16 years ago Reviewed by Tim Omernick.
sullivan [Tue, 11 Apr 2006 04:17:43 +0000 (04:17 +0000)]
    Reviewed by Tim Omernick.

        - fixed <rdar://problem/4139799> Seed: Safari: Private Browsing leaves traces in Icon Cache

        * Misc/WebIconDatabasePrivate.h:
        new ivars: pageURLsBoundDuringPrivateBrowsing, iconURLsBoundDuringPrivateBrowsing, and privateBrowsingEnabled

        * Misc/WebIconDatabase.m:
        (-[NSMutableDictionary init]):
        initialize new ivars, and listen for notifications that WebPreferences changed so we can react to changes
        to private browsing.
        (-[NSMutableDictionary iconForURL:withSize:cache:]):
        Don't remove icon URL from extraRetain dictionary; that's now done in _forgetIconForIconURLString. (I left a
        comment here earlier about why I was worried about this change, but I convinced myself that it's fine.)
        (-[WebIconDatabase removeAllIcons]):
        Removed no-longer-true (and never very clear) comment, and braces. Also remove all objects from the two
        private-browsing-related dictionaries.
        (-[WebIconDatabase _setIcon:forIconURL:]):
        remember icon URL if private browsing is enabled
        (-[WebIconDatabase _setHaveNoIconForIconURL:]):
        remember icon URL if private browsing is enabled
        (-[WebIconDatabase _setIconURL:forURL:]):
        added an assert that helped me out at one point
        (-[WebIconDatabase _clearDictionaries]):
        clear the two new dictionaries too
        (-[WebIconDatabase _loadIconDictionaries]):
        made an existing ERROR not fire in the expected case where there are no icons at all on disk
        (-[WebIconDatabase _updateFileDatabase]):
        when saving the pageURLToIconURL dictionary to disk, first remove any values that were created during
        private browsing
        (-[WebIconDatabase _retainIconForIconURLString:]):
        skip the code that deals with saving changes to disk if private browsing is enabled
        (-[WebIconDatabase _forgetIconForIconURLString:]):
        Remove the icon URL from extraRetain dictionary here. We're forgetting everything about this icon URL
        so we should forget its former extraRetain count too.
        (-[WebIconDatabase _resetCachedWebPreferences:]):
        Cache the new value of private browsing. If it has now been turned off, forget everything we learned
        while it was on. This causes (e.g.) icons for bookmarks or pre-existing history items to be forgotten
        if the icon was only learned during private browsing.

        * History/WebHistoryItem.m:
        removed an unnecessary #import I happened to notice

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13817 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago * fast/dom/gc-4-expected.txt: Add one blank line (test result was failing).
darin [Mon, 10 Apr 2006 23:36:49 +0000 (23:36 +0000)]
    * fast/dom/gc-4-expected.txt: Add one blank line (test result was failing).

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13816 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago - try to fix the Windows build
darin [Mon, 10 Apr 2006 23:30:25 +0000 (23:30 +0000)]
    - try to fix the Windows build

        * WebCore.vcproj/WebCore/WebCore.vcproj: Fix mistaken bad editing of

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13815 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Make focus ring painting respect clips set by WebCore (e.g., overflow).
hyatt [Mon, 10 Apr 2006 22:30:42 +0000 (22:30 +0000)]
    Make focus ring painting respect clips set by WebCore (e.g., overflow).

        Reviewed by darin

        * platform/GraphicsContext.h:
        * platform/mac/GraphicsContextMac.mm:
        * platform/mac/WebCoreGraphicsBridge.h:
        * platform/mac/WebCoreGraphicsBridge.m:
        (-[WebCoreGraphicsBridge drawFocusRingWithPath:radius:color:clipRect:]):
        * rendering/RenderLayer.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13814 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Make the broken CG focus ring painting work when WebCore sets a clip
hyatt [Mon, 10 Apr 2006 22:29:13 +0000 (22:29 +0000)]
    Make the broken CG focus ring painting work when WebCore sets a clip
        (in addition to respecting the dirty rect clip).

        Reviewed by darin

        * WebCoreSupport/WebGraphicsBridge.m:
        (-[WebGraphicsBridge drawFocusRingWithPath:radius:color:clipRect:]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13813 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Geoff.
darin [Mon, 10 Apr 2006 22:12:30 +0000 (22:12 +0000)]
    Reviewed by Geoff.

        - death to khtml/ecma, long live bindings/js

        * khtml/ecma: Removed. Moved all files to bindings/js.

        * bindings/js/JSDOMParser.cpp: Added.
        * bindings/js/JSDOMParser.h: Added.
        * bindings/js/JSXMLHttpRequest.cpp: Added.
        * bindings/js/JSXMLHttpRequest.h: Added.
        * bindings/js/JSXMLSerializer.cpp: Added.
        * bindings/js/JSXMLSerializer.h: Added.
        * bindings/js/JSXSLTProcessor.cpp: Added.
        * bindings/js/JSXSLTProcessor.h: Added.
        * bindings/js/kjs_binding.cpp: Added.
        * bindings/js/kjs_binding.h: Added.
        * bindings/js/kjs_css.cpp: Added.
        * bindings/js/kjs_css.h: Added.
        * bindings/js/kjs_dom.cpp: Added.
        * bindings/js/kjs_dom.h: Added.
        * bindings/js/kjs_events.cpp: Added.
        * bindings/js/kjs_events.h: Added.
        * bindings/js/kjs_html.cpp: Added.
        * bindings/js/kjs_html.h: Added.
        * bindings/js/kjs_navigator.cpp: Added.
        * bindings/js/kjs_navigator.h: Added.
        * bindings/js/kjs_proxy.cpp: Added.
        * bindings/js/kjs_proxy.h: Added.
        * bindings/js/kjs_traversal.cpp: Added.
        * bindings/js/kjs_traversal.h: Added.
        * bindings/js/kjs_views.cpp: Added.
        * bindings/js/kjs_views.h: Added.
        * bindings/js/kjs_window.cpp: Added.
        * bindings/js/kjs_window.h: Added.

        * DerivedSources.make: Removed khtml/ecma from directory list.
        * WebCore.vcproj/WebCore/WebCore.vcproj: Moved files from
        khtml/ecma to bindings/js.
        * WebCore.xcodeproj/project.pbxproj: Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13812 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago - one more test change for http://bugzilla.opendarwin.org/show_bug.cgi?id...
darin [Mon, 10 Apr 2006 21:40:51 +0000 (21:40 +0000)]
    - one more test change for bugzilla.opendarwin.org/show_bug.cgi?id=4884
          Canvas element breaks when RenderObject creation is deferred by external CSS

        * fast/dom/gc-4.html: Test had a bug where it used <canvas>
        without a </canvas> tag. Changed test to include a </canvas> tag.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13810 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Geoff.
darin [Mon, 10 Apr 2006 21:38:16 +0000 (21:38 +0000)]
    Reviewed by Geoff.

        - try to fix the Windows build

        * html/HTMLCanvasElement.cpp: (WebCore::HTMLCanvasElement::reset):
        Put code to release m_drawingContext into an __APPLE__ ifdef.

        * platform/win/TemporaryLinkStubs.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13809 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Beth.
ggaren [Mon, 10 Apr 2006 21:10:00 +0000 (21:10 +0000)]
    Reviewed by Beth.

        * projects/compat/hitlist.html: updated dates

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13808 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Fix by Mitz Pettel, reviewed by Eric.
ap [Mon, 10 Apr 2006 17:02:17 +0000 (17:02 +0000)]
    Fix by Mitz Pettel, reviewed by Eric.

        - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=8295
          Dictionary pop-up panel targets the wrong word in a scrolled IFRAME

        * kwq/WebCoreAXObject.mm:
        (-[WebCoreAXObject doAXTextMarkerForPosition:]): Removed the addition of scroll
        offsets, which is redundant for scrolled views, then changed the first view
        to be the document's scrolled view instead of its scroll view (all subsequent views were
        already scrolled views).
        * manual-tests/dictionary-scrolled-iframe.html: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13754 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoLayoutTests:
darin [Mon, 10 Apr 2006 16:49:04 +0000 (16:49 +0000)]

        - regenerated pixel results because many may be affected by
          the fix for bug 8100


        Reviewed by Darin.

        - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=8100
          REGRESSION (NativeTextField): Right-aligned and RTL text jitters in new text fields

        * WebCoreSupport/WebTextRenderer.m:
        (overrideLayoutOperation): For RTL runs, apply the word-rounding on the left.
        (CG_floatWidthForRun): For RTL runs, apply the last character's rounding
        on the left by adjusting the start position.
        (initializeWidthIterator): Added finalRoundingWidth field to WidthIterator.
        (advanceWidthIterator): For RTL runs, apply rounding on the left of the character,
        by increasing the width of the next character (which is the character to the left). For
        the last character, keep the rounding width in the iterator's finalRoundingWidth, to be used by

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13753 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by eric.
ggaren [Mon, 10 Apr 2006 05:06:35 +0000 (05:06 +0000)]
    Reviewed by eric.

        - Fixed http://bugzilla.opendarwin.org/show_bug.cgi?id=8284
        prevent unnecessary entries in the "nodes with extra refs" hash table

        This patch switches manual RefPtr exchange with use of
        RefPtr::release to ensure that a node's ref count never tops 1
        (in the normal case).

        * kjs/nodes.cpp:
        * kjs/nodes.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13752 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Darin.
ap [Mon, 10 Apr 2006 04:40:28 +0000 (04:40 +0000)]
    Reviewed by Darin.

        - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=7877
        XMLHttpRequest ignores username/password passed to open()

        Test: http/tests/xmlhttprequest/basic-auth.html

        * platform/KURL.cpp:
        (KURL::setUser): Enable a code path that handles non-empty user name -
        it was already present, but commented out and protected with an assertion.
        (KURL::setPass): Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13751 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoRemoved bogus suffix I checked in by accident.
darin [Mon, 10 Apr 2006 02:45:30 +0000 (02:45 +0000)]
Removed bogus suffix I checked in by accident.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13750 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoFirst early steps towards a next, much smaller, round of renames
darin [Sun, 9 Apr 2006 23:48:06 +0000 (23:48 +0000)]
First early steps towards a next, much smaller, round of renames
in WebCore.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13749 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoLayoutTests:
darin [Sun, 9 Apr 2006 23:38:06 +0000 (23:38 +0000)]

        Reviewed by Anders.

        - test changes for http://bugzilla.opendarwin.org/show_bug.cgi?id=4884
          Canvas element breaks when RenderObject creation is deferred by external CSS

        * fast/canvas/canvas-before-css-expected.checksum: Added.
        * fast/canvas/canvas-before-css-expected.png: Added.
        * fast/canvas/canvas-before-css-expected.txt: Added.
        * fast/canvas/canvas-before-css.html: Added.

