git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20579 268f45cc-cd09-0410-ab3c-d52691b4dbfc

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20578 268f45cc-cd09-0410-ab3c-d52691b4dbfc

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20575 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Doesn't seem to affect the Mac build but is a problem for Qt.


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20574 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        
        WebBaseNetscapePluginStream.m => WebBaseNetscapePluginStream.mm, since
        it's ObjC++ now.

        * Plugins/WebBaseNetscapePluginStream.m: Removed.
        * WebKit.xcodeproj/project.pbxproj:



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20573 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        because of getComputedStyle not reporting display:none.
        http://bugs.webkit.org/show_bug.cgi?id=13103

        * fast/css/computed-style-display-none.html: Added.
        * fast/css/computed-style-display-none-expected.txt: Added.

        * fast/css/computed-style-negative-top.html: Updated to work with Gecko.
        * fast/css/computed-style-negative-top-expected.txt: Newline change needed
        because of change above.

        * fast/css/computed-style.html: Updated to work with Gecko.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20572 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        
        Fixed <rdar://problem/5091330> REGRESSION: Repro crash in 
        -[WebBaseNetscapePluginView(WebNPPCallbacks) destroyStream:reason:] 
        navigating away from page with DivX movie plug-in (13203)
        
        The problem was that the DivX plug-in would ask us to destroy an NPStream
        that had already been destroyed, causing us to wander off into freed
        memory. (I believe the reason this was a regression was that we never used 
        to destroy plug-in streams, period.)
        
        The solution here is to track the NPStreams belonging to a plug-in, and 
        guard against plug-ins making calls with NPStreams that don't belong to 
        them. (It turns out that NPN_DestroyStream is the only stream-based 
        plug-in call we support.)

        (CarbonPathFromPOSIXPath): Fixed up a cast to be C++ compatible.
        * Plugins/WebBaseNetscapePluginView.mm:
        (-[WebBaseNetscapePluginView destroyStream:reason:]): The actual fix.
        Use helper method to guard against a plug-in using an NPStream that doesn't
        belong to it.
        * WebKit.xcodeproj/project.pbxproj: Made WebBaseNetscapePluginView ObjC++
        so I could use HashMap.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20571 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Fix for <rdar://problem/4545060>, technorati and arstechnica fail because of getComputedStyle not
        reporting display:none properly when it is set.  This is bugzilla bug 13103.

        Reviewed by mitz

        * css/CSSComputedStyleDeclaration.cpp:
        (WebCore::CSSComputedStyleDeclaration::getPropertyCSSValue):



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20570 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Darin.

        http://bugs.webkit.org/show_bug.cgi?id=12963
        Fix some inconsistencies in the Mozilla JS Array extras implementations
        with respect to the Mozilla implementation:

          - holes in arrays should be skipped, not treated as undefined,
            by all such methods
          - an element with value undefined is not a hole
          - Array.prototype.forEach should return undefined

        * kjs/array_object.cpp:
        (ArrayInstance::getOwnPropertySlot):
        (ArrayProtoFunc::callAsFunction):

LayoutTests:

        Reviewed by Darin.

         http://bugs.webkit.org/show_bug.cgi?id=12963
         Fix some inconsistencies in the Mozilla JS Array extras implementations
         with respect to the Mozilla implementation:

           - holes in arrays should be skipped, not treated as undefined,
             by all such methods
           - an element with value undefined is not a hole
           - Array.prototype.forEach should retur...

        Reviewed by Hyatt.

        Rename DocLoader::setPasteInProgress, etc to more generic 
        setAllowStaleResources, etc

        * editing/Editor.cpp:
        (WebCore::Editor::paste):
        * loader/DocLoader.cpp:
        (WebCore::DocLoader::DocLoader):
        (WebCore::DocLoader::checkForReload):
        * loader/DocLoader.h:
        (WebCore::DocLoader::setAllowStaleResources):
        * page/DragController.cpp:
        (WebCore::DragController::concludeDrag):



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20568 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Anders and Steve.

