GitLab

        Reviewed by Sam Weinig.

        Next step in refactoring JSGlobalObject: Added JSVariableObject class,
        and factored symbol-table-related code into it. (JSGlobalObject doesn't
        use the symbol table code yet, though.)
        
        Layout and JS tests, and testapi, pass. SunSpider reports no regression.

WebCore:

        Reviewed by Sam Weinig.

        Added some namespace qualifications and a forwarding header, now that
        KJS::Node is sometimes #included in WebCore by JavaScriptCore headers.

        * ForwardingHeaders/wtf/ListRefPtr.h: Added.
        * bindings/js/JSXSLTProcessor.cpp:
        (KJS::JSXSLTProcessorPrototypeFunctionTransformToFragment::callAsFunction):
        * bindings/js/kjs_binding.cpp:
        (KJS::ScriptInterpreter::getDOMNodeForDocument):
        (KJS::ScriptInterpreter::forgetDOMNodeForDocument):
        (KJS::ScriptInterpreter::putDOMNodeForDocument):
        (KJS::ScriptInterpreter::markDOMNodesForDocument):
        (KJS::ScriptInterpreter::updateDOMNodeDocument):

WebKit/mac:

        Reviewed by Sam Weinig.
        
        Added a forwarding header, since we now #include nodes.h through some
        JavaScriptCore headers.

        * ForwardingHeaders/wtf/ListRefPtr.h: Added.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28527 268f45cc-cd09-0410-ab3c-d52691b4dbfc

 WebCore:

         Add SoftLinking.h for Windows and use it in a few places

         Reviewed by Oliver.

         * WebCore.vcproj/WebCore.vcproj: Added new file to project.
         * platform/win/PlatformScrollBarSafari.cpp: Use SoftLinking.h.
         (WebCore::PlatformScrollbar::PlatformScrollbar): Removed manual
         soft-linking calls.
         (WebCore::PlatformScrollbar::paintButton): Check for the presence of
         the SafariTheme library now that we can't check for the presence of
         paintThemePart directly.
         (WebCore::PlatformScrollbar::paintTrack): Ditto.
         (WebCore::PlatformScrollbar::paintThumb): Ditto.
         * platform/win/SoftLinking.h: Copied from WebCore/platform/mac/SoftLinking.h.
         * rendering/RenderThemeSafari.cpp: Same basic changes as to
         PlatformScrollBarSafari.cpp.
         (WebCore::RenderThemeSafari::RenderThemeSafari):
         (WebCore::RenderThemeSafari::isControlStyled):
         (WebCore::RenderThemeSafari::paintCapsLockIndicator):
         * rendering/RenderThemeSafari.h: Removed m_themeDLL member.

 WebKit/win:

         Use SoftLinking.h in WebKitClassFactory.

         Reviewed by Oliver.

         * WebKitClassFactory.cpp:
         (WebKitClassFactory::WebKitClassFactory): We don't have to worry about
         not finding STInitialize anymore now that a version of SafariTheme
         containing this function has shipped.


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28526 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        - fix http://bugs.webkit.org/show_bug.cgi?id=16185
          jsRegExpCompile should not add implicit non-capturing bracket

        While this does not make SunSpider faster, it will make many regular
        expressions a bit faster.

        * pcre/pcre_compile.cpp: Moved CompileData struct in here from the
        header since it's private to this file.
        (compile_branch): Updated for function name change.
        (compile_bracket): Renamed from compile_regex, since, for one thing,
        this does not compile an entire regular expression.
        (calculateCompiledPatternLengthAndFlags): Removed unused item_count
        local variable. Renamed CompileData to cd instead of compile_block
        to be consistent with other functions. Added code to set the
        needOuterBracket flag if there's at least one "|" at the outer level.
        (jsRegExpCompile): Renamed CompileData to cd instead of compile_block
        to be consistent with other functions. Removed unneeded "size" field
        from the compiled regular expression. If no outer bracket is needed,
        then use compile_branch to compile the regular expression.

        * pcre/pcre_internal.h: Removed the CompileData struct, which is now
        private to pcre_compile.cpp. Removed the size member from JSRegExp.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28525 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Adele.