        * fast/canvas/image-object-in-canvas.html: Test had a bug where it used <canvas>
        without a </canvas> tag. Changed test to include a </canvas> tag.

        * fast/canvas/patternfill-repeat-expected.txt: Updated for name change from
        RenderCanvasImage to RenderHTMLCanvas.
        * fast/canvas/image-object-in-canvas-expected.txt: Ditto.
        * fast/canvas/fillrect_gradient-expected.txt: Ditto.
        * fast/canvas/quadraticCurveTo-expected.txt: Ditto.


        Reviewed by Anders.

        - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=4884
          Canvas element breaks when RenderObject creation is deferred by external CSS

        Test: fast/canvas/canvas-before-css.html

        This patch makes us match the canvas documentation in Hixie's Web Applications
        draft as far as when the canvas is created and recreated and how it's sized.
        It also gets rid of the compositeOperation attribute of the canvas element.
        We can add that back if we need it. Anders points out that this specifically
        changes behavior for canvas elements where the size is set in CSS and not with
        width and height attributes. The CSS size now determines how big a box the canvas
        is rendered into, but has no effect on the size of the canvas's buffer.

        * html/CanvasRenderingContext2D.h: Added overloads of drawImage that take
        HTMLCanvasElement, which is no longer derived from HTMLImageElement.
        * html/CanvasRenderingContext2D.cpp:
        (WebCore::imageSize): Renamed from imageOrCanvasSize. Now used for images only,
        because canvas is no longer derived from image.
        (WebCore::CanvasRenderingContext2D::drawImage): Split the implementation of this
        for image sources from the implementation for canvas sources.
        (WebCore::CanvasRenderingContext2D::willDraw): Changed to call a new willDraw
        function on the canvas element.
        (WebCore::CanvasRenderingContext2D::drawingContext): Changed to call drawingContext
        on the canvas element rather than the renderer.

        * html/HTMLCanvasElement.h: Changed HTMLCanvasElement to derive from HTMLElement
        instead of HTMLImageElement. Added width, height, setWidth, setHeight, willDraw,
        paint, drawingContext, createDrawingContext, and reset functions. Added m_size,
        m_createdDrawingContext, m_data, and m_drawingContext data members. Removed
        mapToEntry, attach, detach, and isURLAttribute functins.

        * html/HTMLCanvasElement.cpp:
        (WebCore::HTMLCanvasElement::HTMLCanvasElement): Added initializers for new m_size,
        m_createdDrawingContext, m_data, and m_drawingContext data members.
        (WebCore::HTMLCanvasElement::~HTMLCanvasElement): Free m_data and m_drawingContext.
        (WebCore::HTMLCanvasElement::parseMappedAttribute): Got rid of special case for
        srcAttr, which is no longer needed since we aren't deriving from HTMLImageElement.
        Added code that triggers a reset when either width or height is set.
        (WebCore::HTMLCanvasElement::createRenderer): Added code to set the intrinsic
        width and height of the renderer to the size of the element.
        (WebCore::HTMLCanvasElement::setHeight): Added. Sets the height attribute.
        (WebCore::HTMLCanvasElement::setWidth): Added. Sets the width attribute.
        (WebCore::HTMLCanvasElement::willDraw): Added. Tells the renderer to repaint.
        Also has FIXME mentioning we could dirty only the part that has changed in the future.
        (WebCore::HTMLCanvasElement::reset): Added. Sets the size of the canvas and discards
        the old buffer, which is an indirect way of resetting the buffer to transparent black.
        (WebCore::HTMLCanvasElement::paint): Added. Draws the canvas image into the graphics
        context that's passed in.
        (WebCore::HTMLCanvasElement::createDrawingContext): Added. Allocates a buffer for
        the bits, then creates a bitmap context for drawing into the buffer.
        (WebCore::HTMLCanvasElement::drawingContext): Added. Calls createDrawingContext if
        needed, then returns the current drawing context.
        (WebCore::HTMLCanvasElement::createPlatformImage): Changed to always call CGContextFlush
        and to create the image from the context in this class.

        * rendering/RenderHTMLCanvas.h: Remove almost all of the contents of this file.
        Removed ~RenderHTMLCanvas, setNeedsImageUpdate, element, updateDrawnImage, drawingContext,
        createDrawingContext, and drawnImage functions and _drawingContext, _drawingContextData,
        _drawnImage, and _needsImageUpdate booleans. Changed RenderHTMLCanvas to derive from
        RenderReplaced instead of RenderImage.

        * rendering/RenderHTMLCanvas.cpp:
        (WebCore::RenderHTMLCanvas::RenderHTMLCanvas): Changed to only initialize RenderReplaced.
        (WebCore::RenderHTMLCanvas::renderName): Moved this in here, since there's no good reason
        to have this virtual function inlined.
        (WebCore::RenderHTMLCanvas::paint): Changed implementation to use HTMLCanvasElement::paint
        instead ofcalling CGContextDrawImage directly.
        (WebCore::RenderHTMLCanvas::layout): Removed the code that detects changes in width and
        causes the drawing context to be recreated; instead, if the width and height changes we
        scale when we paint the canvas.

        * bindings/js/JSCanvasRenderingContext2DBase.cpp:
        Separated out handling for <canvas> vs. <img> elements in drawRect, since
        HTMLCanvasElement is no longer derived from HTMLImageElement.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13748 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago2006-04-09 Rob Buis <buis@kde.org>
eseidel [Sun, 9 Apr 2006 22:00:40 +0000 (22:00 +0000)]
2006-04-09  Rob Buis  <buis@kde.org>

        Reviewed by eseidel.  Landed by eseidel.

        Fix for http://bugzilla.opendarwin.org/show_bug.cgi?id=6027:
        Dirty rect invalidation issues in mozilla sample

        Make sure the paths calculate the new bounding box and not
        use the cached bbox.

        No automated test case possible.

        * kcanvas/RenderPath.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13747 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago2006-04-09 Rob Buis <buis@kde.org>
eseidel [Sun, 9 Apr 2006 21:55:41 +0000 (21:55 +0000)]
2006-04-09  Rob Buis  <buis@kde.org>

        Reviewed by darin.  Landed by eseidel.

        Fix for http://bugzilla.opendarwin.org/show_bug.cgi?id=6930:
        % width/height on nested <svg> tags do not work

        Set the context correctly for inner <svg>, so calculation of
        width/height for inner <svg> elements is done against the
        viewport element.

        Test: svg/custom/inner-percent.svg

        * ksvg2/svg/SVGSVGElement.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13746 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago2006-04-08 Rob Buis <buis@kde.org>
eseidel [Sun, 9 Apr 2006 20:12:06 +0000 (20:12 +0000)]
2006-04-08  Rob Buis  <buis@kde.org>

        Reviewed by eseidel.

        No automated test case possible.

        Fix for http://bugzilla.opendarwin.org/show_bug.cgi?id=7531:
        hang in SVGPolygonElementImpl::toPathData in polygon test case

        Make sure the points list is cleared, just like the path
        list is cleared first before (re)parsing.

        * ksvg2/svg/SVGPolyElement.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13745 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoForgot to commit ChangeLog for JSC string capitalization fixes.
ap [Sun, 9 Apr 2006 08:03:17 +0000 (08:03 +0000)]
Forgot to commit ChangeLog for JSC string capitalization fixes.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13744 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoLayoutTests:
darin [Sun, 9 Apr 2006 05:08:56 +0000 (05:08 +0000)]

        Reviewed by Beth.

        - tests for http://bugzilla.opendarwin.org/show_bug.cgi?id=7622
          REGRESSION: New text fields should compute maxlength considering composed character sequences

        * fast/forms/input-text-maxlength-expected.checksum: Added.
        * fast/forms/input-text-maxlength-expected.png: Added.
        * fast/forms/input-text-maxlength-expected.txt: Added.
        * fast/forms/input-text-maxlength.html: Added.
        * fast/forms/input-text-paste-maxlength-expected.checksum: Added.
        * fast/forms/input-text-paste-maxlength-expected.png: Added.
        * fast/forms/input-text-paste-maxlength-expected.txt: Added.
        * fast/forms/input-text-paste-maxlength.html: Added.


        Reviewed by Beth.

        - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=7622
          REGRESSION: New text fields should compute maxlength considering composed character sequences

        Test: fast/forms/input-text-maxlength.html
        Test: fast/forms/input-text-paste-maxlength.html

        * html/HTMLInputElement.h: Removed all friend classes (not needed any more). Renamed
        typeEnum to InputType. Made init() function private. Tweaked parameter names and formatting
        in many function declarations. Made canHaveSelection, selectionStart, and selectionEnd
        const. Made data members except for m_name private instead of protected. Added private
        functions constrainValue and recheckValue. Removed unused isEditable function.

        * html/HTMLInputElement.cpp:
        (WebCore::numGraphemeClusters): Added.
        (WebCore::numCharactersInGraphemeClusters): Added.
        (WebCore::HTMLInputElement::isKeyboardFocusable): Use inputType() instead of using m_type
        (WebCore::HTMLInputElement::isMouseFocusable): Ditto.
        (WebCore::HTMLInputElement::focus): Ditto.
        (WebCore::HTMLInputElement::setInputType): Added code to call constrainValue or recheckValue
        so we will enforce maxLen if changing from a type that doesn't have maxLen to one that does.
        (WebCore::HTMLInputElement::type): More-standard formatting for switch statement. Put
        cases into alphabetical order.
        (WebCore::HTMLInputElement::state): Changed switch statement to include all case values and
        not include a default case to take advantage of gcc's missing case warning.
        (WebCore::HTMLInputElement::restoreState): Ditto.
        (WebCore::HTMLInputElement::canHaveSelection): Ditto.
        (WebCore::HTMLInputElement::selectionStart): Ditto.
        (WebCore::HTMLInputElement::selectionEnd): Ditto.
        (WebCore::HTMLInputElement::setSelectionStart): Ditto.
        (WebCore::HTMLInputElement::setSelectionEnd): Ditto.
        (WebCore::HTMLInputElement::select): Ditto.
        (WebCore::HTMLInputElement::setSelectionRange): Ditto.
        (WebCore::HTMLInputElement::click): Ditto.
        (WebCore::HTMLInputElement::accessKeyAction): Ditto.
        (WebCore::HTMLInputElement::parseMappedAttribute): Use inputType() instead of using m_type
        directly. Added call to recheckValue when parsing a new value for the maxlength attribute.
        (WebCore::HTMLInputElement::rendererIsNeeded): Changed switch statement to include all case
        values and not include a default case to take advantage of gcc's missing case warning.
        (WebCore::HTMLInputElement::createRenderer): Ditto.
        (WebCore::HTMLInputElement::attach): Remove code to condition the value attribute when done
        parsing. This is now all handled by constrainValue and recheckValue as needed.
        (WebCore::HTMLInputElement::isSuccessfulSubmitButton): Use inputType() instead of using
        m_type directly.
        (WebCore::HTMLInputElement::appendFormData): Ditto. Rearranged code a little.
        (WebCore::HTMLInputElement::setChecked): Ditto.
        (WebCore::HTMLInputElement::setIndeterminate): Ditto.
        (WebCore::HTMLInputElement::value): Ditto. Call constrainValue when reading the value out
        of the value attribute.
        (WebCore::HTMLInputElement::valueWithDefault): Use inputType() instead of using
        m_type directly. Changed switch statement to include all case values and not include a
        default case to take advantage of gcc's missing case warning.
        (WebCore::HTMLInputElement::setValue): Ditto. Call constrainValue when storing a value.
        (WebCore::HTMLInputElement::setValueFromRenderer): Added an assertion.
        (WebCore::HTMLInputElement::storesValueSeparateFromAttribute): Use inputType() instead of
        using m_type directly.
        (WebCore::HTMLInputElement::preDispatchEventHandler): Ditto.
        (WebCore::HTMLInputElement::postDispatchEventHandler): Ditto.
        (WebCore::HTMLInputElement::defaultEventHandler): Ditto. Changed code to truncate inserted
        text in a BeforeTextInsertedEvent to use the new constrainValue function and also the
        numGraphemeClusters function, so it's based on grapheme clusters instead of characters and
        shares code.
        (WebCore::HTMLInputElement::constrainValue): Added.
        (WebCore::HTMLInputElement::recheckValue): Added.

