation for irony.

        SunSpider reports no change.

        * bytecompiler/BytecodeGenerator.h:
        * interpreter/Interpreter.h:
        (JSC::):

LayoutTests:

2009-03-19  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Cameron Zwarich and Oliver Hunt.

        Re-enabled this test, since it no longer crashes when JavaScriptCore is
        compiled in non-JIT mode.

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

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41852 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-19  Cameron Zwarich  <cwzwarich@uwaterloo.ca>

        Reviewed by Oliver Hunt.

        Bug 24350: REGRESSION: Safari 4 breaks SPAW wysiwyg editor multiple instances
        <https://bugs.webkit.org/show_bug.cgi?id=24350>
        <rdar://problem/6674182>

        The SPAW editor's JavaScript assumes that toString() on a function
        constructed with the Function constructor produces a function with
        a newline after the opening brace.

        JavaScriptCore:

        * runtime/FunctionConstructor.cpp:
        (JSC::constructFunction): Add a newline after the opening brace of the
        function's source code.

        LayoutTests:

        * fast/js/function-constructor-newline-after-brace-expected.txt: Added.
        * fast/js/function-constructor-newline-after-brace.html: Added.
        * fast/js/resources/function-constructor-newline-after-brace.js: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41851 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Fix fast/text/Geeza-Pro-vertical-metrics-adjustment.html on Leopard by moving the results in to the location
that run-webkit-tests currently looks for them.

Reviewed by Stephanie Lewis.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41850 268f45cc-cd09-0410-ab3c-d52691b4dbfc

JavaScriptCore:

2009-03-19  Cameron Zwarich  <cwzwarich@uwaterloo.ca>

        Reviewed by Geoff Garen.

        Bug 23771: REGRESSION (r36016): JSObjectHasProperty freezes on global class without kJSClassAttributeNoAutomaticPrototype
        <https://bugs.webkit.org/show_bug.cgi?id=23771>
        <rdar://problem/6561016>

        * API/tests/testapi.c:
        (main): Add a test for this bug.
        * runtime/JSGlobalObject.cpp:
        (JSC::JSGlobalObject::resetPrototype): Don't set the prototype of the
        last object in the prototype chain to the object prototype when the
        object prototype is already the last object in the prototype chain.

LayoutTests:

2009-03-19  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Oliver Hunt.

        Fixed <rdar://problem/6279213> Regular expression run-time complexity
        limit too low for long inputs (21485)

        Added a test for a complex regexp match that should succeed, and one
        that should fail.

        * fast/js/regexp-overflow-expected.txt:
        * fast/js/resources/regexp-overflow.js:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41849 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Update test results for changes in handling of CMYK.

Reviewed by Sam Weinig.

* fast/canvas/resources/set-colors.js:
* fast/canvas/set-colors-expected.txt:
* platform/mac-leopard/fast/canvas/set-colors-expected.txt: Copied from LayoutTests/fast/canvas/set-colors-expected.txt.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41848 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-19  Anders Carlsson  <andersca@apple.com>

        Reviewed by Dan Bernstein.

        <rdar://problem/6682554> Flash content not being rendered (Shockwave Flash 10.0 r22)

        If calling updateWidget for some reason resulted in another widget being added to m_widgetUpdateSet, then
        that object would never be updated.

        * page/FrameView.cpp:
        (WebCore::FrameView::updateWidgets):
        Factor the widget updating code out into this method. Return true if the update set is empty.

        (WebCore::FrameView::performPostLayoutTasks):
        Loop over the update set multiple times until all widgets have been updated or until we reach the cap.

        * page/FrameView.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41847 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-19  Cameron Zwarich  <cwzwarich@uwaterloo.ca>

        Reviewed by Geoff Garen.

        Bug 23771: REGRESSION (r36016): JSObjectHasProperty freezes on global class without kJSClassAttributeNoAutomaticPrototype
        <https://bugs.webkit.org/show_bug.cgi?id=23771>
        <rdar://problem/6561016>

        * API/tests/testapi.c:
        (main): Add a test for this bug.
        * runtime/JSGlobalObject.cpp:
        (JSC::JSGlobalObject::resetPrototype): Don't set the prototype of the
        last object in the prototype chain to the object prototype when the
        object prototype is already the last object in the prototype chain.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41846 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Make the defered data loading timer honor the Page's scheduled runloop pairs.
        Introduces a new RunLoopTimer class that has an API mimicking Timer but
        allows it to be scheduled with one or more SchedulePairs.

        <rdar://problem/6687342> -[WebView scheduleInRunLoop:forMode:] has no affect on timers

        Reviewed by Darin Adler.

        * WebCore.xcodeproj/project.pbxproj: Adds the new RunLoopTimer.{cpp,h} files.
        * loader/MainResourceLoader.cpp:
        (WebCore::MainResourceLoader::handleDataLoadNow): Use the MainResourceLoaderTimer typedef.
        (WebCore::MainResourceLoader::startDataLoadTimer): Added. Start the timer and on
        Mac platforms also schedule with the Page's SchedulePairs.
        (WebCore::MainResourceLoader::handleDataLoadSoon): Call startDataLoadTimer().
        (WebCore::MainResourceLoader::setDefersLoading): Ditto.
        * loader/MainResourceLoader.h:
        * platform/cf/RunLoopTimerCF.cpp: Added.
        (WebCore::RunLoopTimerBase::~RunLoopTimerBase):
        (WebCore::timerFired):
        (WebCore::RunLoopTimerBase::start):
        (WebCore::RunLoopTimerBase::schedule):
        (WebCore::RunLoopTimerBase::stop):
        (WebCore::RunLoopTimerBase::isActive):
        * platform/RunLoopTimer.h: Added.
        (WebCore::RunLoopTimerBase::RunLoopTimerBase):
        (WebCore::RunLoopTimerBase::startRepeating):
        (WebCore::RunLoopTimerBase::startOneShot):
        (WebCore::RunLoopTimer::RunLoopTimer):
        (WebCore::RunLoopTimer::fired):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41845 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-19 Gustavo Noronha Silva <gns@gnome.org>

Reviewed by Mark Rowe.

        Make determinePassedArchitecture always consume the --32-bit
        option, also in non-AppleMacWebkit platforms. Solution pointed out
        by Mark Rowe.

        * Scripts/webkitdirs.pm:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41844 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Adding scrollbar images for blog post.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41843 268f45cc-cd09-0410-ab3c-d52691b4dbfc

JavaScriptCore:

2009-03-19  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Sam Weinig.

        Fixed <rdar://problem/6603562> REGRESSION (Safari 4): regular expression
        pattern size limit lower than Safari 3.2, other browsers, breaks SAP (14873)

        Bumped the pattern size limit to 1MB, and standardized it between PCRE
        and WREC. (Empirical testing says that we can easily compile a 1MB regular
        expression without risking a hang. Other browsers support bigger regular
        expressions, but also hang.)

        SunSpider reports no change.

        I started with a patch posted to Bugzilla by Erik Corry (erikcorry@google.com).

        * pcre/pcre_internal.h:
        (put3ByteValue):
        (get3ByteValue):
        (put3ByteValueAndAdvance):
        (putLinkValueAllowZero):
        (getLinkValueAllowZero): Made PCRE's "LINK_SIZE" (the number of bytes
        used to record jumps between bytecodes) 3, to accomodate larger potential
        jumps. Bumped PCRE's "MAX_PATTERN_SIZE" to 1MB. (Technically, at this
        LINK_SIZE, we can support even larger patterns, but we risk a hang during
        compilation, and it's not clear that such large patterns are important
        on the web.)