        - fix <rdar://problem/5608795> CrashTracer: 481 crashes in Safari
          at WebCore::HTMLSelectElement::saveState const + 152

        Test: fast/forms/select-set-inner.html

        * dom/ContainerNode.cpp:
        (WebCore::ContainerNode::removeChildren): Added a return value, as with other
        calls that change children, so we can optimize for the case where it does nothing.
        (WebCore::ContainerNode::cloneChildNodes): Changed parameter type to ContainerNode.
        * dom/ContainerNode.h: See above.

        * html/HTMLOptGroupElement.cpp:
        (WebCore::HTMLOptGroupElement::removeChildren): Override removeChildren and call
        recalcSelectOptions in that case.
        (WebCore::HTMLOptGroupElement::childrenChanged): Override childrenChanged instead of
        addChild, for consistency with HTMLSelectElement; no need to override both.
        (WebCore::HTMLOptGroupElement::groupLabelText): Made const.
        * html/HTMLOptGroupElement.h: See above.

        * html/HTMLSelectElement.cpp: Don't override addChild any more, because we already
        override childrenChanged, and addChild calls that.
        (WebCore::HTMLSelectElement::removeChildren): Override removeChildren and call
        recalcSelectOptions in that case.
        (WebCore::HTMLSelectElement::recalcListItems): Tightened up the code a little bit
        by using a for loop and traverseNextSibling. Also added some new comments and
        removed some obsolete ones.
        (WebCore::HTMLSelectElement::checkListItems): Added. Debug-only check to make
        sure we don't have a stale list items vector.
        * html/HTMLSelectElement.h: Changed listItems() to invoke checkListItems().
        This will help us catch cases where we have too few calls to setRecalcListItems.

LayoutTests:

        Reviewed by Adele.

        - test for <rdar://problem/5608795> CrashTracer: 481 crashes in Safari
          at WebCore::HTMLSelectElement::saveState const + 152

        * fast/forms/select-set-inner-expected.txt: Added.
        * fast/forms/select-set-inner.html: Added.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28524 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Darin Adler.

        - WebCore part of fixing <rdar://problem/5499918> REGRESSION: insertion point in input field with custom border cuts holes in focus ring interior edges

        Test: fast/repaint/focus-ring.html

        * editing/SelectionController.cpp:
        (WebCore::SelectionController::caretRepaintRect): Changed to return just
        the caret rect without any padding.
        (WebCore::SelectionController::recomputeCaretRect): Changed to repaint
        just the caret rect without any padding.
        * platform/graphics/GraphicsContext.h: Removed setFocusRingClip() and
        clearFocusRingClip().
        * platform/graphics/cairo/GraphicsContextCairo.cpp: Ditto.
        * platform/graphics/cg/GraphicsContextCG.cpp: Ditto.
        * platform/graphics/cg/GraphicsContextPlatformPrivate.h: Removed
        m_focusRingClip member.
        * platform/graphics/mac/GraphicsContextMac.mm:
        (WebCore::GraphicsContext::drawFocusRing): Changed to call
        wkDrawFocusRing() once without setting up additional clip. On Leopard,
        wkDrawFocusRing() respects the context clip now. On Tiger, a
        transparency layer is used to apply clipping to the focus ring.
        * platform/graphics/qt/GraphicsContextQt.cpp: Removed focus ring clip
        methods and member.
        * platform/graphics/wx/GraphicsContextWx.cpp: Ditto.
        * platform/mac/WebCoreSystemInterface.h: Removed the clipRect argument
        to wkDrawFocusRing().
        * platform/mac/WebCoreSystemInterface.mm: Ditto.
        * rendering/RenderLayer.cpp:
        (WebCore::setClip): Removed call to set the focus ring clip.
        (WebCore::restoreClip): Removed call to reset the focus ring clip.

WebKitLibraries:

        Reviewed by Darin Adler.

        - updated system interface for fixing <rdar://problem/5499918> REGRESSION: insertion point in input field with custom border cuts holes in focus ring interior edges

        * WebKitSystemInterface.h:
        * libWebKitSystemInterfaceLeopard.a:
        * libWebKitSystemInterfaceTiger.a:

LayoutTests:

        Reviewed by Darin Adler.