        * html/HTMLIsIndexElement.cpp: (WebCore::HTMLIsIndexElement::HTMLIsIndexElement):
        Removed unneeded code to set m_type to TEXT, which is already what it gets set to by
        the base class's constructor.

        * html/HTMLGenericFormElement.h: Removed unused isEditable function.
        * html/HTMLGenericFormElement.cpp: Ditto.
        * html/HTMLTextAreaElement.h: Ditto.
        * html/HTMLTextAreaElement.cpp: Ditto.

        * platform/StringImpl.cpp: (WebCore::StringImpl::truncate): Changed > to >= so that
        truncating to the size of the string does nothing, efficiently.

        * rendering/RenderText.h: Added declaration of characterBreakIterator.
        * rendering/RenderText.cpp:
        (WebCore::characterBreakIterator): Made this public so it can be used in other files.
        Maybe we should also move it to another source file later. Also renamed to remove the
        "get" from the title.
        (WebCore::RenderText::previousOffset): Updated for name change.
        (WebCore::RenderText::nextOffset): Updated for name change.

        * rendering/RenderTextField.cpp: (WebCore::RenderTextField::updateFromElement):
        Removed code to implement maxlength checking. That's handled entirely in the DOM now.
        Also moved down the code to get the value into a string so that it's done only in the
        case where the string is used.

        * rendering/render_form.cpp: (WebCore::RenderFileButton::valueChanged): Use setValueFromRenderer
        instead of setting the value directly in the input element. We changed this for all the other
        types a while ago, and it works just as well for the input element.

        * dom/BeforeTextInsertedEvent.h: Added setText function. Previously, clients changed the text
        by modifying the text object in place, but going forward we'd like to avoid that sort of thing.

        * editing/ReplaceSelectionCommand.cpp: (WebCore::ReplacementFragment::ReplacementFragment):
        Changed code to assume clients will change the text in the event rather than mutating the
        text object itself. This is compatible with possible future changes to String to be copy
        on write.
        * editing/TypingCommand.cpp: (WebCore::TypingCommand::insertText): Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13743 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Darin.
ap [Sat, 8 Apr 2006 19:04:29 +0000 (19:04 +0000)]
    Reviewed by Darin.

        One more attempt - use reinterpret_cast, rather than static_cast.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13742 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Darin.
ap [Sat, 8 Apr 2006 18:54:06 +0000 (18:54 +0000)]
    Reviewed by Darin.

        An attempt to fix Win32 build - ICU uses wchar_t on Windows, so we need a type cast.

        * kxmlcore/unicode/icu/UnicodeIcu.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13741 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Darin.
ap [Sat, 8 Apr 2006 17:21:52 +0000 (17:21 +0000)]
    Reviewed by Darin.

        - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=8264
        toLowerCase and toUpperCase don't honor special mappings

        Test: fast/js/string-capitalization.html

        * JavaScriptCore.xcodeproj/project.pbxproj: Added KXMLCore::Unicode headers to the project.
        * icu/unicode/putil.h: Added (copied from WebCore).
        * icu/unicode/uiter.h: Ditto.
        * icu/unicode/ustring.h: Ditto.
        * kjs/string_object.cpp:
        (StringProtoFunc::callAsFunction): Use the new KXMLCore::Unicode::toUpper() and toLower().
        * kjs/ustring.cpp: Removed unused (and evil) UChar::toLower() and toUpper().
        * kjs/ustring.h: Ditto.

        * kxmlcore/unicode/Unicode.h: Corrected capitalization of the word Unicode.
        * kxmlcore/unicode/UnicodeCategory.h: Renamed include guard macro to match file name.

        * kxmlcore/unicode/icu/UnicodeIcu.h:
        (KXMLCore::Unicode::toLower): Work on strings, not individual characters. Use ICU root locale.
        (KXMLCore::Unicode::toUpper): Ditto.
        (KXMLCore::Unicode::isFormatChar): Use int32_t, which can hold a complete code point.
        (KXMLCore::Unicode::isSeparatorSpace): Ditto.
        (KXMLCore::Unicode::category): Ditto.
        * kxmlcore/unicode/qt4/UnicodeQt4.h:
        (KXMLCore::Unicode::toLower): Work on strings, not individual characters.
        (KXMLCore::Unicode::toUpper): Ditto.
        (KXMLCore::Unicode::isFormatChar): Use int32_t, which can hold a complete code point.
        (KXMLCore::Unicode::isSeparatorSpace): Ditto.
        (KXMLCore::Unicode::category): Ditto.

        * tests/mozilla/ecma/String/ Corrected expected results.
        * tests/mozilla/ecma/String/ Corrected expected results.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13740 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Adele Peterson.
sullivan [Sat, 8 Apr 2006 16:49:21 +0000 (16:49 +0000)]
    Reviewed by Adele Peterson.

        - fixed http://bugzilla.opendarwin.org/show_bug.cgi?id=8260
        REGRESSION: Assertion failure: ![_private->iconsToSaveWithURLs containsObject:iconURLString]
        in WebIconDatabase.m:695-[WebIconDatabase(WebInternal) _retainIconForIconURLString:]

        * Misc/WebIconDatabase.m:
        (-[WebIconDatabase _retainIconForIconURLString:]):
        This new assertion was one block too high; moved it in.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13739 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoLayoutTests:
darin [Sat, 8 Apr 2006 00:34:20 +0000 (00:34 +0000)]

        Reviewed by Hyatt.

        - test for http://bugzilla.opendarwin.org/show_bug.cgi?id=8134
          REGRESSION: dragging down from the middle of a text field does not select to end of field

        * fast/forms/input-text-drag-down-expected.checksum: Added.
        * fast/forms/input-text-drag-down-expected.png: Added.
        * fast/forms/input-text-drag-down-expected.txt: Added.
        * fast/forms/input-text-drag-down.html: Added.


        Reviewed by Hyatt.

        - test for http://bugzilla.opendarwin.org/show_bug.cgi?id=8134
          REGRESSION: dragging down from the middle of a text field does not select to end of field

        * editing/Selection.cpp:
        (WebCore::comparePositions): Added. Takes shadow content into account.
        (WebCore::Selection::validate): Changed to call comparePositions instead of calling
        Range::compareBoundaryPoints directly. Also removed unneeded code to redundantly set
        m_start and m_end to null and did a bit of reformatting.

        - some tiny efficiency improvements to the tokenizer -- no measurable speedup, but removes
          a little bit of unneeded code

        * html/HTMLTokenizer.cpp: (WebCore::HTMLTokenizer::parseTag): Changed all the places
        that do "unsigned short x = *c" to "unsigned short x = c->unicode()" when c is a QChar,
        otherwise we do an unnecessary conversion to char (which requires a branch to see if
        the c fits in a char).

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13738 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by harrison
justing [Sat, 8 Apr 2006 00:21:53 +0000 (00:21 +0000)]
    Reviewed by harrison

        Rolled the fix for 8250 back in and fixed a bug:
        The local variables for the first and last nodes in the fragment
        need to be reset when the fragment is changed for plaintext-only mode
        or a change from the beforetextinserted event handler.

        * editing/ReplaceSelectionCommand.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13737 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago A fix that makes coalesced updates work much better (and makes our
hyatt [Sat, 8 Apr 2006 00:16:38 +0000 (00:16 +0000)]
    A fix that makes coalesced updates work much better (and makes our
        single animated GIF timer work better).  The new rect painting algorithm
        for WebHTMLView will use the single unioned rect if the # of rects exceeds
        a threshold (10 is my initial cut), or if the union has enough "wasted"
        additional pixels (conservatively set at 75%).

        Reviewed by darin

        * WebView/WebHTMLView.m:
        (-[NSArray drawSingleRect:]):
        (-[NSArray drawRect:]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13736 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoLayoutTests:
mjs [Fri, 7 Apr 2006 23:55:19 +0000 (23:55 +0000)]

        Reviewed by Darin.

        Test case for this fix:

        - fixed REGRESSION: offsetParent on element with no offset parent crashes

        * fast/dom/Element/null-offset-parent.html: Added.


        Reviewed by Darin.

        - fixed REGRESSION: offsetParent on element with no offset parent crashes

        * dom/Element.cpp:
        (WebCore::Element::offsetParent): Add missing null check.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13735 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Darin.
mjs [Fri, 7 Apr 2006 23:50:17 +0000 (23:50 +0000)]
    Reviewed by Darin.

        - simplify archive loading

        * WebView/WebDataSource.m:
        (-[WebDataSource representation]): fix whitespace
        * WebView/WebHTMLRepresentation.m:
        (-[WebHTMLRepresentation loadArchive]): Use WebFrame's loadArchive: instead
        of trying to do a manual load here.
        (-[WebHTMLRepresentation documentSource]): No more special case for WebArchive.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13734 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Tim Omernick.
sullivan [Fri, 7 Apr 2006 23:01:57 +0000 (23:01 +0000)]
    Reviewed by Tim Omernick.

        - fixed <rdar://problem/4507054> If an icon file is removed from disk behind Safari's back,
        Safari will not try to refetch it

        In addition to letting WebIconDatabase recover from disk/memory mismatches as per 4507054, I also found
        a bug in the way icons are remembered that could account for some of the other cases where
        site icons didn't appear (and you'd get an ERROR on debug builds about WebIconDatabase saying it
        had some icon when it really doesn't).