        * wrec/WREC.cpp:
        (JSC::WREC::Generator::compileRegExp): Match PCRE's maximum pattern size,
        to avoid quirks between platforms.

LayoutTests:

2009-03-19  Geoffrey Garen  <ggaren@apple.com>

        Reviewed by Sam Weinig.

        Made two layout tests less agressive, to accomodate a change I made
        for <rdar://problem/6603562> REGRESSION (Safari 4): regular expression
        pattern size limit lower than Safari 3.2, other browsers, breaks SAP (14873)

        * fast/js/regexp-charclass-crash-expected.txt:
        * fast/js/regexp-charclass-crash.html: Explicitly limit the number of iterations
        in the test loop. Otherwise, regular expression engines supporting very
        long patterns take a very very very long time to run this test.

        * fast/js/resources/regexp-overflow.js: Made the "too big" regexp in
        this test even bigger, to match our new limit.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41842 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-19  Dimitri Glazkov  <dglazkov@chromium.org>

        Reviewed by Darin Fisher.

        https://bugs.webkit.org/show_bug.cgi?id=24702
        Upstream miscellaneous bindings changes.

        * bindings/v8/ScheduledAction.cpp:
        (WebCore::ScheduledAction::execute): Changed to call lower-case evaluate.
        * bindings/v8/ScriptCallStack.h: Added an extra include.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41841 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-19  Simon Fraser  <simon.fraser@apple.com>

        Reviewed by Dave Hyatt

        https://bugs.webkit.org/show_bug.cgi?id=24686

        When hit testing a RenderLayer whose parent lives in a preserves-3D hierarchy,
        we need to compare the computed z-offset with the depth-test z-offset before
        deciding that such a RenderLayer was hit. This fixes an issue, tested by the
        3d-point-mapping-overlapping.html test, where the child of a transformed element
        is found by hit testing, even when some other element with greater Z overlaps
        them both.

        Improved the code by adding a utility method, isHitCandidate(), which computes and tests
        z-depth when necessary.

        Tests: transforms/3d/point-mapping/3d-point-mapping-coplanar.html
               transforms/3d/point-mapping/3d-point-mapping-overlapping.html

        * rendering/RenderLayer.cpp:
        (WebCore::isHitCandidate):
        (WebCore::RenderLayer::hitTestLayer):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41840 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-19  Jeremy Moskovich  <jeremy@chromium.org>

        Reviewed by Dimitri Glazkov.

        https://bugs.webkit.org/show_bug.cgi?id=24456
        Split ColorChromium.cpp into Mac & Windows variants.
        Remove Chromium Dependency on platform/graphics/mac/ColorMac.mm since we
        ultimately need to take a different approach.  For now, createCGColor()
        is copied from ColorMac.mm.

        No observable change in behavior, so no test.

        * platform/graphics/chromium/ColorChromium.cpp:
        (WebCore::focusRingColor):
        * platform/graphics/chromium/ColorChromiumMac.mm: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41839 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Fix WebCore/ChangeLog entry, add missing reviewer line.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41838 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-19  Scott Violet  <sky@google.com>

        Reviewed by Simon Fraser.

        https://bugs.webkit.org/show_bug.cgi?id=24678
        Transitions tests somewhat flakey on Chromium.

        Modifies transitions/transition-end-event-helpers.js to only process
        results once.

        * transitions/transition-end-event-helpers.js:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41837 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-19  Pavel Feldman  <pfeldman@chromium.org>

        Reviewed by Dimitri Glazkov.

        https://bugs.webkit.org/show_bug.cgi?id=24675
        Unforking frontend: add custom InspectorController methods
        implementation.

        * bindings/v8/custom/V8InspectorControllerCustom.cpp: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41836 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-19  Jay Campan  <jcampan@google.com>

        https://bugs.webkit.org/show_bug.cgi?id=24625
        Adding an accessor to the currently selected index in the PopupMenuChromium.
        This is required for implementing the deletion of an autocomplete entry in Chromium.

        * platform/chromium/PopupMenuChromium.cpp:
        (WebCore::PopupContainer::selectedIndex):
        * platform/chromium/PopupMenuChromium.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41835 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-19  Evan Stade  <estade@chromium.org>

        Reviewed by Dimitri Glazkov.

        https://bugs.webkit.org/show_bug.cgi?id=24526
        Improve windows skia text stroking.

        Test: LayoutTests/svg/custom/struct-use-09-b.svg

        * platform/graphics/skia/SkiaFontWin.cpp: Close the path representing
        each font glyph polygon, rather than only closing the path once per
        letter. This fixes stroking for letters with multiple polygons, such
        as 'A' or 'D'.
        (WebCore::getPathForGlyph):
        (WebCore::skiaDrawText):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41834 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Oops, bail out last patch, that was incorrect.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41833 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-19  Laurent Montel  <montel@kde.org>

        Reviewed by Simon Hausmann.

        Another fix for pedantic compilation in QtWebKit

        (patch from qt-copy)

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41832 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-19  Ariya Hidayat <ariya.hidayat@trolltech.com>

        Reviewed by Simon Hausmann.

        Fixes pedantic compilation in QtWebKit.

        There are no semi-colons after namespace declarations.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41831 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-19  David Boddie  <dboddie@trolltech.com>

        Reviewed by Simon Hausmann.

        Doc: Removed obsolete warning about Flash and other plugins.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41830 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-19  Paul Olav Tvete  <paul.tvete@nokia.com>

        Reviewed by Simon Hausmann.

        Properly escape tooltip text

        ManualTest: http://xkcd.com/554/

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41829 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-19  Simon Hausmann  <simon.hausmann@nokia.com>

        Reviewed by Tor Arne Vestbø.

        Fixed support for doing calls from JavaScript into NPAPI Plugins for the Qt port on Windows.

        Removed dead code for distinguishing between Widget and PluginView in the Qt port.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41828 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Work around <rdar://problem/6698023> by activating fonts from disk.

Reviewed by Oliver Hunt.

* DumpRenderTree/DumpRenderTree.xcodeproj/project.pbxproj: Copy fonts into DumpRenderTree.resources
in the built products directory.
* DumpRenderTree/mac/DumpRenderTree.mm:
(activateFonts): Activate the fonts from disk.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41827 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-19  Cameron Zwarich  <cwzwarich@uwaterloo.ca>

        Reviewed by Oliver Hunt.

        Bug 24596: ASSERT in JSC::PropertySlot::slotBase @ iGoogle homepage
        <https://bugs.webkit.org/show_bug.cgi?id=24596>
        <rdar://problem/6686493>

        JSDOMWindow::customGetOwnPropertySlot() does an access check after calling
        JSGlobalObject::getOwnPropertySlot(). This causes the PropertySlot to be
        set twice, once to the value that is illegal to access, and then to undefined
        This causes an assertion failure in property access caching code.

        The fix is to do the access check before calling JSGlobalObject::getOwnPropertySlot().

        WebCore:

        * bindings/js/JSDOMWindowCustom.h:
        (WebCore::JSDOMWindow::customGetOwnPropertySlot):

        LayoutTests:

        * http/tests/security/cross-frame-access-get-custom-property-cached-expected.txt: Added.
        * http/tests/security/cross-frame-access-get-custom-property-cached.html: Added.
        * http/tests/security/resources/cross-frame-access-get-custom-property-cached-iframe.html: Added.
        * http/tests/security/resources/cross-frame-access.js:
        (shouldBeUndefined):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41826 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by Sam Weinig.

        https://bugs.webkit.org/show_bug.cgi?id=24676
        Simple cross-origin requests shouldn't dispatch upload progress events

        Test: http/tests/xmlhttprequest/simple-cross-origin-progress-events.html

        * loader/CrossOriginAccessControl.cpp: (WebCore::passesAccessControlCheck): Added a comment
        explaining the somewhat unexpected behavior of this function.