        - repaint test for <rdar://problem/5499918> REGRESSION: insertion point in input field with custom border cuts holes in focus ring interior edges

        * fast/repaint/focus-ring.html: Added.
        * platform/mac/fast/repaint/focus-ring-expected.checksum: Added.
        * platform/mac/fast/repaint/focus-ring-expected.png: Added.
        * platform/mac/fast/repaint/focus-ring-expected.txt: Added.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28523 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        * misc/DatabaseExample.html: Changed to differentiate failure between running with an
          unsupported browser or some other error that prevented the database from opening.
          In the future we can use ExceptionCodes returned from openDatabase(), but a good list
          of those don't exist yet



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28522 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        - fix <rdar://problem/5601586> QtKit should be dynamically loaded upon need, not linked at startup

        Also did a lot of small tweaks to MediaPlayerPrivateQTKit.

        * WebCore.xcodeproj/project.pbxproj: Don't link to QTKit.

        * platform/graphics/mac/MediaPlayerPrivateQTKit.h: Omit unneeded includes and declarations.
        Made a lot more functions const. Made a few more members private and a couple inline.
        * platform/graphics/mac/MediaPlayerPrivateQTKit.mm: Added soft linking machinery for all the
        things we currently use in QTKit. It's a little more awkward for classes and other data objects
        than it is for functions, but still relatively straightforward, with no changes needed to the
        client code. Added using namespace directives. Made a cuePointTimerInterval constant and put
        it at the top of the file. Use 0 consistently instead of sometimes 0 and sometimes 0.0f.
        (WebCore::MediaPlayerPrivate::MediaPlayerPrivate): Removed unneeded initialization of RetainPtr
        members to nil.
        (WebCore::MediaPlayerPrivate::createQTMovie): Use adoptNS instead of autorelease.
        (WebCore::MediaPlayerPrivate::createQTMovieView): Ditto. Also use -[NSColor clearColor].
        (WebCore::MediaPlayerPrivate::createQTTime): Remove an unneeded type cast that had no effect.
        Changed to use long instead of int because that's the type for a QTTime time scale anyway.
        (WebCore::MediaPlayerPrivate::duration): Use a static_cast instead of a C-style cast.
        (WebCore::MediaPlayerPrivate::currentTime): Ditto. Also merged into a single expression.
        (WebCore::MediaPlayerPrivate::cuePointTimerFired): Added code to make a copy of the cue
        points set to avoid a potential problem with a set being modified as we iterate it.
        (WebCore::MediaPlayerPrivate::bytesLoaded): Removed unneeded null check of m_qtMovie.
        (WebCore::MediaPlayerPrivate::updateStates): Instead of comments explaining the numeric
        values, used the constants from the headers directly.
        (WebCore::MediaPlayerPrivate::getSupportedTypes): Instead of (QTMovieFileTypeOptions)0,
        pass the named constant with value 0, QTIncludeCommonTypes. Skipped the intermediate type
        of NSString to remove one cast. Replaced C-style cast with reinterpret_cast (arguably
        no better). Used RetainPtr instead of explicit CFRelease calls.

        * platform/mac/SoftLinking.h: Added macros to do soft linking for classes and for pointers.
        It's not quite as automatic as the soft linking we can do for functions, since these define
        functions to get the values, so you need to define macros to make what look like variable
        accesses turn into function calls. See MediaPlayerPrivateQTKit for the details.

        * html/HTMLMediaElement.h:
        * html/TimeRanges.h:
        * html/VoidCallback.h:
        * platform/graphics/MediaPlayer.h:
        Use angle brackets for wtf includes. Omit unneeded includes.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28521 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        <rdar://problem/5556378> Implemented database related UI delegate methods (prompts for new/enlarged databases)

        Reviewed by Darin.

        * Interfaces/IWebUIDelegatePrivate.idl:
        * WebChromeClient.cpp:
        (WebChromeClient::requestQuotaIncreaseForNewDatabase):
        (WebChromeClient::requestQuotaIncreaseForDatabaseOperation):



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28520 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Darin Adler.

        - fix http://bugs.webkit.org/show_bug.cgi?id=16334
          <rdar://problem/5634923> REGRESSION (r28299): Homepage of any DotMac Web Gallery won't load completely

        Test: fast/dynamic/subtree-no-common-root-static-y.html

        * rendering/RenderObject.cpp:
        (WebCore::RenderObject::markContainingBlocksForLayout): Changed the call
        to setChildNeedsLayout() to not mark containing blocks and added a
        separate call to markContainingBlocksForLayout() that will not schedule
        a layout if we are already in the middle of scheduleRelayoutOfSubtree().