        * Misc/WebIconDatabase.m:
        (-[WebIconDatabase iconForURL:withSize:cache:]):
        When the icon file for a previously-saved site icon isn't found, forget about the darn icon URL thoroughly
        so that this situation is self-correcting. Formerly, once you got into this state WebKit would never find
        the icon again (short of removing the entire icon database cleanly). Note that this does *not* change
        the behavior of sites that didn't have a site icon when WebKit checked -- these will continue to not
        return a site icon indefinitely (that's a separate, possibly performance-sensitive issue).
        (-[WebIconDatabase _retainIconForIconURLString:]):
        This code did the wrong thing in the case where an icon was in the process of being forgotten about. In
        that case, the icon would still be in _private->iconsOnDiskWithURLs, so we wouldn't try to save it here,
        so the new icon wouldn't stick.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13733 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoLayoutTests:
justing [Fri, 7 Apr 2006 22:25:43 +0000 (22:25 +0000)]

        Reviewed by adele

        * fast/forms/input-maxlength-2-expected.checksum: Added.
        * fast/forms/input-maxlength-2-expected.png: Added.
        * fast/forms/input-maxlength-2-expected.txt: Added.
        * fast/forms/input-maxlength-2.html: Added.


        Reviewed by adele

        maxlength truncation in text fields didn't work if the fragment was a
        single text node.

        * editing/ReplaceSelectionCommand.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13732 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoRolling out the patch I just checked in.
justing [Fri, 7 Apr 2006 21:57:55 +0000 (21:57 +0000)]
Rolling out the patch I just checked in.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13731 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoLayoutTests:
justing [Fri, 7 Apr 2006 20:46:44 +0000 (20:46 +0000)]

        Reviewed by harrison
        REGRESSION: Interchange newlines aren't passed with the khtmlBeforeTextInsertedEvent

        * fast/forms/8250-expected.checksum: Added.
        * fast/forms/8250-expected.png: Added.
        * fast/forms/8250-expected.txt: Added.
        * fast/forms/8250.html: Added.


        Reviewed by harrison

        REGRESSION: Interchange newlines aren't passed with the khtmlBeforeTextInsertedEvent

        * editing/ReplaceSelectionCommand.cpp:
        Interchange content removal happened before khtmlBeforeTextInsertedEvent was sent.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13730 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoLayoutTests:
justing [Fri, 7 Apr 2006 19:36:59 +0000 (19:36 +0000)]

        Reviewed by adele
        REGRESSION: Two extra newlines added when pasting a single styled line into a plaintext-only region

        Updated to relfect fix:
        * editing/pasteboard/8145-1-expected.checksum:
        * editing/pasteboard/8145-1-expected.png:
        * editing/pasteboard/8145-1-expected.txt:
        * editing/pasteboard/8145-1.html:


        Reviewed by adele

        REGRESSION: Two extra newlines added when pasting a single styled line into a plaintext-only region

        * editing/ReplaceSelectionCommand.cpp:
        Converting the fragment to plaintext introduced an extraneous newline because
        the range passed to plainText ended after the paragraph containing the fragment
        built from the markup that TextEdit put on the paste board.  TextIterator will
        emit a newline when it exits a paragraph.
        Two extra newlines were added because the '\n' in the plaintext string turns
        into an interchange newline, which isn't removed because of 8250, and the
        interchange newline looks like inline content that requires the insertion of
        a paragraph separator during paste.
        Fixed by creating a range using VisiblePositions at the start and the end of
        the node that holds the fragment during paste's test rendering.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13729 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoLayoutTests:
justing [Fri, 7 Apr 2006 18:41:26 +0000 (18:41 +0000)]

        Reviewed by harrison

        REGRESSION: Pasting text from TextEdit with a bold word into text field results in crash

        * editing/pasteboard/8145-1-expected.checksum: Added.
        * editing/pasteboard/8145-1-expected.png: Added.
        * editing/pasteboard/8145-1-expected.txt: Added.
        * editing/pasteboard/8145-1.html: Added.
        * editing/pasteboard/8145-2-expected.checksum: Added.
        * editing/pasteboard/8145-2-expected.png: Added.
        * editing/pasteboard/8145-2-expected.txt: Added.
        * editing/pasteboard/8145-2.html: Added.
        * editing/pasteboard/8145-3-expected.checksum: Added.
        * editing/pasteboard/8145-3-expected.png: Added.
        * editing/pasteboard/8145-3-expected.txt: Added.
        * editing/pasteboard/8145-3.html: Added.
        * fast/forms/plaintext-mode-1-expected.txt:
        * fast/forms/plaintext-mode-1.html:


        Reviewed by harrison

        REGRESSION: Pasting text from TextEdit with a bold word into text field results in crash

        * editing/AppendNodeCommand.cpp:
        Assert that the node will be placed somewhere that's contenteditable.
        * editing/InsertNodeBeforeCommand.cpp:
        (WebCore::InsertNodeBeforeCommand::doApply): Ditto
        * editing/JSEditor.cpp:
        Enabled insertHTML for plaintext-only regions since it's useful for debugging rich
        content pastes into the new text fields.
        * editing/ReplaceSelectionCommand.cpp:
        1) After the first paragraph of the fragment has been merged with the first part of the
        paragraph where the paste occured, if the next node to be inserted is inline, we put it
        in a new paragraph because it was at the start of a paragraph in the fragment.  The change
        is to insert a paragraph separator if insertionPos.next() is null or outside of the current
        editable region.
        2) Before the paste begins, a paragraph separator is inserted in order to avoid
        nesting blocks from the fragment to be pasted inside the block where the paste will
        occur.  I made two fixes to the code that decides whether or not to insert the
        paragraph separator and added testcases for each.  Added a fixme because it appears that
        this code is also used to ensure that the aforementioned insertionPos will be at the end of
        a paragraph.  This code should only be about preventing nesting.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13728 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoLayoutTests:
mjs [Fri, 7 Apr 2006 08:31:17 +0000 (08:31 +0000)]

        Reviewed by Anders.

        Added test case for incidental fix from this change:

        - Convert Element JS bindings to be almost completely autogenerated

        * fast/dom/Element/dimension-properties-unrendered-expected.txt: Added.
        * fast/dom/Element/dimension-properties-unrendered.html: Added.
        * fast/dom/Element/resources/dimension-properties-unrendered.js: Added.


        Reviewed by Anders.

        - Convert Element JS bindings to be almost completely autogenerated

        * dom/Element.idl: Declare full interface in IDL.
        * dom/Element.h:
        (WebCore::Element::tagQName): Renamed from tagName, so the real DOM
        method can be called tagName.
        (WebCore::Element::tagName): inline alias for nodeName.
        * dom/Element.cpp:
        (WebCore::Element::scrollByUnits): Moved logic from JS bindings to
        core DOM.
        (WebCore::Element::scrollByLines): ditto
        (WebCore::Element::scrollByPages): ditto
        (WebCore::Element::offsetLeft): ditto, plus make unrendered elements return 0 not undefined
        (WebCore::Element::offsetTop): ditto, plus make unrendered elements return 0 not undefined
        (WebCore::Element::offsetWidth): ditto, plus make unrendered elements return 0 not undefined
        (WebCore::Element::offsetHeight): ditto, plus make unrendered elements return 0 not undefined
        (WebCore::Element::offsetParent): ditto
        (WebCore::Element::clientWidth): ditto, plus make unrendered elements return 0 not undefined
        (WebCore::Element::clientHeight): ditto, plus make unrendered elements return 0 not undefined
        (WebCore::Element::scrollLeft): ditto
        (WebCore::Element::scrollTop): ditto
        (WebCore::Element::setScrollLeft): ditto
        (WebCore::Element::setScrollTop): ditto
        (WebCore::Element::scrollWidth): ditto, plus make unrendered elements return 0 not undefined
        (WebCore::Element::scrollHeight): ditto, plus make unrendered elements return 0 not undefined

        * bindings/scripts/CodeGeneratorJS.pm: Added support for
        ConvertUndefinedToTrue, for benefit of scroll methods where
        omitted arguments should be treated as true.  However, maybe
        explicit overloading in the IDL would be a better long-term
        approach for optional arguments.

        * khtml/ecma/kjs_binding.cpp:
        (KJS::valueToStringWithNullCheck): fixed formatting
        (KJS::valueToBooleanTreatUndefinedAsTrue): added for binding of scrollIntoView
        (isn't this lame?)
        * khtml/ecma/kjs_binding.h:
        (KJS::toJS): fixed formatting, added new stuff
        * khtml/ecma/kjs_dom.cpp:
        (KJS::DOMElement::getValueProperty): removed most of contents
        (KJS::DOMElement::putValueProperty): ditto
        (KJS::DOMElementProtoFunc::callAsFunction): ditto

        * css/cssstyleselector.cpp:
        (WebCore::CSSStyleSelector::canShareStyleWithElement): updated for rename
        of tagName to tagQName
        (WebCore::CSSStyleSelector::checkOneSelector): ditto
        * dom/Document.cpp:
        (WebCore::Document::importNode): ditto
        * editing/ApplyStyleCommand.cpp:
        (WebCore::ApplyStyleCommand::removeInlineStyle): ditto
        (WebCore::areIdenticalElements): ditto
        * html/HTMLElement.cpp:
        (WebCore::HTMLElement::inEitherTagList): ditto
        (WebCore::HTMLElement::inInlineTagList): ditto
        (WebCore::HTMLElement::inBlockTagList): ditto

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13726 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Re-enable coalesced updates. Our move to a single timer has essentially
hyatt [Fri, 7 Apr 2006 08:04:02 +0000 (08:04 +0000)]
    Re-enable coalesced updates.  Our move to a single timer has essentially
        implemented coalescing of updates anyway, so the regression has already
        come back.  Since we have to deal with that now anyway, there's no reason
        to visually tear also. :)

        Reviewed by darin

        * ChangeLog:
        * WebView/WebView.m:
        (-[WebView _commonInitializationWithFrameName:groupName:]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13725 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoAdd the ignore property for inside this project.
darin [Fri, 7 Apr 2006 04:52:32 +0000 (04:52 +0000)]
Add the ignore property for inside this project.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13724 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Darin.
bdakin [Thu, 6 Apr 2006 18:45:52 +0000 (18:45 +0000)]
    Reviewed by Darin.

        Fix for http://bugzilla.opendarwin.org/show_bug.cgi?id=8203
        REGRESSION: "Invite a friend" text field in GMail page spills out
        of table

        RenderReplaced::calcMinMaxWidth() sets m_minWidth to 0 when the
        width is a percent, so this patch copies that behavior into

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13723 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago - try to fix Windows build
darin [Thu, 6 Apr 2006 07:26:44 +0000 (07:26 +0000)]
    - try to fix Windows build

        * WebCore.vcproj/WebCore/WebCore.vcproj: Add JSDocument.cpp.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13722 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Changes requested by Mark Rowe.
darin [Thu, 6 Apr 2006 07:23:34 +0000 (07:23 +0000)]
    Changes requested by Mark Rowe.