        * xml/XMLHttpRequestUpload.cpp: (WebCore::XMLHttpRequestUpload::hasListeners):
        * xml/XMLHttpRequestUpload.h:
        Report whether there are any event listeners registered.

        * xml/XMLHttpRequest.h: Added m_uploadEventsAllowed.

        * xml/XMLHttpRequest.cpp:
        (WebCore::XMLHttpRequest::createRequest): Set m_uploadEventsAllowed flag.
        (WebCore::XMLHttpRequest::makeSameOriginRequest): Ditto.
        (WebCore::XMLHttpRequest::makeCrossOriginAccessRequest): Ditto.
        (WebCore::XMLHttpRequest::makeSimpleCrossOriginAccessRequest): Set request body - it can be
        non-empty for POST requests.
        (WebCore::XMLHttpRequest::makeCrossOriginAccessRequestWithPreflight): Set m_uploadEventsAllowed flag.
        (WebCore::XMLHttpRequest::handleAsynchronousPreflightResult): Ditto.
        (WebCore::XMLHttpRequest::abort): Only dispatch upload progress events if allowed.
        (WebCore::XMLHttpRequest::networkError): Ditto.
        (WebCore::XMLHttpRequest::abortError): Ditto.
        (WebCore::XMLHttpRequest::didSendData): Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41825 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-18  Marc-Antoine Ruel  <maruel@chromium.org>

        Reviewed by Darin Fisher.

        https://bugs.webkit.org/show_bug.cgi?id=24398
        Fix a crash when loading a svg file in Chromium's test_shell and
        then reloading the page.

        * history/BackForwardListChromium.cpp:
        (WebCore::BackForwardList::BackForwardList):
        (WebCore::BackForwardList::close):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41824 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Rolling out r41818 since it broke the windows build.
        Error: ..\..\runtime\DatePrototype.cpp(30) : fatal error C1083: Cannot open include file: 'langinfo.h': No such file or directory

        * runtime/DatePrototype.cpp:
        (JSC::formatLocaleDate):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41823 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Skip plugins/netscape-throw-exception.html due to <rdar://problem/6530010>.

Reviewed by Sam Weinig.

* platform/mac-snowleopard/Skipped:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41822 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Disable pluginDocumentView-deallocated-dataSource.html as it has not tested
what it was designed to test for some time, and is failing on some machines.
Fixing the test is tracked by <https://bugs.webkit.org/show_bug.cgi?id=24690>.

Rubber-stamped by Dan Bernstein.

* platform/mac/plugins/pluginDocumentView-deallocated-dataSource.html-disabled: Renamed from LayoutTests/platform/mac/plugins/pluginDocumentView-deallocated-dataSource.html.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41821 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-18  Anders Carlsson  <andersca@apple.com>

        Reviewed by Oliver Hunt.

        WebKit side of <rdar://problem/6688244>.

        Try reinitializing the vendor port if it's invalid.

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

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41820 268f45cc-cd09-0410-ab3c-d52691b4dbfc

<rdar://problem/6693300> Don't rely on printf from TestNetscapePlugIn appearing in test results

Reviewed by Dan Bernstein.

Switch from using printf to using the NPAPI to invoke console.log so that plug-in messages appear
in test results even when the plug-in's stdout differs from DumpRenderTree's stdout.

* DumpRenderTree/TestNetscapePlugIn.subproj/main.cpp:
(log): Invoke console.log via the NPAPI.
(NPP_Destroy): Call log instead of printf.
(NPP_SetWindow): Ditto.
(handleEventCarbon): Ditto.
(handleEventCocoa): Ditto.
(NPP_HandleEvent): Pass the instance in to the event handler.

Update test results for change to address <rdar://problem/6693300>.

* plugins/mouse-events-expected.txt:
* plugins/netscape-destroy-plugin-script-objects-expected.txt:
* plugins/netscape-plugin-setwindow-size-2-expected.txt:
* plugins/netscape-plugin-setwindow-size-expected.txt:
* plugins/open-and-close-window-with-plugin-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41819 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-18  Gustavo Noronha Silva  <gns@gnome.org> and Thadeu Lima de Souza Cascardo <cascardo@holoscopio.com>

        Reviewed by Darin Adler.

        https://bugs.webkit.org/show_bug.cgi?id=24674
        Crashes in !PLATFORM(MAC)'s formatLocaleDate, in very specific situations

        Make sure strftime never returns 2-digits years to avoid ambiguity and
        a crash.

        * runtime/DatePrototype.cpp:
        (JSC::formatLocaleDate):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41818 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Add webViewClosing to private UI delegate.

        Reviewed by Adam Roben.

        * Interfaces/IWebUIDelegatePrivate.idl:
        * Interfaces/WebKit.idl:
        * WebView.cpp:
        (WebView::close):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41817 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-18  Eric Carlson  <eric.carlson@apple.com>

        Reviewed by Simon Fraser.

        Fix for <rdar://problem/6685235>
        <video> element poster cannot be set dynamically if not originally set up in HTML

        Allocate the media engine immediately so the plug-in proxy is available for
        scripting right away.

        * platform/graphics/MediaPlayer.cpp:
        (WebCore::MediaPlayer::MediaPlayer):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41816 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-18  Beth Dakin  <bdakin@apple.com>

        Rubber-stamped by Dan Bernstein.

        Updating test to reflect our joy.

        * fast/dom/offset-parent-positioned-and-inline-expected.txt:
        * fast/dom/offset-parent-positioned-and-inline.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41815 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WebCore:

2009-03-18  Beth Dakin  <bdakin@apple.com>

        Reviewed by Dave Hyatt.

        Fix for <rdar://problem/6636747> REGRESSION (Safari 4 PB-r41326):
        Popup menu appears at the wrong location on page at http://
        www.signonsandiego.com/

        This was a regression from http://trac.webkit.org/changeset/40769,
        which changed the base class of RenderInline to
        RenderBoxModelObject rather than RenderBox.

        * rendering/RenderObject.cpp:
        (WebCore::RenderObject::offsetParent): offsetParent should return a
        RenderBoxModelObject rather than just a RenderBox, which is more
        restrictive.

LayoutTests:

2009-03-18  Beth Dakin  <bdakin@apple.com>

        Reviewed by Dave Hyatt.

        Test for <rdar://problem/6636747> REGRESSION (Safari 4 PB-r41326):
        Popup menu appears at the wrong location on page at http://
        www.signonsandiego.com/

        * fast/dom/offset-parent-positioned-and-inline-expected.txt: Added.
        * fast/dom/offset-parent-positioned-and-inline.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41814 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WebCore:

2009-03-18  Anders Carlsson  <andersca@apple.com>

        Reviewed by Darin Adler.

        <rdar://problem/6504776>
        CrashTracer: [USER] 188 crashes in Safari at com.apple.WebCore â\80¢ WTF::HashTableIterator<WTF::RefPtr<WebCore::ResourceLoader>, ...

        Return early in case calling the client ends up spinning the run loop and completing/cancelling the load.

        * loader/NetscapePlugInStreamLoader.cpp:
        (WebCore::NetscapePlugInStreamLoader::didCancel):

WebKit/mac:

2009-03-18  Anders Carlsson  <andersca@apple.com>

        Reviewed by Darin Adler.