LayoutTests:

        Reviewed by Darin Adler.

        - test for http://bugs.webkit.org/show_bug.cgi?id=16334
          <rdar://problem/5634923> REGRESSION (r28299): Homepage of any DotMac Web Gallery won't load completely

        * fast/dynamic/subtree-no-common-root-static-y.html: Added.
        * platform/mac/fast/dynamic/subtree-no-common-root-static-y-expected.checksum: Added.
        * platform/mac/fast/dynamic/subtree-no-common-root-static-y-expected.png: Added.
        * platform/mac/fast/dynamic/subtree-no-common-root-static-y-expected.txt: Added.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28519 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Reviewed by Darin Adler.


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28518 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Reviewed by Darin Adler.


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28517 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Alp Toker.

        http://bugs.webkit.org/show_bug.cgi?id=16333
        [GTK] Key bindings must work with focused frames.

        There are still more cases where getFrameFromView() is mis-used that
        need to be fixed, but this is a good start.

        * WebView/webkitwebview.cpp:
        (webkit_web_view_real_select_all):
        (webkit_web_view_real_cut_clipboard):
        (webkit_web_view_real_copy_clipboard):
        (webkit_web_view_real_paste_clipboard):


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28516 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        http://bugs.webkit.org/show_bug.cgi?id=16325
        <rdar://problem/5632997> REGRESSION: www.xerox.ru doesn't work
        
        Test: http/tests/misc/empty-cookie.html

        * platform/mac/CookieJar.mm:
        (WebCore::setCookies): Don't store empty cookies.
        (WebCore::cookies): Filter out empty cookies if we have them, as they could have been set
        with an earlier version of Leopard!



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28515 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        returning true when there's an error).

        Reviewed by Steve.

        * platform/win/FileSystemWin.cpp:
        (WebCore::makeAllDirectories):



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28514 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        http://bugs.webkit.org/show_bug.cgi?id=16091
        JSCRE needs to import the PCRE test suite

        We can use two files from PCRE test suite, as others either test for features not supported
        in JS, or use internal debugging facilities of PCRE.

        The files are checked in almost unmodified, I have only converted them to UTF-8 and removed two
        huge test cases that were completely incompatible with JSC, and were polluting the output.

        Many of the failures in results are just legitimate differences in behavior or possibly
        problems with the test runner, but some may be genuine bugs and even regressions in JSC.

        * fast/regex: Added.
        * fast/regex/pcre-test-runner.js: Added.
        * fast/regex/test1-expected.txt: Added.
        * fast/regex/test1.html: Added.
        * fast/regex/test4-expected.txt: Added.
        * fast/regex/test4.html: Added.
        * fast/regex/testinput1: Added.
        * fast/regex/testinput4: Added.
        * fast/regex/testoutput1: Added.
        * fast/regex/testoutput4: Added.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28513 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Let's go ahead and call the correct UI Delegate method, shall we?

        * WebCoreSupport/WebChromeClient.mm:
        (WebChromeClient::requestQuotaIncreaseForDatabaseOperation): Call the correct UI delegate



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28512 268f45cc-cd09-0410-ab3c-d52691b4dbfc

Linux build fix - ensure that webcore is linked before jscore so that the linker will know which symbols it needs to link in. Also fix MSVC project file name typo for sample app.


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28511 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        * bindings/js/kjs_binding.cpp:
        (KJS::setDOMException): Oops, this was just supposed to be PERMISSION_DENIED.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28510 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        - fix http://bugs.webkit.org/show_bug.cgi?id=16332
          ObjC DOM exception object descriptions should include the exception name

        * WebCore.pro: Added ExceptionCode.cpp.
        * WebCore.vcproj/WebCore.vcproj: Added ExceptionCode.cpp.
        * WebCore.xcodeproj/project.pbxproj: Added ExceptionCode.cpp.
        * WebCoreSources.bkl: Added ExceptionCode.cpp.