        Add a flag to DocLoader to prevent it from reloading 
        when performing a paste or drop operation.
        
        Remove m_cachePolicy from CachedResource as it is never used
        
        Fixes rdar://problem/5044366 : REGRESSION: Repro ASSERT failures 
        dragging image into GMail message in _web_writeImage and 
        documentFragmentFromDragData

        * editing/Editor.cpp:
        (WebCore::Editor::paste):
          prevent the docloader from reloading resources on paste
        * html/HTMLImageLoader.cpp:
        (WebCore::HTMLImageLoader::updateFromElement):
        * loader/Cache.cpp:
        (WebCore::createResource):
        * loader/CachedCSSStyleSheet.cpp:
        (WebCore::CachedCSSStyleSheet::CachedCSSStyleSheet):
        * loader/CachedCSSStyleSheet.h:
        * loader/CachedImage.cpp:
        (WebCore::CachedImage::CachedImage):
        * loader/CachedImage.h:
        * loader/CachedResource.cpp:
        (WebCore::CachedResource::CachedResource):
        * loader/CachedResource.h:
        * loader/CachedScript.cpp:
        (WebCore::CachedScript::CachedScript):
        * loader/CachedScript.h:
        * loader/CachedXBLDocument.cpp:
        (WebCore::CachedXBLDocument::CachedXBLDocument):
        * loader/CachedXBLDocument.h:
        * loader/CachedXSLStyleSheet.cpp:
        (WebCore::CachedXSLStyleSheet::CachedXSLStyleSheet):
        * loader/CachedXSLStyleSheet.h:
        * loader/DocLoader.cpp:
        (WebCore::DocLoader::DocLoader):
           Tidy up constructor initialise new field
        (WebCore::DocLoader::checkForReload):
           Prevent reload when pasting
        * loader/DocLoader.h:
        (WebCore::DocLoader::setPasteInProgress):
          new field and setter
        * page/DragController.cpp:
        (WebCore::DragController::concludeDrag):
          prevent the docloader from reloading resources on drop



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20567 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        * platform/gdk/ScrollViewGdk.cpp:
        (WebCore::ScrollView::contentsToWindow):


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20566 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Darin.
        
        - <rdar://problem/4820817> Autoscroll is broken in Mail.app (message jumps around when trying to select text)

        * fast/overflow/scrollRevealButton-expected.checksum:
        * fast/overflow/scrollRevealButton-expected.png:
        * fast/overflow/scrollRevealButton-expected.txt:

WebCore:

        Reviewed by Darin.
        
        - fixed <rdar://problem/4820817> Autoscroll is broken in Mail.app (message jumps around when trying to select text)
                
        Handle scrolling external scrollviews around the webview during
        autoscroll or when scrolling into view.
        
        This should also restore performance on BenchJS test 6 lost due to Tim's partial fix.

        * page/FrameView.cpp:
        (WebCore::FrameView::scrollRectIntoViewRecursively):
        * page/FrameView.h:
        * platform/ScrollView.h:
        * platform/gdk/ScrollViewGdk.cpp:
        (WebCore::ScrollView::visibleContentRectConsideringExternalScrollers):
        (WebCore::ScrollView::scrollRectIntoViewRecursively):
        * platform/mac/ScrollViewMac.mm:
        (WebCore::ScrollView::visibleContentRect):
        (WebCore::ScrollView::visibleContentRectConsideringExternalScrollers):
        (WebCore::ScrollView::scrollRectIntoViewRecursively):
        (WebCore::ScrollView::updateContents):
        * platform/qt/ScrollViewQt.cpp:
        (WebCore::ScrollView::visibleContentRectConsideringExternalScrollers):
        (WebCore::ScrollView::scrollRectIntoViewRecursively):
        * platform/win/ScrollViewWin.cpp:
        (WebCore::ScrollView::visibleContentRectConsideringExternalScrollers):
        * platform/win/TemporaryLinkStubs.cpp:
        (WebCore::ScrollView::scrollRectIntoViewRecursively):
        * rendering/RenderLayer.cpp:
        (WebCore::RenderLayer::scrollRectToVisible):



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20565 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Update to last fix.