        <rdar://problem/6504776>
        CrashTracer: [USER] 188 crashes in Safari at com.apple.WebCore â\80¢ WTF::HashTableIterator<WTF::RefPtr<WebCore::ResourceLoader>, ...

        If the m_pluginView member was zeroed out as a result of making a call into the plug-in, the pluginFunctionCallDepth would be off
        causing the plug-in never to be stopped. Simplify the code by using a RAII object.

        * Plugins/WebBaseNetscapePluginStream.mm:
        (PluginStopDeferrer::PluginStopDeferrer):
        (PluginStopDeferrer::~PluginStopDeferrer):
        (WebNetscapePluginStream::startStream):
        (WebNetscapePluginStream::wantsAllStreams):
        (WebNetscapePluginStream::destroyStream):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41813 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-18  David Levin  <levin@chromium.org>

        Reviewed by Dimitri Glazkov.

        https://bugs.webkit.org/show_bug.cgi?id=24664
        Upstreaming v8 collection.h

        No change in behavior, so no test.

        * bindings/v8/V8Collection.h: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41812 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WebCore:

2009-03-18  Sam Weinig  <sam@webkit.org>

        Reviewed by Simon Fraser.

        Fix for https://bugs.webkit.org/show_bug.cgi?id=23966
        REGRESSION: Custom arrow navigation functionality doesn't work at dropular.net
        <rdar://problem/6589657>

        Match the CSSOM spec for getClientRects and getBoundingClientRect by
        returning a 0x0 rect at the correct top/left position for empty inline
        boxes.

        * rendering/RenderInline.cpp:
        (WebCore::RenderInline::absoluteRects):
        (WebCore::RenderInline::absoluteQuads):

LayoutTests:

2009-03-18  Sam Weinig  <sam@webkit.org>

        Reviewed by Simon Fraser.

        Update test results for https://bugs.webkit.org/show_bug.cgi?id=23966
        REGRESSION: Custom arrow navigation functionality doesn't work at dropular.net
        <rdar://problem/6589657>

        * fast/dom/getBoundingClientRect-expected.txt:
        * fast/dom/getClientRects-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41811 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by Darin Adler.

        https://bugs.webkit.org/show_bug.cgi?id=24670
        Add tests for XHR progress events during cross-origin requests.

        * http/tests/xmlhttprequest/redirect-cross-origin-2-expected.txt: Added.
        * http/tests/xmlhttprequest/redirect-cross-origin-2.html: Added.
        * http/tests/xmlhttprequest/redirect-cross-origin-expected.txt: Added.
        * http/tests/xmlhttprequest/redirect-cross-origin-post-expected.txt: Added.
        * http/tests/xmlhttprequest/redirect-cross-origin-post-sync-expected.txt: Added.
        * http/tests/xmlhttprequest/redirect-cross-origin-post-sync.html: Added.
        * http/tests/xmlhttprequest/redirect-cross-origin-post.html: Added.
        * http/tests/xmlhttprequest/redirect-cross-origin-sync-double-expected.txt: Added.
        * http/tests/xmlhttprequest/redirect-cross-origin-sync-double.html: Added.
        * http/tests/xmlhttprequest/redirect-cross-origin-sync-expected.txt: Added.
        * http/tests/xmlhttprequest/redirect-cross-origin-sync.html: Added.
        * http/tests/xmlhttprequest/redirect-cross-origin.html: Added.
        * http/tests/xmlhttprequest/simple-cross-origin-denied-events-expected.txt: Added.
        * http/tests/xmlhttprequest/simple-cross-origin-denied-events-post-expected.txt: Added.
        * http/tests/xmlhttprequest/simple-cross-origin-denied-events-post-sync-expected.txt: Added.
        * http/tests/xmlhttprequest/simple-cross-origin-denied-events-post-sync.html: Added.
        * http/tests/xmlhttprequest/simple-cross-origin-denied-events-post.html: Added.
        * http/tests/xmlhttprequest/simple-cross-origin-denied-events-sync-expected.txt: Added.
        * http/tests/xmlhttprequest/simple-cross-origin-denied-events-sync.html: Added.
        * http/tests/xmlhttprequest/simple-cross-origin-denied-events.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41810 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-18  David Levin  <levin@chromium.org>

        Reviewed by Eric Seidel.

        Match the changes done for windows in the bug
        https://bugs.webkit.org/show_bug.cgi?id=24530.

        No change in behavior, so no test.

        * rendering/RenderThemeChromiumLinux.cpp:
        (WebCore::RenderThemeChromiumLinux::extraDefaultStyleSheet):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41809 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-18  Dirk Schulze  <krit@webkit.org>

        Reviewed by Holger Freyther.

        Added support for SVG's stroked texts to Qt.

        [Qt] SVGs stroked text support missing
        https://bugs.webkit.org/show_bug.cgi?id=24262

        * platform/graphics/qt/FontQt.cpp:
        (WebCore::Font::drawComplexText):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41808 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Fix fast/encoding/char-decoding.html with recent ICU.

Reviewed by Alexey Proskuryakov.

* platform/text/TextCodecICU.cpp:
(WebCore::TextCodecICU::decode): Look for gb18030 case-insensitively,
as newer versions of ICU use GB18030 as the canonical name.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41807 268f45cc-cd09-0410-ab3c-d52691b4dbfc

<rdar://problem/6692138> REGRESSION (Safari 4): Incorrect function return value when using IE "try ... finally" memory leak work-around (24654)
<https://bugs.webkit.org/show_bug.cgi?id=24654>

Reviewed by Cameron Zwarich.

If the return value for a function is in a local register we need
to copy it before executing any finalisers, otherwise it is possible
for the finaliser to clobber the result.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41806 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-17  David Levin  <levin@chromium.org>

        Reviewed by Dimitri Glazkov.

        https://bugs.webkit.org/show_bug.cgi?id=24662
        Chromium build fixes.

        Bug https://bugs.webkit.org/show_bug.cgi?id=24663 tracks
        the invalid implementation of ScriptCallStack.

        No change in behavior, so no test.

        * bindings/v8/ScriptCallStack.h:
        (WebCore::ScriptCallStack::state):
        * platform/graphics/skia/PlatformContextSkia.cpp:
        (PlatformContextSkia::drawRect):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41805 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Use the normal search rules for tests inside the platform directory.

Reviewed by Oliver Hunt.

This allows tests inside the platform directory to have different results
on different versions of Mac OS X.

* Scripts/run-webkit-tests:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41804 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Update layout test to avoid depending on formatting of -[NSAttributedString description].

Rubber-stamped by Dan Bernstein.

The attributes aren't important to the test.

* platform/mac/fast/text/attributed-substring-from-range-001-expected.txt:
* platform/mac/fast/text/attributed-substring-from-range-001.html:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41803 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Land updated results for tests involving Geeza Pro.

Rubber-stamped by Dan Bernstein.