        * BuildSlaveSupport/build-launcher-app: Set executable bit, removed property change
        trash at the end of the file.
        * BuildSlaveSupport/build-launcher-dmg: Ditto.

        * WebKitLauncher/main.m: Removed trash at end of file.

        * WebKitLauncher/WebKitLauncher.xcodeproj/project.pbxproj: Let Xcode update comments
        to make name match.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13721 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago - regenerated test results
darin [Thu, 6 Apr 2006 07:11:46 +0000 (07:11 +0000)]
    - regenerated test results

        * fast/forms/input-table-expected.txt: Regenerated. Seems right, matches Gecko.

        * fast/forms/form-element-geometry-expected.png: Regenerated.
        * fast/forms/form-element-geometry-expected.checksum: Regenerated.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13720 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago - attempt to fix Windows build
darin [Thu, 6 Apr 2006 07:02:00 +0000 (07:02 +0000)]
    - attempt to fix Windows build

        * kxmlcore/HashMap.h: (KXMLCore::HashMap::remove): Use (*it). instead of it->.
        * kxmlcore/HashSet.h: (KXMLCore::HashSet::remove): Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13719 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago - bug http://bugzilla.opendarwin.org/show_bug.cgi?id=8187 seems to have disap...
darin [Thu, 6 Apr 2006 06:41:49 +0000 (06:41 +0000)]
    - bug bugzilla.opendarwin.org/show_bug.cgi?id=8187 seems to have disappeared

        * fast/forms/input-truncate-newline-expected.txt: Restored good expected results.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13718 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago - attempt to fix Windows build
darin [Thu, 6 Apr 2006 06:07:42 +0000 (06:07 +0000)]
    - attempt to fix Windows build

        * os-win32/stdint.h: Add int8_t, uint8_t, int64_t.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13717 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago - fixed a storage leak from that last check-in
darin [Thu, 6 Apr 2006 06:01:20 +0000 (06:01 +0000)]
    - fixed a storage leak from that last check-in

        * WebView/WebUnarchivingState.m: (-[WebUnarchivingState dealloc]):
        Release the two dictionaries.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13716 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoLayoutTests:
darin [Thu, 6 Apr 2006 05:57:30 +0000 (05:57 +0000)]

        Reviewed by Adele.

        - test for http://bugzilla.opendarwin.org/show_bug.cgi?id=8111
          REGRESSION (NativeTextField): first click in form field on weather.com leaves focus
          but no caret

        - updated results for improved dumping of selection location in DumpRenderTree

        * fast/forms/input-text-self-emptying-click-expected.checksum: Added.
        * fast/forms/input-text-self-emptying-click-expected.png: Added.
        * fast/forms/input-text-self-emptying-click-expected.txt: Added.
        * fast/forms/input-text-self-emptying-click.html: Added.

        * editing/selection/3690703-2-expected.txt: Updated.
        * fast/forms/input-appearance-focus-expected.txt: Updated.
        * fast/forms/input-appearance-readonly-expected.txt: Updated.
        * fast/forms/input-appearance-selection-expected.txt: Updated.
        * fast/forms/input-paste-undo-expected.txt: Updated.
        * fast/forms/input-text-double-click-expected.txt: Updated.
        * fast/forms/input-text-option-delete-expected.txt: Updated.
        * fast/forms/input-text-scroll-left-on-blur-expected.txt: Updated.
        * fast/forms/plaintext-mode-2-expected.txt: Updated.


        Reviewed by Adele.

        - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=8111
          REGRESSION (NativeTextField): first click in form field on weather.com leaves focus
          but no caret

        Test: fast/forms/input-text-self-emptying-click.html

        * page/MouseEventWithHitTestResults.h: Removed url, target, m_url and m_target,
        replacing them with isOverLink and m_isOverLink. Replaced innerNode with targetNode,
        and added logic to handle the case where the target node is removed from the document
        but the element the target node was in is still inside the document.
        * page/MouseEventWithHitTestResults.cpp: Added.

        * WebCore.xcodeproj/project.pbxproj: Added MouseEventWithHitTestResults.cpp.
        * WebCore.vcproj/WebCore/WebCore.vcproj: Ditto.

        * kwq/RenderTreeAsText.cpp: (nodePosition): Added a call to shadowParentNode so we
        correctly dump positions within shadow trees. This was needed to give a good result
        from my new test for this bug.

        * dom/Document.cpp:
        (WebCore::Document::prepareMouseEvent): Update to pass fewer parameters to the
        constructor for MouseEventWithHitTestResults. Now takes isOverLink boolean, and no
        longer takes href and target parameters.

        * page/Frame.cpp:
        (WebCore::Frame::handleMousePressEventDoubleClick): Use new name targetNode, instead
        of old name innerNode.
        (WebCore::Frame::handleMousePressEventTripleClick): Ditto.
        (WebCore::Frame::handleMousePressEventSingleClick): Ditto. Also change code to check
        if over a link to use !isOverLink instead of url.isNull.
        (WebCore::Frame::handleMousePressEvent): Ditto. Also remove unused "url" local variable.
        (WebCore::Frame::handleMouseMoveEvent): Ditto.
        (WebCore::Frame::handleMouseReleaseEvent): Ditto.
        (WebCore::Frame::passWidgetMouseDownEventToWidget): Ditto.

        * bridge/mac/FrameMac.mm:
        (WebCore::FrameMac::handleMousePressEvent): Use new name targetNode, instead of old
        name innerNode.
        (WebCore::FrameMac::handleMouseMoveEvent): Ditto.
        (WebCore::FrameMac::passSubframeEventToSubframe): Ditto.
        (WebCore::FrameMac::sendContextMenuEvent): Ditto.

        * page/FrameView.cpp:
        (WebCore::subframeForEvent): Use new name targetNode, instead of old name innerNode.
        (WebCore::FrameView::handleMousePressEvent): Ditto.
        (WebCore::FrameView::handleMouseDoubleClickEvent): Ditto.
        (WebCore::selectCursor): Ditto. Also change code to check if over a link to use
        isOverLink instead of !url.isNull.
        (WebCore::FrameView::handleMouseMoveEvent): Ditto.
        (WebCore::FrameView::handleMouseReleaseEvent): Ditto.
        (WebCore::FrameView::updateDragAndDrop): Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13715 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Eric.
mjs [Thu, 6 Apr 2006 04:49:28 +0000 (04:49 +0000)]
    Reviewed by Eric.

        - start on factoring WebArchive unpacking more into a separate class.

        * WebKit.xcodeproj/project.pbxproj:
        * WebView/WebDataSource.m:
        (-[WebDataSourcePrivate dealloc]):
        (-[WebDataSource _addToUnarchiveState:]):
        (-[WebDataSource _popSubframeArchiveWithName:]):
        (-[WebDataSource _documentFragmentWithArchive:]):
        (-[WebDataSource _setupForReplaceByMIMEType:]):
        (-[WebDataSource subresourceForURL:]):
        * WebView/WebDataSourcePrivate.h:
        * WebView/WebFrame.m:
        (-[WebFrame _loadRequest:archive:]):
        (-[WebFrame loadRequest:]):
        (-[WebFrame loadArchive:]):
        * WebView/WebFramePrivate.h:
        * WebView/WebHTMLRepresentation.m:
        (-[WebHTMLRepresentation loadArchive]):
        * WebView/WebUnarchivingState.h: Added.
        * WebView/WebUnarchivingState.m: Added.
        (-[WebUnarchivingState init]):
        (-[WebUnarchivingState addArchive:]):
        (-[WebUnarchivingState archivedResourceForURL:]):
        (-[WebUnarchivingState popSubframeArchiveWithFrameName:]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13714 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoFixed date in ChangeLog entries for last checkin.
mjs [Thu, 6 Apr 2006 04:44:42 +0000 (04:44 +0000)]
Fixed date in ChangeLog entries for last checkin.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13713 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoLayoutTests:
mjs [Thu, 6 Apr 2006 04:40:07 +0000 (04:40 +0000)]

        Reviewed and landed by Maciej.

        - updated test result for http://bugzilla.opendarwin.org/show_bug.cgi?id=8184
          REGRESSION (r13655): Layer outline not clipped where it should be

        * fast/clip/outline-overflowClip-expected.txt: Reverted to the correct
        expected result.


        Reviewed and landed by Maciej.

        - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=8184
          REGRESSION (r13655): Layer outline not clipped where it should be

        * rendering/RenderLayer.cpp:
        (WebCore::RenderLayer::calculateRects): Reverted this part of the fix for bug 7943.
        The outlineRect should not be inflated.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13712 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoLayoutTests:
mjs [Thu, 6 Apr 2006 04:36:14 +0000 (04:36 +0000)]

        Reviewed by Eric, landed by Maciej.

        * svg/custom/class-selector.svg: Added.


        Reviewed by Eric, landed by Maciej.

        Fix for http://bugzilla.opendarwin.org/show_bug.cgi?id=7627:
        SVG from the W3C SVG 1.1 test suite (cubic01.svg) renders as all black

        Fix this svg stylesheet to use strict-mode, as this makes sure
        class selecting is done correctly.

        * ksvg2/svg/SVGStyleElement.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13711 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Darin, landed by Maciej.
mjs [Thu, 6 Apr 2006 04:14:24 +0000 (04:14 +0000)]
    Reviewed by Darin, landed by Maciej.

        * BuildSlaveSupport/build-launcher-app: Added.  Builds WebKit.app from WebKitLauncher
          and bundles the WebKit frameworks inside it.
        * BuildSlaveSupport/build-launcher-dmg: Added.  Builds, and optionally uploads, a disk image
          containing WebKit.app.
        * BuildSlaveSupport/run-performance-tests:  Use currentSVNRevision.
        * Scripts/webkitdirs.pm:  Add currentSVNRevision to retrieve the revision
          number of the SVN working copy.
        * WebKitLauncher: Added.
        * WebKitLauncher/Info.plist: Added.
        * WebKitLauncher/WebKitLauncher.xcodeproj: Added.
        * WebKitLauncher/WebKitLauncher.xcodeproj/project.pbxproj: Added.
        * WebKitLauncher/WebKitNightlyEnabler.m: Added.  This makes
          up the WebKitNightlyEnabler dylib which is loaded into Safari's address space
          to cause LaunchServices to treat Safari.app as WebKit.app when it is run from
          the nightly launcher.
        * WebKitLauncher/main.m: Added.  The WebKit.app launcher.
          It sets up the environment to have Safari use the bundled frameworks
          and load the WebKitNightlyEnabler dylib before exec'ing Safari
        * WebKitLauncher/start.html: Added.
        * WebKitLauncher/webkit.icns: Added.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13710 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Maciej.
darin [Thu, 6 Apr 2006 00:13:50 +0000 (00:13 +0000)]
    Reviewed by Maciej.