        * Misc/WebKitVersionChecks.h: Added WEBKIT_FIRST_VERSION_WITHOUT_VITALSOURCE_QUIRK.
        * WebView/WebView.mm: (-[WebView stringByEvaluatingJavaScriptFromString:]):
        Added check to only use the VitalSource workaround if the app is not linked on or after
        the defined WEBKIT_FIRST_VERSION_WITHOUT_VITALSOURCE_QUIRK version number.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20560 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Get rid of some misuses of |islower| and |tolower| for non-ASCII characters.
        http://bugs.webkit.org/show_bug.cgi?id=13138

        * platform/StringImpl.cpp:
        (WebCore::StringImpl::isLower):
        (WebCore::StringImpl::lower):
        (WebCore::StringImpl::equalIgnoringCase):



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20559 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Fix <rdar://problem/5092653>
        REGRESSION: Image decoding making scrolling slow
        
        If decoder gave back a null frame (which I think happens regularly if you don't have full frame 
        data yet for image that does not decode incrementally) m_decodedSize was incremented anyway. This
        led to massive growth in calculated image size and eventually made cache prune constantly. 
        
        No test, I don't know how to make one for this.

        * platform/graphics/BitmapImage.cpp:
        (WebCore::BitmapImage::cacheFrame):



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20558 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Kevin M.

        WebCore part of fix for <rdar://problem/5095515> VitalSource Bookshelf should not pass return statements into stringByEvaluatingJavaScriptFromString

        * page/mac/WebCoreFrameBridge.mm: (-[WebCoreFrameBridge stringByEvaluatingJavaScriptFromString:forceUserGesture:]):
        Convert return value to a string.  This matches the behavior in Safari 2.0.

WebKit:

        Reviewed by Kevin M.

        WebKit part of fix for <rdar://problem/5095515> VitalSource Bookshelf should not pass return statements into stringByEvaluatingJavaScriptFromString

        Added an app specific workaround for VitalSource Bookshelf that strips "return" from the beginning of their script strings.  We used to allow this
        but now we throw a JavaScript exception for return statements that aren't in functions.

        Filed this evangelism bug so we can notify VitalSource of the problem:
        <rdar://problem/5095515> VitalSource Bookshelf should not pass return statements into stringByEvaluatingJavaScriptFromString

        * WebView/WebView.mm: (-[WebView stringByEvaluatingJavaScriptFromString:]):



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20555 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        http://bugs.webkit.org/show_bug.cgi?id=13190
        XPath incorrectly handles namespaces on attributes

WebCore:
        * xml/XPathStep.cpp:
        (WebCore::XPath::Step::nodesInAxis): Added a special case for faster attribute lookup; gives a slight but 
        measurable performance improvement for bug 13021.
        (WebCore::XPath::Step::nodeMatches): Fixed NameTest for attribute nodes.

        * xml/XPathStep.h:
        (WebCore::XPath::Step::NodeTest::NodeTest):
        (WebCore::XPath::Step::NodeTest::namespaceURI):
        (WebCore::XPath::Step::nodeTest):
        (WebCore::XPath::Step::setNodeTest):
        Move m_namespaceURI to NodeTest, where it belongs. Removed unused m_nodeTestData (oops!).

        * xml/XPathGrammar.y:
        * xml/XPathPath.cpp:
        (WebCore::XPath::LocationPath::optimizeStepPair):
        Accounted for the above change.

LayoutTests:
        * fast/xpath/attr-namespace-expected.txt: Added.
        * fast/xpath/attr-namespace.html: Added.

        * fast/xpath/xpath-namespaces-expected.txt:
        * fast/xpath/xpath-namespaces.html:
        Cleaned up; added a couple more cases (which passed anyway, but weren't tested for).



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20554 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        * libWebKitSystemInterface.a:



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20552 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        rs=Hyatt.

        Rollout local changes :-/

        * loader/Cache.cpp:
        (WebCore::Cache::pruneAllResources):
        (WebCore::Cache::remove):



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20551 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Hyatt.