* platform/mac-leopard/fast/text/international/Geeza-Pro-vertical-metrics-adjustment-expected.checksum: Renamed from LayoutTests/platform/mac/fast/text/international/Geeza-Pro-vertical-metrics-adjustment-expected.checksum.
* platform/mac-leopard/fast/text/international/Geeza-Pro-vertical-metrics-adjustment-expected.png: Renamed from LayoutTests/platform/mac/fast/text/international/Geeza-Pro-vertical-metrics-adjustment-expected.png.
* platform/mac-leopard/fast/text/international/Geeza-Pro-vertical-metrics-adjustment-expected.txt: Copied from LayoutTests/platform/mac/fast/text/international/Geeza-Pro-vertical-metrics-adjustment-expected.txt.
* platform/mac-leopard/fast/text/international/bidi-AN-after-L-expected.checksum: Renamed from LayoutTests/platform/mac/fast/text/international/bidi-AN-after-L-expected.checksum.
* platform/mac-leopard/fast/text/international/bidi-AN-after-L-expected.png: Renamed from LayoutTests/platform/mac/fast/text/international/bidi-AN-after-L-expected.png.
* platform/mac-leopard/fast/text/international/bidi-AN-after-L-expected.txt: Copied from LayoutTests/platform/mac/fast/text/international/bidi-AN-after-L-expected.txt.
* platform/mac-leopard/fast/text/international/bidi-AN-after-empty-run-expected.checksum: Renamed from LayoutTests/platform/mac/fast/text/international/bidi-AN-after-empty-run-expected.checksum.
* platform/mac-leopard/fast/text/international/bidi-AN-after-empty-run-expected.png: Renamed from LayoutTests/platform/mac/fast/text/international/bidi-AN-after-empty-run-expected.png.
* platform/mac-leopard/fast/text/international/bidi-AN-after-empty-run-expected.txt: Copied from LayoutTests/platform/mac/fast/text/international/bidi-AN-after-empty-run-expected.txt.
* platform/mac-leopard/fast/text/international/bidi-CS-after-AN-expected.checksum: Renamed from LayoutTests/platform/mac/fast/text/international/bidi-CS-after-AN-expected.checksum.
* platform/mac-leopard/fast/text/international/bidi-CS-after-AN-expected.png: Renamed from LayoutTests/platform/mac/fast/text/international/bidi-CS-after-AN-expected.png.
* platform/mac-leopard/fast/text/international/bidi-CS-after-AN-expected.txt: Copied from LayoutTests/platform/mac/fast/text/international/bidi-CS-after-AN-expected.txt.
* platform/mac-leopard/fast/text/international/bidi-neutral-run-expected.checksum: Renamed from LayoutTests/platform/mac/fast/text/international/bidi-neutral-run-expected.checksum.
* platform/mac-leopard/fast/text/international/bidi-neutral-run-expected.png: Renamed from LayoutTests/platform/mac/fast/text/international/bidi-neutral-run-expected.png.
* platform/mac-leopard/fast/text/international/bidi-neutral-run-expected.txt: Copied from LayoutTests/platform/mac/fast/text/international/bidi-neutral-run-expected.txt.
* platform/mac/fast/text/international/Geeza-Pro-vertical-metrics-adjustment-expected.txt:
* platform/mac/fast/text/international/bidi-AN-after-L-expected.txt:
* platform/mac/fast/text/international/bidi-AN-after-empty-run-expected.txt:
* platform/mac/fast/text/international/bidi-CS-after-AN-expected.txt:
* platform/mac/fast/text/international/bidi-neutral-run-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41802 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Move the results out of platform/mac since they're a cross-platform progression,
and non-Mac platforms are already skipping the tests.

Rubber-stamped by Dan Bernstein.

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

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41801 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Update layout test results now that the test passes.

Rubber-stamped by Sam Weinig.

* fast/xsl/mozilla-tests-expected.txt:
* platform/mac-leopard/fast/xsl/mozilla-tests-expected.txt: Copied from LayoutTests/fast/xsl/mozilla-tests-expected.txt.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41800 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Add platform-specific layout test results to accommodate geolocation.

Reviewed by Sam Weinig.

* platform/mac-leopard/fast/dom/Window/window-properties-expected.txt: Copied from LayoutTests/fast/dom/Window/window-properties-expected.txt.
* platform/mac-leopard/fast/dom/navigator-detached-no-crash-expected.txt: Copied from LayoutTests/fast/dom/navigator-detached-no-crash-expected.txt.
* platform/mac/fast/dom/Window/window-properties-expected.txt: Copied from LayoutTests/fast/dom/Window/window-properties-expected.txt.
* platform/mac/fast/dom/navigator-detached-no-crash-expected.txt: Copied from LayoutTests/fast/dom/navigator-detached-no-crash-expected.txt.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41799 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Remove geolocation-not-implemented.html.

Reviewed by Sam Weinig.

* geolocation/geolocation-not-implemented-expected.txt: Removed.
* geolocation/geolocation-not-implemented.html: Removed.
* geolocation/geolocation-test.js: Removed.
* platform/gtk/Skipped:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41798 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Update layout test results to accommodate the updated Unicode text segmentation rules.

Rubber-stamped by Sam Weinig.

* platform/mac-leopard/css1/text_properties/text_transform-expected.checksum: Renamed from LayoutTests/platform/mac/css1/text_properties/text_transform-expected.checksum.
* platform/mac-leopard/css1/text_properties/text_transform-expected.png: Renamed from LayoutTests/platform/mac/css1/text_properties/text_transform-expected.png.
* platform/mac-leopard/css1/text_properties/text_transform-expected.txt: Copied from LayoutTests/platform/mac/css1/text_properties/text_transform-expected.txt.
* platform/mac-leopard/css2.1/t1605-c545-txttrans-00-b-ag-expected.checksum: Renamed from LayoutTests/platform/mac/css2.1/t1605-c545-txttrans-00-b-ag-expected.checksum.
* platform/mac-leopard/css2.1/t1605-c545-txttrans-00-b-ag-expected.png: Renamed from LayoutTests/platform/mac/css2.1/t1605-c545-txttrans-00-b-ag-expected.png.
* platform/mac-leopard/css2.1/t1605-c545-txttrans-00-b-ag-expected.txt: Copied from LayoutTests/platform/mac/css2.1/t1605-c545-txttrans-00-b-ag-expected.txt.
* platform/mac/css1/text_properties/text_transform-expected.txt:
* platform/mac/css2.1/t1605-c545-txttrans-00-b-ag-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41797 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-17  David Carson  <dacarson@apple.com>

        Reviewed by David Hyatt.

        Add anchor elements to the simple style sheet so that applications that
        are using WebKit just for simple text and links will also take advantage
        of the fast path.

        * css/CSSStyleSelector.cpp:
        (WebCore::elementCanUseSimpleDefaultStyle):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41796 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    resolve-ChangeLogs should not die on unmerged non-ChangeLog files

        Reviewed by Adam Roben.

        Fixes the following bug in resolve-ChangeLogs:

            Use of uninitialized value in -e at ./WebKitTools/Scripts/resolve-ChangeLogs line 132.
            Died at ./WebKitTools/Scripts/resolve-ChangeLogs line 164.

        * Scripts/resolve-ChangeLogs:
        (findUnmergedChangeLogs): Check the result of findChangeLog() to
        make sure we don't add undef values to the list of files being
        returned.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41795 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WebCore:

2009-03-17  Darin Adler  <darin@apple.com>

        Reviewed by Adele Peterson.

        <rdar://problem/6687005> Need support for new move-left/right selectors.

        * editing/EditorCommand.cpp:
        (WebCore::executeMoveToLeftEndOfLine): Added.
        (WebCore::executeMoveToLeftEndOfLineAndModifySelection): Added.
        (WebCore::executeMoveToRightEndOfLine): Added.
        (WebCore::executeMoveToRightEndOfLineAndModifySelection): Added.
        Added command entries for the functions above.

WebKit/mac:

2009-03-17  Darin Adler  <darin@apple.com>

        Reviewed by Adele Peterson.

        <rdar://problem/6687005> Need support for new move-left/right selectors.