        - fix memory leak introduced by the previous change

        * kxmlcore/HashTable.h: Specialize NeedsRef so that it correctly returns true when
        the value in question is a pair where one of the pair needs a ref and the other
        of the pair does not.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13709 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by mjs
justing [Thu, 6 Apr 2006 00:00:24 +0000 (00:00 +0000)]
    Reviewed by mjs

        I forgot to convert one of the implicit remove/inserts to an explicit remove/insert,
        so we were hitting the assert I just added to InsertNodeBeforeCommand that checked for
        implicit removes.

        * editing/ReplaceSelectionCommand.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13708 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoShould fix the build
justing [Wed, 5 Apr 2006 22:43:40 +0000 (22:43 +0000)]
Should fix the build

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13707 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoLayoutTests:
mjs [Wed, 5 Apr 2006 22:26:43 +0000 (22:26 +0000)]

        Reviewed by Anders.

        Update prototype chain for the following fix:

        - autogenerate bindings for all of the methods and properties of Document

        * fast/dom/prototype-chain-expected.txt:


        Reviewed by Anders.

        - autogenerate bindings for all of the methods and properties of Document

        - also removed document.actualEncoding since it is not in any spec
        or implemented by any other browser

        * DerivedSources.make: add JSDocument.h to results
        * WebCore.xcodeproj/project.pbxproj: Added new files to project
        * bindings/scripts/CodeGeneratorJS.pm: Support for new stuff needed by Document.
        * dom/Document.cpp:
        (WebCore::Document::readyState): moved impl here from JS bindings
        (WebCore::Document::inputEncoding): ditto
        (WebCore::Document::defaultCharset): ditto
        (WebCore::Document::setCharset): ditto
        * dom/Document.h:
        (WebCore::Document::charset): added, synonym for inputEncoding.
        (WebCore::Document::characterSet): ditto
        * dom/Document.idl: Added. Full interface for the Document object.
        * khtml/ecma/JSXMLHttpRequest.cpp:
        (KJS::JSXMLHttpRequestProtoFunc::callAsFunction): JSDocument, not DOMDocument
        * khtml/ecma/JSXSLTProcessor.cpp:
        (KJS::XSLTProcessorProtoFunc::callAsFunction): ditto
        * khtml/ecma/kjs_binding.cpp:
        (KJS::jsStringOrFalse): Added this convenience for the method on Document
        that bizzarely returns false on failure and a string otherwise.
        * khtml/ecma/kjs_binding.h:
        * khtml/ecma/kjs_css.cpp:
        (KJS::toJS): renamed for consistency
        * khtml/ecma/kjs_css.h:
        * khtml/ecma/kjs_dom.cpp:
        - removed all traces of DOMDocument
        (KJS::toJS): JSDocument, not DOMDocument
        * khtml/ecma/kjs_dom.h:
        * khtml/ecma/kjs_html.cpp:
        (KJS::JSHTMLDocument::JSHTMLDocument): inherit from JSDocument
        (KJS::JSHTMLDocument::getOwnPropertySlot): ditto
        (KJS::JSHTMLDocument::put): ditto
        * khtml/ecma/kjs_html.h:
        * khtml/ecma/kjs_traversal.cpp:
        (KJS::toJS): added overloads
        (KJS::toNodeFilter): handle JS functions as well as impl NodeFilter objects
        * khtml/ecma/kjs_traversal.h:
        * platform/AtomicString.h:
        (WebCore::AtomicString::AtomicString): Allow implicit conversion from String.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13706 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoLayoutTests:
justing [Wed, 5 Apr 2006 21:53:23 +0000 (21:53 +0000)]

        Reviewed by darin

        Hitting an assert on undo paste

        * editing/pasteboard/undoable-fragment-removes-expected.checksum: Added.
        * editing/pasteboard/undoable-fragment-removes-expected.png: Added.
        * editing/pasteboard/undoable-fragment-removes-expected.txt: Added.
        * editing/pasteboard/undoable-fragment-removes.html: Added.


        Reviewed by darin

        Hitting an assert on undo paste

        ReplaceSelectionCommand was doing a combination of undoable and non-undoable
        removes from the ReplacementFragment.  On Undo Paste, the undoable removes
        couldn't be undone because the tree was in a different state than it was
        at the time of the remove.  This patch makes all the removes from the fragment
        non-undoable.  We could make them all undoable, but I can't think of any reason
        why we'd want the fragment to be reconstructed on an Undo Paste.

        * editing/AppendNodeCommand.cpp:
        Assert that the node to append isn't already in a tree, since if it is, it will
        be removed in a non-undoable way.
        * editing/InsertNodeBeforeCommand.cpp:
        (WebCore::InsertNodeBeforeCommand::doApply): Ditto.
        * editing/ReplaceSelectionCommand.cpp:
        Nodes were being moved from the fragment to the document with undoable inserts.
        Undoable inserts implicitly remove the node (in a non-undoable way) from its
        old location if it is already in a tree.  I now explicitly remove the nodes
        from the fragment before inserting them into the document to make it clear that
        they are being removed in a non-non-undoable way.  I also changed the one undoable
        remove from the fragment to a non-undoable remove.
        * editing/ReplaceSelectionCommand.h:
        Made ReplacementFragment's non-undoable removeNode public.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13705 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago - fixed the build
darin [Wed, 5 Apr 2006 21:51:11 +0000 (21:51 +0000)]
    - fixed the build

        * WebCore.xcodeproj/project.pbxproj: Oops! Resolved merge conflict.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13704 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoJavaScriptCore:
darin [Wed, 5 Apr 2006 21:19:57 +0000 (21:19 +0000)]

        Reviewed by Maciej.

        - JavaScriptCore part of fix for http://bugzilla.opendarwin.org/show_bug.cgi?id=8049
          StringImpl hash traits deleted value creates an init routine for WebCore
          <rdar://problem/4442248> REGRESSION: WebCore has init routines (8049)

        Change HashMap and HashSet implementation so they fold various types together.
        This allows us to implement maps and sets that use RefPtr<WebCore::StringImpl>
        and WebCore::String in terms of the underlying raw pointer type, and hence use
        -1 for the deleted value.

        * kxmlcore/HashTraits.h: Added a new type to HashTraits, StorageTraits, which is a
        type to be used when storing a value that has the same layout as the type itself.
        This is used only for non-key cases. In the case of keys, the hash function must also
        be considered. Moved emptyValue out of GenericHashTraitsBase into GenericHashTraits.
        Added a new bool to HashTraits, needsRef, which indicates whether the type needs
        explicit reference counting. If the type itself has needsRef true, but the storage
        type has needsRef false, then the HashSet or HashMap has to handle the reference
        counting explicitly. Added hash trait specializations for all signed integer values
        that give -1 as the deleted value. Gave all integers StorageTraits of the canonical
        integer type of the same size so int and long will share code. Gave all pointers and
        RefPtrs StorageTraits of the appropriately sized integer type. Removed redundant
        TraitType and emptyValue definitions in the pointer specialization for HashTraits.
        Added PairBaseHashTraits, which doesn't try to set up needsDestruction and deletedValue.
        Useful for types where we don't want to force the existence of deletedValue, such as
        the type of a pair in a HashMap which is not the actual storage type. Removed an
        unneeded parameter from the DeletedValueAssigner template. Added HashKeyStorageTraits
        template, which determines what type can be used to store a given hash key type with
        a given hash function, and specialized it for pointers and RefPtr so that pointer
        hash tables share an underlying HashTable that uses IntHash.

        * kxmlcore/HashTable.h: Added HashTableConstIteratorAdapter, HashTableIteratorAdapter,
        NeedsRef, RefCountManagerBase, RefCountManager, HashTableRefCountManagerBase, and
        HashTableRefCountManager. All are used by both HashSet and HashMap to handle hash
        tables where the type stored is not the same as the real value type.

        * kxmlcore/HashFunctions.h: Added a new struct named IntTypes that finds an
        integer type given a sizeof value. Renamed pointerHash to intHash and made it
        use overloading and take integer parameters. Added an IntHash struct which is
        a hash function that works for integers. Changed PtrHash to call IntHash with
        an appropriately sized integer. Made IntHash the default hash function for
        many integer types. Made PtrHash the default hash function for RefPtr as well
        as for raw pointers.

        * kxmlcore/HashSet.h: Changed implementation to use a separate "storage type"
        derived from the new traits. The HashTable will use the storage type and all
        necessary translation and ref/deref is done at the HashSet level. Also reorganized
        the file so that the HashSet is at the top and has no inline implementation inside
        it so it's easy to read the interface to HashSet.

        * kxmlcore/HashMap.h: Changed implementation to use a separate "storage type"
        derived from the new traits. The HashTable will use the storage type and all
        necessary translation and ref/deref is done at the HashMap level. Also reorganized
        the file so that the HashMap is at the top and has no inline implementation inside
        it so it's easy to read the interface to HashMap.

        * kxmlcore/HashMapPtrSpec.h: Removed. Superceded by optimizations in HashMap itself.

        * JavaScriptCore.xcodeproj/project.pbxproj: Remove HashMapPtrSpec.h, resort files,
        and also remove some unnecessary build settings from the aggregate target that
        generates derived sources.
        * JavaScriptCore.vcproj/JavaScriptCore/JavaScriptCore.vcproj: Ditto.


        Reviewed by Maciej.

        - fix for http://bugzilla.opendarwin.org/show_bug.cgi?id=8049
          StringImpl hash traits deleted value creates an init routine for WebCore
          <rdar://problem/4442248> REGRESSION: WebCore has init routines (8049)

        * platform/StringHash.h: Added. Moved hash functions and such for
        WebCore::String and friends into this file so we don't have to include
        the hash traits header everywhere. Changed hashing for WebCore::StringImpl
        and WebCore::String so that they use a raw pointer for the underlying
        storage type, taking advantage of the new feature added in JavaScriptCore.

        * platform/AtomicString.h: Moved StrHash specialization to StringHash.h.
        * platform/PlatformString.h: Moved StrHash specialization to StringHash.h.
        * platform/StringImpl.h: Moved StrHash, CaseInsensitiveHash, and HashTraits
        to StringHash.h. Left DefaultHash behind so that you can't get the wrong
        hash function by accident if you forget to include "StringHash.h".

        * platform/StringImpl.cpp: Added include of StringHash.h and removed
        RefPtr<StringImpl> HashTraits<RefPtr<StringImpl> >::_deleted, which is
        the object with a global initializer causing all the trouble!

        * kwq/AccessibilityObjectCache.h: Changed hash function to be IntHash
        instead of PtrHash.

        * dom/StyledElement.cpp: Changed MappedAttributeKeyTraits to inherit from
        the generic traits in KXMLCore so we get a StorageType. Also cleaned up a
        tiny bit by adding default values to the MappedAttributeKey constructor.