        Roll out r20511 due to incredible badness

        * loader/Cache.cpp:
        (WebCore::Cache::pruneAllResources):
        (WebCore::Cache::remove):



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20550 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        different results due to the painting Adele added to DumpRenderTree. Long term,
        we just have to fix SVG so that painting doesn't cause changes to the render tree!



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20546 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Adam.

        Don't enforce text-selection drag delay when we are
        going to drag an image anyway.

        Fixes rdar://problem/4668935

        * page/EventHandler.cpp:
        (WebCore::EventHandler::handleDrag):
Layout tests:
        Test to ensure there is no delay forced when dragging a selected image

        * editing/pasteboard/drag-selected-image-to-contenteditable-expected.checksum: Added.
        * editing/pasteboard/drag-selected-image-to-contenteditable-expected.png: Added.
        * editing/pasteboard/drag-selected-image-to-contenteditable-expected.txt: Added.
        * editing/pasteboard/drag-selected-image-to-contenteditable.html: Added.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20545 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Geoff.

        * bindings/NP_jsobject.cpp:
        (_NPN_InvokeDefault):
        Call JSObject:call for native JavaScript objects.

LayoutTests:

        Reviewed by Geoff.

        Add NPN_InvokeDefault test.
        
        * plugins/netscape-invoke-default-expected.txt: Added.
        * plugins/netscape-invoke-default.html: Added.

WebKitTools:

        Reviewed by Geoff.

        * DumpRenderTree/TestNetscapePlugIn.subproj/PluginObject.c:
        (pluginInvoke):
        Add new function which takes a function and calls it using NPN_InvokeDefault.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20544 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        
        * editing/markup.cpp:
        (WebCore::startMarkup):



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20543 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by darin
        
        <rdar://problem/5091898> REGRESSION: GMail Editor: A line of underlined text copied with Select All pastes with wrong font family

        Demonstrates the bug fixed:
        * editing/style/5091898-expected.checksum: Added.
        * editing/style/5091898-expected.png: Added.
        * editing/style/5091898-expected.txt: Added.
        * editing/style/5091898.html: Added.
        Demonstrates a related bug fixed (deleting
        the line break before a paragraph could remove
        its underlining):
        * editing/deleting/5091898-expected.checksum: Added.
        * editing/deleting/5091898-expected.png: Added.
        * editing/deleting/5091898-expected.txt: Added.
        * editing/deleting/5091898.html: Added.
        Demonstrates that a link now (correctly) exists, 
        instead of just underlined text:
        * editing/pasteboard/4840662-expected.txt:

WebCore:

        Reviewed by darin
        
        <rdar://problem/5091898> REGRESSION: GMail Editor: A line of underlined text copied with Select All pastes with wrong font family
        
        createMarkup skipped elements if they were blocks
        when called from moveParagraphs (because that function 
        must receive only inline content).  This patch adds 
        code to inline these blocks instead of skipping them 
        so that we don't lose any of the style that they 
        contribute to the copied markup.

        * editing/markup.cpp:
        (WebCore::startMarkup): Add an inlineBlocks option.
        Make sure to overwrite display:block coming from
        a style sheet or the inline style declaration.
        (WebCore::createMarkup): Don't refuse to include a 
        specialCommonAncestor that's a block if we were asked 
        to include only inline content, since we can now inline 
        block elements in startMarkup.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20542 268f45cc-cd09-0410-ab3c-d52691b4dbfc

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20541 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        * DumpRenderTree/DumpRenderTree.m:
        (-[WaitUntilDoneDelegate webView:didFinishLoadForFrame:]):



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20540 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        * DumpRenderTree/DumpRenderTree.m:
        (-[WaitUntilDoneDelegate webView:didFinishLoadForFrame:]):



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20539 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        
        - fixed <rdar://problem/5092556> Default UA spoofing is always off until explicitly toggled

        * WebView/WebView.mm:
        (-[WebView _commonInitializationWithFrameName:groupName:]):
        initialize the cached value of _private->useSiteSpecificSpoofing here; formerly it would not
        be initialized correctly in the common case of WebViews that use [WebPreferences standardPreferences]



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20537 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Fix for <rdar://problem/5026141> REGRESSION: Aperture Viewer Flashes when switching from Index Page to Detail page in Web Gallery

        This disables the paint that occurs after WebCore finishes loading the document.  We should reenable this when we
        move the dispatchDidFinishLoad delegate callback to happen around the same time as the JS onload call.  See <rdar://problem/5092361>.