        * WebView/WebHTMLView.mm: Added the four new selectors to the command-forwarding list.
        * WebView/WebView.mm: Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41794 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-17  Darin Adler  <darin@apple.com>

        Reviewed by Adele Peterson.

        Bug 24477: REGRESSION (r41467): Page Down key scrolls two pages
        https://bugs.webkit.org/show_bug.cgi?id=24477
        rdar://problem/6674184

        * WebView/WebHTMLView.mm:
        (responderChainRespondsToSelector): Added.
        (-[WebHTMLView doCommandBySelector:]): Set eventWasHandled based on whether we
        can find a responder that responds to this selector rather than always assuming
        the selector will not be handled.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41793 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-17  Ojan Vafai  <ojan@chromium.org>

        Reviewed by Adele Peterson.

        Put the padding inside scrollbars on textareas.
        https://bugs.webkit.org/show_bug.cgi?id=24370

        Put the padding inside the scrollbars on textareas. This is done by moving the
        overflow from the shadowNode to the RenderTextControlMultiline. As a result,
        all of the scroll handing methods that RenderTextControl overrides can be moved
        down into RenderTextControlSingleLine since RenderTextControlMultiline can now
        just use RenderBlock's versions. This also allows RenderTextControlMultiLine to
        no longer need a custom layout method since the shadowNode can now just size
        like a regular DIV.

        * css/CSSStyleSelector.cpp:
        (WebCore::CSSStyleSelector::adjustRenderStyle):
        * rendering/RenderTextControl.cpp:
        (WebCore::RenderTextControl::styleDidChange):
        (WebCore::RenderTextControl::selection):
        (WebCore::RenderTextControl::calcHeight):
        (WebCore::RenderTextControl::hitInnerTextElement):
        * rendering/RenderTextControl.h:
        * rendering/RenderTextControlMultiLine.cpp:
        (WebCore::RenderTextControlMultiLine::nodeAtPoint):
        (WebCore::RenderTextControlMultiLine::createInnerTextStyle):
        * rendering/RenderTextControlMultiLine.h:
        * rendering/RenderTextControlSingleLine.cpp:
        (WebCore::RenderTextControlSingleLine::styleDidChange):
        (WebCore::RenderTextControlSingleLine::autoscroll):
        (WebCore::RenderTextControlSingleLine::scrollWidth):
        (WebCore::RenderTextControlSingleLine::scrollHeight):
        (WebCore::RenderTextControlSingleLine::scrollLeft):
        (WebCore::RenderTextControlSingleLine::scrollTop):
        (WebCore::RenderTextControlSingleLine::setScrollLeft):
        (WebCore::RenderTextControlSingleLine::setScrollTop):
        (WebCore::RenderTextControlSingleLine::scroll):
        * rendering/RenderTextControlSingleLine.h:
        * rendering/TextControlInnerElements.cpp:
        (WebCore::RenderTextControlInnerBlock::RenderTextControlInnerBlock):
        (WebCore::RenderTextControlInnerBlock::positionForPoint):
        (WebCore::TextControlInnerTextElement::createRenderer):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41792 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-17  Darin Adler  <darin@apple.com>

        Reviewed by David Hyatt.

        Bug 24517: REGRESSION (r41552): innerHTML does an updateLayout -- unneeded and can be slow
        https://bugs.webkit.org/show_bug.cgi?id=24517

        * editing/DeleteButtonController.cpp:
        (WebCore::DeleteButtonController::enable): Added a call to updateRendering, since
        determining whether to display the delete button involves style and updateRendering
        also updates style (should probably be named updateStyle, in fact). Not needed to fix
        this bug, but would have prevented the crash that led to this bug in the first place.

        * editing/EditCommand.cpp:
        (WebCore::EditCommand::EditCommand): Get rid of unneeded null check. All frames have
        delete button controllers.
        * editing/Editor.cpp:
        (WebCore::Editor::rangeForPoint): Ditto.

        * editing/markup.cpp:
        (WebCore::appendStartMarkup): Changed a "&" to a "&&" so that generating markup
        doesn't depend on renderers at all when the convertBlocksToInlines boolean is false.
        This allows us to omit the call to updateLayoutIgnorePendingStylesheets in the
        createMarkup function that's called by innerHTML.
        (WebCore::MarkupAccumulator::appendMarkup): Turned this into a class with a member
        function. Added a feature where the accumulator will skip a node. Moved arguments
        that don't change during recursion into an object. This function still is a bit
        inefficient, since it creates a new HashMap at every level as it recurses, but for now
        I did not tackle that. Also replaced the onlyIncludeChildren boolean with EChildrenOnly
        for consistency and clarity.
        (WebCore::createMarkup): Removed the call to updateLayoutIgnorePendingStylesheets.
        Instead of calling disable/enable on the delete button controller's container element,
        pass it in to the markup accumulator as a node to skip.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41791 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-17  Gustavo Noronha Silva  <gns@gnome.org>

        Reviewed by Simon Fraser.

        Fix the usage of the $architecture variable for non-Apple-Mac
        ports.

        * Scripts/webkitdirs.pm:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41790 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-17  Scott Violet  <sky@google.com>

        Reviewed by Eric Seidel.

        https://bugs.webkit.org/show_bug.cgi?id=24651
        Skia does not always render text fill/stroke pattern/gradient/color correctly

        Changes Skia's font rendering to only render gradient/pattern if current
        color space indicates the gradient/pattern should be used.
        This is covered by LayoutTests/fast/canvas/canvas-text-alignment.html .

        * platform/graphics/GraphicsContext.cpp:
        (WebCore::GraphicsContext::strokeColorSpace):
        (WebCore::GraphicsContext::fillColorSpace):
        * platform/graphics/GraphicsContext.h:
        (WebCore::):
        * platform/graphics/GraphicsContextPrivate.h:
        * platform/graphics/skia/SkiaFontWin.cpp:
        (WebCore::paintSkiaText):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41789 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WebCore:

2009-03-17  Adele Peterson  <adele@apple.com>

        Reviewed by Darin Adler.

        Fix for https://bugs.webkit.org/show_bug.cgi?id=24655
        <rdar://problem/6633727> Hitting return at the end of a line with an anchor jumps me to the bottom of the message

        Test: editing/inserting/6633727.html

        This changes does a few things:
        1) Renames pos to insertionPosition.
        2) Eliminates "startNode".  It doesn't work well to consider the node separately from the insertionPosition.
           The insertionPosition gets updated at various times, and it seems likely that startNode can get out of sync.
        3) Before building up a list of ancestors to move around when we insert the new block, make sure to use the deepest
           representation of the insertionPosition, so all ancestor nodes are correctly included.

        * editing/InsertParagraphSeparatorCommand.cpp: (WebCore::InsertParagraphSeparatorCommand::doApply):

LayoutTests:

2009-03-17  Adele Peterson  <adele@apple.com>

        Reviewed by Darin Adler.

        Test for https://bugs.webkit.org/show_bug.cgi?id=24655
        <rdar://problem/6633727> Hitting return at the end of a line with an anchor jumps me to the bottom of the message

        * editing/inserting/6633727.html: Added.
        * platform/mac/editing/inserting/6633727-expected.checksum: Added.
        * platform/mac/editing/inserting/6633727-expected.png: Added.
        * platform/mac/editing/inserting/6633727-expected.txt: Added.

        Updated results.  The behavior for these tests doesn't change, but the render tree is slightly different.

        * platform/mac/editing/inserting/insert-div-021-expected.txt:
        * platform/mac/editing/style/block-styles-007-expected.txt:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41788 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Fix the build.