        * platform/CharsetNames.cpp: Changed hash traits here to be a new
        TextEncodingIDHashTraits struct rather than defining new default traits
        for the integer type since more integer types have default traits in
        HashTraits.h now. Also added a specialization so this class will share
        the underlying implementation (since InvalidEncoding happens to be -1).

        * bridge/mac/FrameMac.h:
        * dom/Document.h:
        * dom/xml_tokenizer.h:
        * khtml/xsl/XSLTProcessor.h:
        * kwq/JavaAppletWidget.h:
        * page/FramePrivate.h:
        * page/Page.cpp:
        * platform/AtomicString.cpp:
        * platform/TransferJob.h:
        * rendering/render_applet.h:
        Added include of StringHash.h.

        * WebCore.xcodeproj/project.pbxproj: Added StringHash.h. Remove unneeded
        CREATE_HASH_TABLE variable in build settings. Re-sorted some file lists.
        Added quotes to the CREATE_HASH_TABLE initialization in the rule that
        builds generated files. Removed various unneeded build settings for that
        target as well.

        * ForwardingHeaders/kxmlcore/HashTraits.h: Added.

        - other minor cleanup

        * bridge/mac/FrameMac.mm: Sorted includes.
        * dom/Node.cpp: Removed bogus symbol after #endif.

        * khtml/xsl/XSLTProcessor.cpp: Sorted includes. Removed redundant using
        namespace WebCore.
        * loader/Cache.cpp: Ditto.


        Reviewed by Maciej.

        * Scripts/check-for-global-initializers: Remove StringImpl from the list of files that
        are allowed to have global initializers.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13703 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Darin.
bdakin [Wed, 5 Apr 2006 18:05:12 +0000 (18:05 +0000)]
    Reviewed by Darin.

        Fix for <rdar://problem/4502311> text-transform:capitalize needs to
        treat nbsp as a regular space when ICU changes

        There will be future changes in ICU to match the Unicode 4.1
        standard which no longer recognizes &nbsp as a word separator. We
        need to work around this with text-transform:capitalize because
        words after non-breaking spaces still need to be capitalized.

        No layout tests added because existing layout tests cover this.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13702 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Darin.
ap [Wed, 5 Apr 2006 18:02:19 +0000 (18:02 +0000)]
    Reviewed by Darin.

        - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=8110
          Define navigator.vendorSub (bcms.gov.uk doesn't allow access to login page)

        Test: fast/dom/navigator-vendorSub.html

        * khtml/ecma/kjs_navigator.cpp:
        (KJS::Navigator::getValueProperty): Return an empty string for vendorSub property.
        * khtml/ecma/kjs_navigator.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13701 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by OMG BETH
ggaren [Wed, 5 Apr 2006 18:02:07 +0000 (18:02 +0000)]
    Reviewed by OMG BETH

        * Scripts/run-testkjs:
        - pipe STDERR to /dev/null by default; new --verbose option overrides
        this behavior
        - set DYLD_FRAMEWORK_PATH to the webkit-configured path
        - output run command in a format that can be copied and pasted into the
        terminal to run manually

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13700 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoLayoutTests:
darin [Wed, 5 Apr 2006 03:55:40 +0000 (03:55 +0000)]

        Reviewed by Justin and Adele.

        * fast/forms/plaintext-mode-2-expected.txt: Updated because there's no longer
        and extra <div> element added.


        Reviewed by Justin (editing parts) and Adele (the rest).

        - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=8182
          some text-field-related layout tests are failing

        The smart paste code was getting confused and adding extra spaces.

        * editing/ReplaceSelectionCommand.cpp:
        (WebCore::ReplaceSelectionCommand::doApply): Use isStartOfParagraph instead of
        (WebCore::ReplaceSelectionCommand::removeLinePlaceholderIfNeeded): Ditto.

        * editing/InsertParagraphSeparatorCommand.cpp: (WebCore::enclosingEmptyListItem):
        Change to call isStart/EndOfParagraph instead of Line.

        * editing/InsertTextCommand.cpp: (WebCore::InsertTextCommand::input):
        Add a comment about how isStartOfLine is almost certainly wrong here.

        - clean up some loose ends in the Frame class from the recent renaming

        * page/Frame.h: Removed declarations of deleteMe1, deleteMe2, and
        * page/Frame.cpp: (WebCore::Frame::handleMouseMoveEvent): Removed
        handleMouseMoveEventPart2 by renaming it to handleMouseMoveEvent and removing
        handleMouseMoveEvent itself.

        - invoke the makefile directly, removing the generate-derived-sources script

        * WebCore.vcproj/WebCore/build-generated-files.sh: Call make directly.
        * WebCore.xcodeproj/project.pbxproj: Ditto.
        * generate-derived-sources: Removed.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13699 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoUpdate table results after my change to create fewer RenderTexts.
hyatt [Wed, 5 Apr 2006 02:26:51 +0000 (02:26 +0000)]
Update table results after my change to create fewer RenderTexts.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13698 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoUpdate fast results after my change to create fewer RenderTexts.
hyatt [Wed, 5 Apr 2006 02:19:22 +0000 (02:19 +0000)]
Update fast results after my change to create fewer RenderTexts.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13697 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoLayoutTests:
adele [Wed, 5 Apr 2006 02:06:51 +0000 (02:06 +0000)]

        - Tests for http://bugzilla.opendarwin.org/show_bug.cgi?id=8158
        REGRESSION: Clicking past RTL text in a new text field puts the caret on the wrong side of the text

        * editing/selection/caret-rtl-expected.checksum: Added.
        * editing/selection/caret-rtl-expected.png: Added.
        * editing/selection/caret-rtl-expected.txt: Added.
        * editing/selection/caret-rtl.html: Added.
        * editing/selection/caret-rtl-2-expected.checksum: Added.
        * editing/selection/caret-rtl-2-expected.png: Added.
        * editing/selection/caret-rtl-2-expected.txt: Added.
        * editing/selection/caret-rtl-2.html: Added.


        Reviewed by Justin.

        - Fix for http://bugzilla.opendarwin.org/show_bug.cgi?id=8158
        REGRESSION: Clicking past RTL text in a new text field puts the caret on the wrong side of the text


        * rendering/RenderText.cpp: (WebCore::RenderText::positionForCoordinates):
        When calculating the position for the beginning or end of an InlineTextBox,
        we now use offsetForPosition instead of just using m_start and m_len, because
        offsetForPosition will take rtl text into account.  I also made some formatting

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13696 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago New results for
adele [Wed, 5 Apr 2006 01:49:30 +0000 (01:49 +0000)]
    New results for
        * fast/forms/input-truncate-newline-expected.txt

        Filed bug:
        input-truncate-newline layout test fails when run as part of larger set of tests

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13695 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Fix for bug 8065, inline blocks incorrectly loses spaces between them.
hyatt [Tue, 4 Apr 2006 23:34:04 +0000 (23:34 +0000)]
    Fix for bug 8065, inline blocks incorrectly loses spaces between them.

        Reviewed by beth

        * dom/Text.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13689 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoLayoutTests:
adele [Tue, 4 Apr 2006 22:23:01 +0000 (22:23 +0000)]

         - Test for:
        REGRESSION (NativeTextField): table contents misaligned in Netflix queue

        REGRESSION: Native text field fails to wrap inside table

        REGRESSION: text fields at connect.apple.com spill out of the containing box

        * fast/forms/input-table-expected.checksum: Added.
        * fast/forms/input-table-expected.png: Added.
        * fast/forms/input-table-expected.txt: Added.
        * fast/forms/input-table.html: Added.


        Reviewed by Hyatt.

        - Fix for:
        REGRESSION (NativeTextField): table contents misaligned in Netflix queue

        REGRESSION: Native text field fails to wrap inside table

        REGRESSION: text fields at connect.apple.com spill out of the containing box

        Test: fast/forms/input-table.html

        Rewrote calcMinMaxWidth for text fields so it considers width,
        min-width, and max-width settings as well as the size attribute.

        * rendering/RenderTextField.cpp: (WebCore::RenderTextField::calcMinMaxWidth):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13688 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Darin.
bdakin [Tue, 4 Apr 2006 22:11:17 +0000 (22:11 +0000)]
    Reviewed by Darin.

        This is a followup to my fix for <rdar://problem/4493218>

        This patch re-names computeIntLength() and computerShortLength() to
        be computeLengthInt() and computeLengthShort(), respectively, to
        match the pre-existing computeLengthFloat(). This patch also adds
        the slightly confusing-ly named computeLengthIntForLength() which
        uses the max and min values of a 28-bit integer as bounds for
        overflow. This function is necessary because Length objects expect
        28-bit integers.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13687 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoJavaScriptCore:
thatcher [Tue, 4 Apr 2006 21:21:35 +0000 (21:21 +0000)]

        Reviewed by Darin.

        The Debug and Release frameworks are now built with install paths relative to the build products directory.
        This removes the need for other projects to build with -framework WebCore and -framework JavaScriptCore.

        * JavaScriptCore.xcodeproj/project.pbxproj:


        Reviewed by Darin.

        The Debug and Release frameworks are now built with install paths relative to the build products directory.
        This removes the need for other projects to build with -framework WebCore and -framework JavaScriptCore.

        * JavaScriptGlue.xcodeproj/project.pbxproj:


        Reviewed by Darin.

        The Debug and Release frameworks are now built with install paths relative to the build products directory.
        This removes the need for other projects to build with -framework WebCore and -framework JavaScriptCore.

        * WebCore.xcodeproj/project.pbxproj:


        Reviewed by Darin.

        The Debug and Release frameworks are now built with install paths relative to the build products directory.
        This removes the need for other projects to build with -framework WebCore and -framework JavaScriptCore.

        * WebKit.xcodeproj/project.pbxproj:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13686 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoLayoutTests:
justing [Tue, 4 Apr 2006 20:52:10 +0000 (20:52 +0000)]

        Reviewed by darin

        REGRESSION: Line disappears when deleting

        * editing/deleting/merge-different-styles-expected.checksum: Added.
        * editing/deleting/merge-different-styles-expected.png: Added.
        * editing/deleting/merge-different-styles-expected.txt: Added.
        * editing/deleting/merge-different-styles.html: Added.
        * editing/deleting/merge-no-br-expected.checksum: Added.
        * editing/deleting/merge-no-br-expected.png: Added.
        * editing/deleting/merge-no-br-expected.txt: Added.
        * editing/deleting/merge-no-br.html: Added.
        * editing/deleting/merge-whitespace-pre-expected.checksum: Added.
        * editing/deleting/merge-whitespace-pre-expected.png: Added.
        * editing/deleting/merge-whitespace-pre-expected.txt: Added.
        * editing/deleting/merge-whitespace-pre.html: Added.