        * bindings/js/kjs_binding.cpp: (KJS::setDOMException): Moved the code to decompose an
        ExceptionCode into ExceptionCode.h/cpp -- getExceptionCodeDescription. Also removed
        the many unneeded includes that were here. Had to keep one special case here, for
        SECURITY_ERR.

        * bindings/objc/ExceptionHandlers.mm: (WebCore::raiseDOMException): Changed to use the
        new getExceptionCodeDescription function so that this shares the exception name
        information that was previously only available to JavaScript.

        * dom/ExceptionCode.cpp: Copied from bindings/js/kjs_binding.cpp.
        (WebCore::getExceptionCodeDescription): Added some assertions, and made the function
        handle exception names in a slightly more robust way that is not subject to integer
        overflow. (Not a real world issue since we should never receive a bad exception code.)

        * dom/ExceptionCode.h: Added the ExceptionCodeDescription struct and the
        getExceptionCodeDescription function.

        * svg/SVGException.h: Added a missing #include and got rid of some comments. Some of
        the comments were mildly helpful, but others were incorrect. This now matches the other
        exception-related headers such as RangeException.h.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28509 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Fixed a glaring bug that would prevent a statement from getting run a second time

        * storage/SQLStatement.cpp:
        (WebCore::SQLStatement::execute): Clear failure due to quota *before* we check the error
          and return early
        (WebCore::SQLStatement::clearFailureDueToQuota): Only clear the error if it was a quota error



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28508 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Use keydown instead of keypress so keyIdentifier can be used.

        * page/inspector/ConsolePanel.js: Use keydown instead of keypress.
        * page/inspector/DatabasePanel.js: Ditto.
        * page/inspector/inspector.js: Ditto. Plus call removeEventListener
        before deleting windowLoaded.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28507 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Rubberstamped by Hyatt.

        * platform/graphics/win/FontDatabase.cpp:


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28506 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Darin.

        * Interfaces/IWebViewPrivate.idl:
        * WebView.cpp:
        (WebView::shouldClose):
        * WebView.h:



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28505 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Tweaked a comment and a few assertions from my last checkin

        * storage/SQLStatement.cpp:
        (WebCore::SQLStatement::execute):
        (WebCore::SQLStatement::clearFailureDueToQuota):
        (WebCore::SQLStatement::lastExecutionFailedDueToQuota):



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28504 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WebKit/mac:

        Remove some assertions we know can fire and replace them with a FIXME

        Reviewed by Anders.

        * WebCoreSupport/WebFrameLoaderClient.mm:

WebKit/win:

        Fix <rdar://5615283> Crash in WebFrame::receivedPolicyDecision

        It's possible for a second policy listener to be created while we're
        waiting for a decision on the first policy listener. In this case we
        would crash when the first policy listener had a decision made. Mac
        WebKit already handles this by invalidating any existing policy
        listener when creating a new one.

        A test case is coming in a future commit.

        Reviewed by Anders.

        * WebFrame.cpp:
        (WebFrame::setUpPolicyListener): Copy the Mac behavior of invalidating
        any existing policy listener, and remove some assertions that we know
        can fire.


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28503 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Finished hooking up UI Delegate for databases - Database operations will now enforce a size quota and
        will ask the UI Delegate for more space when that quota is met

        * platform/sql/SQLiteDatabase.cpp: Add the new SQLResultFull constant
        * platform/sql/SQLiteDatabase.h: Ditto

        * storage/Database.cpp:
        (WebCore::Database::securityOriginData): Added this accessor, copying for thread safety
        (WebCore::Database::stringIdentifier): Ditto
        * storage/Database.h:

        * storage/SQLStatement.cpp:
        (WebCore::SQLStatement::SQLStatement):
        (WebCore::SQLStatement::execute): Change to return an enum that represents 3 states - success, error, and quota.
          If the result is quota, this statement expects that it might be run again, presumably after the user increases 
          the quota
        (WebCore::SQLStatement::setFailureDueToQuota): Setup a quota failure, including a flag and the error
        (WebCore::SQLStatement::clearFailureDueToQuota): Clear a quota failure, for when the statement is rerun
        (WebCore::SQLStatement::lastExecutionFailedDueToQuota):
        * storage/SQLStatement.h:

        * storage/SQLTransaction.cpp:
        (WebCore::SQLTransaction::SQLTransaction):
        (WebCore::SQLTransaction::performPendingCallback): Added an acceptable callback pointer
        (WebCore::SQLTransaction::openTransactionAndPreflight): Setup the quota in the database that will remain for this 
          transaction.  Note that in this patch, the quota being set is wrong - it makes sense to fix that in a follow up patch
        (WebCore::SQLTransaction::runStatements): Modified to add the ability to re-run a statement based on the UI delegate
          decision and whether the current statement was already run
        (WebCore::SQLTransaction::runCurrentStatement): Added another result condition - the Quota result - and handle it
        (WebCore::SQLTransaction::handleCurrentStatementError): Statements can now error-out from two places, so the code
          that handles a statement error was moved here
        (WebCore::SQLTransaction::deliverQuotaIncreaseCallback): Added - Consult the UI delegate for more quota, then 
          reschedule the current statement on the database thread
        * storage/SQLTransaction.h:



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28502 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Only assert if the major version is zero, since zero is handled in the other cases.

        * WebView/WebView.mm:
        (callGestalt): Remove the ASSERT.
        (createMacOSXVersionString): ASSERT that major is not zero.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28501 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        
        Don't use callback-based timers, since these cause Windows to eat Windows crashes
        in code the timers call.
        
        Windows appears to be defending against "shatter" attacks partially by setting
        up a structured exception block while dispatching callback-based WM_TIMERs.
        
        I verified this by adding a divide by zero into some timer callback code.
        In the case where the timer was dispatched via a callback, the divide by zero
        exception was silently handled and ignored, with execution continuing after
        our call to DispatchMessage.  When processed via the WNDPROC, no SEH
        block was established by Windows, and our divide by zero generated a real
        crash (which is what we wanted).
        
        Windows handling our crashes for us led us to leave the timer data structures
        in an invalid state so the next time a timer was set, we'd crash accessing an
        invalid HashMap of timer data.
        
        Reviewed by Hyatt.

        * platform/win/SharedTimerWin.cpp:
        (WebCore::TimerWindowWndProc):
        (WebCore::setSharedTimerFireTime):



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28500 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        - fix <rdar://problem/5513394> No way to detect Tiger vs Leopard from Safari's user agent string

        * WebView/WebView.mm:
        (callGestalt): Added.
        (createMacOSXVersionString): Added.
        (-[WebView _userAgentWithApplicationName:andWebKitVersion:]): Added Mac OS X version string,
        right after the string "Mac OS X", but with underscores instead of dots to avoid the dreaded
        "4." problem (old libraries that think a "4." anywhere in the user agent means Netscape 4).
        (-[WebView _userAgentForURL:]): Fixed incorrect bug numbers.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28499 268f45cc-cd09-0410-ab3c-d52691b4dbfc

WebCore:

        Fix <rdar://5108390> Feed title is too low in blue banner

        Way back in r23069 we started applying the same font ascent hack that
        Mac WebKit applies to Helvetica, Times, and Courier. We did this so
        that those fonts would match the Mac metrics when we run the
        regression tests. However, this hack was applying to Arial on Windows
        when a site would specify the Helvetica font face because Windows will
        alias the font names. Instead of removing the hack entirely, we
        turn it off by default but provide some SPI so that DumpRenderTree can
        turn it on.

        Reviewed by Hyatt.

        All tests pass.

        * platform/graphics/FontData.h: Add a new static method to turn on the
        hack on Windows only.
        * platform/graphics/win/FontDataWin.cpp:
        (WebCore::FontData::setShouldApplyMacAscentHack): Added.
        (WebCore::FontData::platformInit): Only perform the hack if
        shouldApplyMacAscentHack is true.

WebKit/win:

        Add SPI so that DumpRenderTree can turn on the Mac font ascent hack

        Reviewed by Hyatt.

        * Interfaces/IWebViewPrivate.idl: Added new method declaration.
        * WebView.cpp:
        (WebView::setShouldApplyMacFontAscentHack): Added. Just calls down to
        FontData.
        * WebView.h:

WebKitTools:

        Explicitly turn on the Mac font ascent hack on Windows

        This keeps our font metrics matching those from Mac.

        Reviewed by Hyatt.