* Plugins/Hosted/HostedNetscapePluginStream.mm:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41787 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Reviewed by Kevin Ollivier.

https://bugs.webkit.org/show_bug.cgi?id=24115
Introduce platform independent stubs for plugins.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41786 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Bug 24645: bisect-builds script doesn't work with Safari 4 Public Beta (version string)

        <https://bugs.webkit.org/show_bug.cgi?id=24645>

        Reviewed by Mark Rowe.

        * Scripts/bisect-builds:
        (makeNightlyList): Added checks for Safari 4 Public Beta on
        Tiger and Leopard.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41785 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Fixed ChangeLog date.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41784 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Use -[NSURLResponse(WebCoreURLResponse) _webcore_MIMEType] consistently

        Reviewed by Darin Adler.

        WebKit r30323 added -_webcore_MIMEType to fix issues with
        incorrect MIME types in NS[HTTP]URLResponse objects.  However,
        uses of -[NSURLResponse MIMEType] still persist in WebKit that
        should be switched to use -_webcore_MIMEType.  Note that
        -[WebDataSource _responseMIMEType] calls back into WebCore to
        get the MIME type from the ResourceResponse object, which has
        already retrieved it via -_webcore_MIMEType.

        * Plugins/Hosted/HostedNetscapePluginStream.mm:
        (WebKit::HostedNetscapePluginStream::didReceiveResponse): Use
        -_webcore_MIMEType.
        * Plugins/WebBaseNetscapePluginStream.mm:
        (WebNetscapePluginStream::didReceiveResponse): Ditto.
        * Plugins/WebNetscapePluginView.mm:
        (-[WebNetscapePluginView pluginView:receivedData:]): Ditto.
        * Plugins/WebPluginController.mm:
        (-[WebPluginController pluginView:receivedResponse:]): Ditto.
        * WebView/WebHTMLRepresentation.mm:
        (-[WebHTMLRepresentation _isDisplayingWebArchive]): Use
        -[WebDataSource _responseMIMEType] instead.
        * WebView/WebPDFRepresentation.m:
        (-[WebPDFRepresentation finishedLoadingWithDataSource:]): Ditto.
        * WebView/WebPDFView.mm:
        (-[WebPDFView menuForEvent:]): Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41783 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-17  Simon Fraser  <simon.fraser@apple.com>

        Reviewed by Dave Hyatt

        https://bugs.webkit.org/show_bug.cgi?id=24632

        Fix repaint issues when composited layers come and go (only applies
        when ACCELERATED_COMPOSITING is turned on).

        * rendering/RenderLayer.cpp:
        (WebCore::RenderLayer::RenderLayer):
        * rendering/RenderLayer.h:
        (WebCore::RenderLayer::mustOverlayCompositedLayers):
        (WebCore::RenderLayer::setMustOverlayCompositedLayers):
        * rendering/RenderLayerBacking.cpp:
        (WebCore::RenderLayerBacking::RenderLayerBacking):
        * rendering/RenderLayerBacking.h:

        Move what used to be the 'forceCompositingLayer' flag from RenderLayerBacking
        to RenderLayer, because we don't want the side-effects of creating RenderLayerBacking
        when setting this flag.

        * rendering/RenderLayerCompositor.cpp:
        (WebCore::RenderLayerCompositor::updateLayerCompositingState):
        When a RenderLayer flips into or out of compositing mode, compute a repaint
        rect relative to the containerForRepaint, and repaint it.

        (WebCore::RenderLayerCompositor::computeCompositingRequirements):
        Call layer->setMustOverlayCompositedLayers() rather than setForcedCompositingLayer().

        (WebCore::RenderLayerCompositor::needsToBeComposited):
        (WebCore::RenderLayerCompositor::requiresCompositingLayer):
        (WebCore::RenderLayerCompositor::requiresCompositingForTransform):
        (WebCore::RenderLayerCompositor::requiresCompositingForAnimation):
        * rendering/RenderLayerCompositor.h:

        Rename requiresCompositingLayerForTransform() to requiresCompositingForTransform()
        and make it a class static method to match requiresCompositingForAnimation(). Both
        now take RenderObjects, rathern than RenderLayers.

        * rendering/style/RenderStyle.h:
        (WebCore::InheritedFlags::hasTransformRelatedProperty):
        Minor tidyup using convenience methods added in an earlier commit.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41782 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-17  Darin Adler  <darin@apple.com>

        Earlier version reviewed by Adele Peterson.

        Bug 24304: REGRESSION (r39864): Hitting the space bar to select an <input type=radio>
        or push an <input type=button> or <button> causes the page to scroll down.

        Would be best to add a regression test for Windows eventually; tested that this has
        no effect on the Mac OS X platform.

        * html/HTMLInputElement.cpp:
        (WebCore::HTMLInputElement::defaultEventHandler): Added FIXMEs and tweaked formatting.
        Use the code that calls the base class's defaultEventHandler early only in the cases
        where it's needed: keydown and keypress events in text fields. In other cases, do the
        more typical thing and call the default handler only at the end of the function.
        This function already had code to make sure the keypress event for space never gets
        through, but it was running too late since the scrolling code was moved into the
        base class default event handler.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41781 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-17  Simon Fraser  <simon.fraser@apple.com>

        Reviewed by Darin Adler

        https://bugs.webkit.org/show_bug.cgi?id=24396

        * config.h:
        Add WTF_USE_ACCELERATED_COMPOSITING, defined to 0 for now, and add some
        comments to make the #ifdefs more readable.

        * css/CSSComputedStyleDeclaration.cpp:
        (WebCore::computedTransform):
        Add a comment to mention that we don't flatten the matrix.

        * css/MediaQueryEvaluator.cpp:
        (WebCore::transform_3dMediaFeatureEval):
        Have the 'transform-3d' media query evaluate to 'true' if 3d-rendering
        is supported.

        * platform/graphics/mac/GraphicsLayerCA.mm:
        (WebCore::GraphicsLayerCA::animateTransform):
        No need for the #ifdef here. If we don't support 3d, we will have already flattened
        the matrix.

        * platform/graphics/transforms/TransformationMatrix.cpp:
        (WebCore::TransformationMatrix::makeAffine):
        * platform/graphics/transforms/TransformationMatrix.h:
        New method to convert the matrix to an affine matrix by throwing a way the non-affine
        parts.

        * rendering/RenderLayer.cpp:
        (WebCore::RenderLayer::updateTransform):
        (WebCore::RenderLayer::currentTransform):
        * rendering/RenderLayerBacking.cpp:
        (WebCore::RenderLayerBacking::updateLayerTransform):
        If 3d rendering is not supported, convert the matrix to an affine matrix
        which can be rendered, and used for hit testing.

        * rendering/RenderLayerCompositor.cpp:
        Change the name of the exported symbol that webkitdirs.pm uses to know if
        3d rendering is supported. There is no other 3d-rendering-specific symbol we can sniff.

        * rendering/RenderObject.cpp:
        (WebCore::RenderObject::transformFromContainer):
        Only take perspective into account if 3d rendering is supported.

        * rendering/RenderObject.h:
        (WebCore::makeMatrixRenderable):
        Utility method that flattens a matrix if 3d rendering is not supported.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41780 268f45cc-cd09-0410-ab3c-d52691b4dbfc

wx build fix. Fix typo after mouse wheel changes.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41779 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Update failing layout test after float painting fix.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41778 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-17  Darin Adler  <darin@apple.com>

        * inspector/ConsoleMessage.cpp:
        (WebCore::ConsoleMessage::isEqual): Fix build, remove stray parenthesis.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41777 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-17  Gustavo Noronha Silva  <gns@gnome.org>

        Reviewed by Mark Rowe.