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



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20535 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by darin
        
        <rdar://problem/5065910> REGRESSION: In Mail and GMail, Can't apply color to pasted line of text
        <rdar://problem/5084241> After deleting a paragraph of colored text, can't change text color
        <rdar://problem/5065605> REGRESSION: Text copied with Select All + Copy loses color on Paste
        <rdar://problem/4916887> GMAIL: Can't apply color to text once it becomes indented

        * editing/pasteboard/5065605-expected.checksum: Added.
        * editing/pasteboard/5065605-expected.png: Added.
        * editing/pasteboard/5065605-expected.txt: Added.
        * editing/pasteboard/5065605.html: Added.
        * editing/style/4916887-expected.checksum: Added.
        * editing/style/4916887-expected.png: Added.
        * editing/style/4916887-expected.txt: Added.
        * editing/style/4916887.html: Added.
        * editing/style/5065910-expected.checksum: Added.
        * editing/style/5065910-expected.png: Added.
        * editing/style/5065910-expected.txt: Added.
        * editing/style/5065910.html: Added.
        * editing/style/5084241-expected.checksum: Added.
        * editing/style/5084241-expected.png: Added.
        * editing/style/5084241-expected.txt: Added.
        * editing/style/5084241.html: Added.

WebCore:

        Reviewed by darin
        
        <rdar://problem/5065910> REGRESSION: In Mail and GMail, Can't apply color to pasted line of text
        <rdar://problem/5084241> After deleting a paragraph of colored text, can't change text color
        <rdar://problem/5065605> REGRESSION: Text copied with Select All + Copy loses color on Paste
        <rdar://problem/4916887> GMAIL: Can't apply color to text once it becomes indented

        * css/CSSComputedStyleDeclaration.cpp:
        (WebCore::CSSComputedStyleDeclaration::copyInheritableProperties):
        If a node's text fill color is invalid, then its children use 
        their font-color as their text fill color (they don't inherit 
        it).  Likewise for stroke color.



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20520 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Use wtf/MathExtras.h instead of math.h



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20518 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Dave Harrison.

        * Configurations/WebKit.xcconfig: Include UMBRELLA_FRAMEWORKS_DIR in framework search path.


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20513 268f45cc-cd09-0410-ab3c-d52691b4dbfc

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20512 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Reviewed by Maciej.

        The WebArchiver needs all the subresources of a document to be accessible
        through the DocLoader, so we can't remove a CachedResource simultaneously
        from the global Cache and all DocLoaders, just because it needs to be 
        reloaded in one.

        Fixes rdar://problem/5044366

        * loader/Cache.cpp:
        (WebCore::Cache::remove):
           Don't evict the resource from all DocLoaders
        * manual-tests/test-iframes-loading-the-same-resource.html: Added.
           Alas it wasn't possible to make a nice layout test, even this manual test
           is not entirely deterministic



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20511 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        Build fix.

        * platform/GlyphBuffer.h:
        (WebCore::GlyphBuffer::clear):


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20510 268f45cc-cd09-0410-ab3c-d52691b4dbfc

asserts and improve a bit the layout scheduling in the
Qt port


git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20509 268f45cc-cd09-0410-ab3c-d52691b4dbfc

git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20508 268f45cc-cd09-0410-ab3c-d52691b4dbfc

        * platform/mac/PlatformScrollBarMac.mm:
        (WebCore::PlatformScrollbar::scrollbarHit):



git-svn-id: http://svn.webkit.org/repository/webkit/trunk@20507 268f45cc-cd09-0410-ab3c-d52691b4dbfc