        * DumpRenderTree/win/DumpRenderTree.cpp:
        (main):


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28498 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        * svg/batik/text/textEffect2-expected.txt:
        * svg/batik/text/textProperties-expected.txt:



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28497 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Sam Weinig.

        Fixed http://bugs.webkit.org/show_bug.cgi?id=16328
        REGRESSION (r28470): Crash expanding a GMail conversation
        
        * page/Frame.cpp:
        (WebCore::Frame::scriptProxy): Only return 0 if JS seems disabled *and*
        we haven't created the proxy yet. If we've created the proxy already, a
        script may be in the midst of execution, even though we've lost our
        settings object. During execution, scripts assume they have free access
        to the proxy.

LayoutTests:

        Reviewed by Sam Weinig.

        Test for http://bugs.webkit.org/show_bug.cgi?id=16328
        REGRESSION (r28470): Crash expanding a GMail conversation

        * fast/frames/location-put-after-removal.html: Added.
        * fast/frames/resources/location-put-after-removal-frame.html: Added.
        * fast/frames/resources/location-put-after-removal-frame-expected.txt: Updated.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28496 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        
        * DumpRenderTree/win/TestNetscapePlugin/main.cpp: Copied from DumpRenderTree/win/TestNetscapePlugin/main.c.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28495 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        * Scripts/do-webcore-rename: Don't rename kjs_css twice.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28494 268f45cc-cd09-0410-ab3c-d52691b4dbfc

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28493 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        removed somehow (possibly when I made PluginObject be a cpp file).
        
        * DumpRenderTree/TestNetscapePlugIn.subproj/PluginObject.cpp:
        (pluginInvoke):



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28492 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Eric Seidel.

        - fix http://bugs.webkit.org/show_bug.cgi?id=16321
          new RegExp("[\u0097]{4,6}", "gmy") crashes in DEBUG builds
          <rdar://problem/5632992>

        Test: fast/js/regexp-oveflow.html

        * pcre/pcre_compile.cpp:
        (calculateCompiledPatternLengthAndFlags): In the case where a single character
        character class is optimized to not use a character class at all, the preflight
        code was not setting the lastitemlength variable.

LayoutTests:

        Reviewed by Eric Seidel.

        - fix http://bugs.webkit.org/show_bug.cgi?id=16321
          new RegExp("[\u0097]{4,6}", "gmy") crashes in DEBUG builds
          <rdar://problem/5632992>

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



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28491 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Alp Toker.

        http://bugs.webkit.org/show_bug.cgi?id=16329
        [GTK] Two small cleanups

        * WebCoreSupport/ChromeClientGtk.cpp:
        (WebKit::ChromeClient::createWindow):
        (WebKit::ChromeClient::addMessageToConsole):
        (WebKit::ChromeClient::runJavaScriptAlert):
        (WebKit::ChromeClient::runJavaScriptConfirm):
        (WebKit::ChromeClient::runJavaScriptPrompt):
        (WebKit::ChromeClient::setStatusbarText):
        (WebKit::ChromeClient::mouseDidMoveOverElement):
        (WebKit::ChromeClient::setToolTip):
        * WebCoreSupport/ChromeClientGtk.h:
        * WebView/webkitprivate.cpp:
        (WebKit::kit):
        s/m_webPage/m_webView/
        * WebView/webkitwebview.cpp:
        Chain up to the parent class to activate bindings instead
        of doing it explicitely.


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28490 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Oliver.

        Fix for <rdar://problem/5631356> Toggling display property on video causes controls to get lost

        * html/HTMLMediaElement.cpp: (WebCore::HTMLMediaElement::attach): Call updateFromElement.
        * html/HTMLMediaElement.h:

LayoutTests:

        Reviewed by Oliver.

        Fix <rdar://problem/5631356> Toggling display property on video causes controls to get lost

        * media/video-display-toggle.html: Added.
        * platform/mac/media/video-display-toggle-expected.checksum: Added.
        * platform/mac/media/video-display-toggle-expected.png: Added.
        * platform/mac/media/video-display-toggle-expected.txt: Added.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28489 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Geoff.

        * WebScriptCallFrame.cpp:
        (WebScriptCallFrame::valueByEvaluatingJavaScriptFromString):



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@28488 268f45cc-cd09-0410-ab3c-d52691b4dbfc