        Enable HTML5 media elements support by default in the GTK+ port.

        * configure.ac:

WebKitTools:

2009-03-17  Gustavo Noronha Silva  <gns@gnome.org>

        Reviewed by Mark Rowe.

        Enable HTML5 media elements support by default also for the GTK+
        port.

        * Scripts/build-webkit:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41776 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-17  Darin Adler  <darin@apple.com>

        Fix crash seen right away when running run-webkit-tests.

        * inspector/ConsoleMessage.cpp:
        (WebCore::ConsoleMessage::isEqual): Restore assertion to its behavior pre-refactoring.
        Also tweaked formatting a bit.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41775 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by Darin Adler.

        HTMLSelectElement::add() doesn't look at exception code returned from insertBefore(), so
        it doesn't need to zero it out before calling.

        * html/HTMLSelectElement.cpp: (WebCore::HTMLSelectElement::add): Removed "ec = 0" line.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41774 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WebCore:

        Reviewed by Adam Roben.

        - WebCore part of adding a mechanism for controlling the caching of
          responses through WebFrameLoaderClient

        Mac already has such a mechanism, and this adds one for CFNetwork ports.

        * WebCore.vcproj/WebCore.vcproj: Added EmptyClients.h
        and ResourceLoaderCFNet.cpp.

        * loader/EmptyClients.h:
        (WebCore::EmptyFrameLoaderClient::shouldCacheResponse): Added an
        implementation that always returns true.

        * loader/FrameLoaderClient.h: Declared shouldCacheResponse().

        * loader/ResourceLoader.h: Ditto.

        * loader/cf/ResourceLoaderCFNet.cpp: Added.
        (WebCore::ResourceLoader::shouldCacheResponse): Added. Calls through to
        FrameLoaderClient::shouldCacheResponse().

        * platform/network/ResourceHandleClient.h:
        (WebCore::ResourceHandleClient::shouldCacheResponse): Added an
        implementation that always returns true.

        * platform/network/cf/ResourceHandleCFNet.cpp:
        (WebCore::willCacheResponse): Added a call to
        ResourceHandleClient::shouldCacheResponse(). If the client returns
        false, return 0, which will prevent CFNetwork from caching the response.

WebKit/win:

        Reviewed by Adam Roben.

        - WebKit part of adding a mechanism for controlling the caching of
          responses through WebFrameLoaderClient

        * Interfaces/IWebResourceLoadDelegatePrivate.idl: Added the
        IWebResourceLoadDelegatePrivate3 interface, containing the new
        delegate method shouldCacheResponse().

        * Interfaces/IWebURLRequest.idl: Added the isEqual() method.

        * WebCoreSupport/WebFrameLoaderClient.cpp:
        (WebFrameLoaderClient::shouldCacheResponse): Implemented this new
        FrameLoaderDelegate method by calling through to the resource load
        delegate, if it implements shouldCacheResponse(), and returning true
        otherwise.

        * WebCoreSupport/WebFrameLoaderClient.h:

        * WebMutableURLRequest.cpp:
        (WebMutableURLRequest::isEqual): Added.
        (WebMutableURLRequest::setValue): Implemented.

        * WebMutableURLRequest.h: Declared isEqual().

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41773 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    Reviewed by Darin Adler.

        https://bugs.webkit.org/show_bug.cgi?id=13287
        Cannot change SELECT to a dynamically created option

        Tests: fast/forms/add-and-remove-option.html
               fast/forms/add-remove-option-modification-event.html
               fast/forms/add-selected-option.html
               fast/forms/select-cache-desynchronization.html

        * dom/ContainerNode.cpp:
        (WebCore::dispatchChildInsertionEvents): Increment DOM tree version. This will happen when
        dispatching DOMSubtreeModified again, but the version should be incremented for event
        listeners to have an up to date view of the DOM.
        (WebCore::dispatchChildRemovalEvents): Ditto.

        * html/HTMLOptionElement.cpp: (WebCore::HTMLOptionElement::insertedIntoTree):
        Make sure that the select element knows about its new selected option.

        * html/HTMLOptionElement.h: Use insertedIntoTree() instead of insertedIntoDocument(),
        because DOM also needs to be updated for forms that are not in document yet. Similar
        problems exist for node removing, but removedFromTree() is called at a wrong time, so
        those problems cannot be fixed without deeper refactoring.

        * html/HTMLSelectElement.cpp:
        (WebCore::HTMLSelectElement::setRecalcListItems): Reset m_activeSelectionAnchorIndex - it
        doesn't make sense to keep the anchor after programmatically changing the selection, and
        keeping it was causing a failure in fast/forms/listbox-selection.html.

        * html/HTMLSelectElement.h: Removed overrides for ContainerNode methods that only called
        base class versions.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41772 268f45cc-cd09-0410-ab3c-d52691b4dbfc

    <rdar://problem/6690324> Accessing FTP sites reads unallocated memory, can result in garbled entries or crashes

        Reviewed by Darin Adler.

        * loader/FTPDirectoryDocument.cpp:
        (WebCore::FTPDirectoryTokenizer::parseAndAppendOneLine): Assign CString to a local while we hold pointers into it.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41771 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Fix ChangeLog comment.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41770 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WebCore:

2009-03-16  David Hyatt  <hyatt@apple.com>

        <rdar://problem/6648411> REGRESSION: Layout of page is wrong at http://www.popcap.com/

        Make sure that the initial shouldPaint check that looks at enclosingLayers properly skips over
        layers that don't paint themselves.  This is done by adding a new boolean parameter to enclosingLayer
        so that RenderObjects can walk up the enclosing layer chain and skip any layers that don't paint
        themselves.

        Reviewed by Darin Adler.

        Added fast/block/float/overlapping-floats-with-overflow-hidden.html

        * WebCore.base.exp:
        * rendering/RenderBlock.cpp:
        (WebCore::RenderBlock::addOverhangingFloats):
        * rendering/RenderObject.cpp:
        (WebCore::RenderObject::enclosingSelfPaintingLayer):
        * rendering/RenderObject.h:

LayoutTests:

2009-03-16  David Hyatt  <hyatt@apple.com>

        <rdar://problem/6648411> REGRESSION: Layout of page is wrong at http://www.popcap.com/

        Make sure that the initial shouldPaint check that looks at enclosingLayers properly skips over
        layers that don't paint themselves.  This is done by adding a new enclosingSelfPaintingLayer method
        so that RenderObjects can walk up the enclosing layer chain and skip any layers that don't paint
        themselves.

        Reviewed by Darin Adler.

        * fast/block/float/overlapping-floats-with-overflow-hidden.html: Added.
        * platform/mac/fast/block/float/overlapping-floats-with-overflow-hidden-expected.checksum: Added.
        * platform/mac/fast/block/float/overlapping-floats-with-overflow-hidden-expected.png: Added.
        * platform/mac/fast/block/float/overlapping-floats-with-overflow-hidden-expected.txt: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41769 268f45cc-cd09-0410-ab3c-d52691b4dbfc

2009-03-17  Xan Lopez  <xlopez@igalia.com>

        Reviewed by Holger Freyther.

        https://bugs.webkit.org/show_bug.cgi?id=24592
        [GTK] Crash in FcPatternHash

        Style fixes.

        * platform/graphics/gtk/FontPlatformDataGtk.cpp:
        (WebCore::FontPlatformData::operator=):
        * platform/graphics/gtk/FontPlatformDataPango.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@41768 268f45cc-cd09-0410-ab3c-d52691b4dbfc