        Fixes (not enough style on nodes for the fixes to be reflected in pixel results):
        * editing/deleting/delete-block-merge-contents-005-expected.txt:
        * editing/deleting/delete-block-merge-contents-006-expected.txt:
        * editing/deleting/delete-block-merge-contents-008-expected.txt:

        Equivalent render trees:
        * editing/deleting/delete-3857753-fix-expected.txt:
        * editing/inserting/insert-div-026-expected.txt:

        Forgot to checkin these new expected results after fixing the DRT bug:
        * fast/lists/drag-into-marker-expected.checksum:
        * fast/lists/drag-into-marker-expected.png:
        * fast/lists/drag-into-marker-expected.txt:


        Reviewed by darin

        REGRESSION: Line disappears when deleting

        Rewrote moveNodesAfterNode to address these problems:
        It moved nodes without preserving their style.
        It traversed over siblings looking for a br to know when
        to stop merging.  If the br was burried inside a span, it
        wouldn't find it.  If the text is whitespace:pre, it wouldn't
        In theory it would crash if the "enclosingInlineElements" of the start of the
        selection to delete and the end of the selection to delete were the
        same.  We think that this will fix these:
        CrashTracer: 2116 crashes in Mail at com.apple.WebCore: khtml::CompositeEditCommand::insertNodeAfter + 32
        CrashTracer: 1569 crashes in Mail at com.apple.WebCore: khtml::DeleteSelectionCommand::moveNodesAfterNode + 340
        But we haven't been able to construct a reproducible case.

        * editing/CompositeEditCommand.cpp:
        (WebCore::CompositeEditCommand::removeNodeAndPruneAncestors): Moved from ReplaceSelectionCommand.
        (WebCore::CompositeEditCommand::prune): Ditto.
        * editing/CompositeEditCommand.h:
        * editing/DeleteSelectionCommand.cpp:
        * editing/DeleteSelectionCommand.h:
        * editing/ReplaceSelectionCommand.cpp:
        * editing/ReplaceSelectionCommand.h:
        * editing/markup.cpp:
        Was crashing when passed a collapsed range.  I early return an empty string instead.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13685 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago2006-04-04 Eric Seidel <eseidel@apple.com>
eseidel [Tue, 4 Apr 2006 18:10:12 +0000 (18:10 +0000)]
2006-04-04  Eric Seidel  <eseidel@apple.com>

        Reviewed by ggaren.

        Fix win32 build.
        Disable ASSERT redefinition warnings for now.

        * JavaScriptCore.vcproj/testkjs/testkjs.vcproj:
        * kxmlcore/Assertions.h:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13684 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Adele Peterson.
sullivan [Tue, 4 Apr 2006 17:48:02 +0000 (17:48 +0000)]
    Reviewed by Adele Peterson.

        - WebKit part of <rdar://problem/4498418> "Autosaved" searchterms are saved during private browsing

        * WebView/WebView.m:
        (-[WebView _updateWebCoreSettingsFromPreferences:]):
        Pass private browsing setting down to WebCore.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13683 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Adele Peterson.
sullivan [Tue, 4 Apr 2006 17:47:49 +0000 (17:47 +0000)]
    Reviewed by Adele Peterson.

        - WebCore part of <rdar://problem/4498418> "Autosaved" searchterms are saved during private browsing

        * bridge/mac/WebCoreSettings.h:
        * bridge/mac/WebCoreSettings.mm:
        (-[WebCoreSettings setPrivateBrowsingEnabled:]):
        (-[WebCoreSettings privateBrowsingEnabled]):
        Teach WebCoreSettings about private browsing (WebKit knew, but WebCore didn't)

        * kwq/KWQKHTMLSettings.h:
        Teach KHTMLSettings about private browsing

        * kwq/KWQLineEdit.mm:
        Fix wrong class in a category method declaration; the compiler didn't seem to mind.

        * kwq/WebCoreTextField.mm:
        (-[KWQSearchFieldCell _addStringToRecentSearches:]):
        Override this method to bail out if private browsing is enabled.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13682 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Hyatt.
darin [Tue, 4 Apr 2006 16:15:07 +0000 (16:15 +0000)]
    Reviewed by Hyatt.

        REGRESSION: Safari crashes when printing a google map w/directions

        Tests: none, because I believe it only happens when printing, due to the relayouts

        * rendering/RenderTable.cpp:
        (WebCore::RenderTable::recalcSectionsIfNeeded): Add new utility to let the cells
        ensure the sections' grid data is up to date.
        * rendering/RenderTable.h:
        * rendering/RenderTableCell.cpp:
        (WebCore::RenderTableCell::calcMinMaxWidth):  Call above method.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13681 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago2006-04-04 Bjrn Graf <bjoern.graf@gmail.com>
eseidel [Tue, 4 Apr 2006 12:11:08 +0000 (12:11 +0000)]
2006-04-04  Bjrn Graf  <bjoern.graf@gmail.com>

        Reviewed by ggaren & darin.  Landed by eseidel.

        Integrate CURL version of gettimeofday
        Disable crash report dialogs for testkjs.exe in Release mode

        * kjs/testkjs.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13680 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago2006-04-03 Justin Haygood <jhaygood@spsu.edu>
eseidel [Tue, 4 Apr 2006 11:54:46 +0000 (11:54 +0000)]
2006-04-03  Justin Haygood <jhaygood@spsu.edu>

        Reviewed by eseidel.  Landed by eseidel.

        - WIN32: maximumScroll() is the maximum scroll delta, not the maximum scroll position. Update to use
          the real maximum scroll position.

        * platform/win/ScrollViewWin.cpp:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13679 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago2006-04-04 Eric Seidel <eseidel@apple.com>
eseidel [Tue, 4 Apr 2006 11:48:37 +0000 (11:48 +0000)]
2006-04-04  Eric Seidel  <eseidel@apple.com>

        Reviewed by andersca.

        Work-around spaces-in-pathnames issue in gnumake on win32.

        * WebCore.vcproj/WebCore/build-generated-files.sh:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13678 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago2006-04-04 Eric Seidel <eseidel@apple.com>
eseidel [Tue, 4 Apr 2006 10:08:50 +0000 (10:08 +0000)]
2006-04-04  Eric Seidel  <eseidel@apple.com>

        Reviewed by mjs.

        * kjs/number_object.cpp:
        (NumberProtoFunc::callAsFunction): remove trunc() to fix win32.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13677 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Darin.
mjs [Tue, 4 Apr 2006 09:03:26 +0000 (09:03 +0000)]
    Reviewed by Darin.

        - fixed "toPrecision sometimes messes up the last digit on intel Macs"

        * kjs/number_object.cpp:
        (intPow10): Compute integer powers of 10 using exponentiation by squaring.
        (NumberProtoFunc::callAsFunction): Use intPow10(n) in place of all pow(10.0, n),
        plus a bit of refactoring.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13676 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago - tried to fix build again
darin [Tue, 4 Apr 2006 06:54:07 +0000 (06:54 +0000)]
    - tried to fix build again

        * WebCore.xcodeproj/project.pbxproj: Removed JSStyleSheet files.
        * WebCore.vcproj/WebCore/WebCore.vcproj: Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13675 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago - fixed properties on a bunch of files
darin [Tue, 4 Apr 2006 06:30:13 +0000 (06:30 +0000)]
    - fixed properties on a bunch of files
          (removed allow-tabs and svn:executable from many)

        * css/css_base.cpp:
        * html/html_headimpl.cpp:
        * khtml/ecma/kjs_traversal.cpp:
        * kwq/DeprecatedPtrListImpl.cpp:
        * kwq/DeprecatedValueListImpl.cpp:
        * loader/CachedScript.h:
        * platform/ArrayImpl.cpp:
        * platform/StringImpl.cpp:
        * rendering/DataRef.h:
        * rendering/RenderContainer.cpp:
        * rendering/RenderTableCell.cpp:
        * rendering/bidi.h:
        * rendering/render_list.cpp:
        * rendering/render_style.cpp:
        * rendering/table_layout.h:
        Converted tabs to spaces.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13674 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Darin.
ap [Tue, 4 Apr 2006 05:21:08 +0000 (05:21 +0000)]
    Reviewed by Darin.

        - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=7118
          Property values with extra items do not get treated as invalid (they should)

        Tests: fast/css/invalidation-errors.html

        * css/CSSGrammar.y: Rollback the properties added by parseValue() when it returns false.
        * css/cssparser.h: Moved shorthand counting to ShorthandScope, a new class in cssparser.cpp.
        * css/cssparser.cpp:
        (WebCore::CSSParser::rollbackLastProperties): Added.
        (WebCore::CSSParser::parseValue): Return false if there are too many properties in the list.
        (WebCore::CSSParser::parseBackgroundShorthand): Use ShorthandScope.
        (WebCore::CSSParser::parseShorthand): Ditto.
        (WebCore::CSSParser::parse4Values): Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13673 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoBumping the PLT count back to 5.
thatcher [Tue, 4 Apr 2006 04:47:53 +0000 (04:47 +0000)]
Bumping the PLT count back to 5.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13672 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago - changed StyleSheet back to hand-generated since the generated toJS
darin [Tue, 4 Apr 2006 04:07:26 +0000 (04:07 +0000)]
    - changed StyleSheet back to hand-generated since the generated toJS
          function was not making the right type of wrapper for CSS style sheets
          (fixes failing layout tests)

        * DerivedSources.make: Removed JSStyleSheet.h.
        * css/StyleSheet.idl: Removed.
        * khtml/ecma/kjs_css.cpp: Added DOMStyleSheet back in.
        * khtml/ecma/kjs_css.h: Ditto.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13671 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago - tweak config.h and Platform.h to try to get buildbot working
darin [Tue, 4 Apr 2006 03:55:09 +0000 (03:55 +0000)]
    - tweak config.h and Platform.h to try to get buildbot working
          (making some small changes at the same time)

        * kjs/config.h: Removed now-unneeded HAVE_ICU.
        * kxmlcore/Platform.h: Tweak how platform gets set up. Move all the
        USE stuff to the end.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13670 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago - fixed Macintosh build
darin [Tue, 4 Apr 2006 03:34:33 +0000 (03:34 +0000)]
    - fixed Macintosh build

        * WebCore.xcodeproj/project.pbxproj: Fixed paths of some files that were absolute
        paths from my machine.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13669 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago - try to fix Windows build
darin [Tue, 4 Apr 2006 03:14:15 +0000 (03:14 +0000)]
    - try to fix Windows build

        * WebCore.vcproj/WebCore/WebCore.vcproj: Add three new generated files as source files.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13668 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoReviewed by Maciej.
staikos [Tue, 4 Apr 2006 02:45:19 +0000 (02:45 +0000)]
Reviewed by Maciej.

Fix Win32 build breakage and remove an unused forward.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13667 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoLayoutTests:
darin [Tue, 4 Apr 2006 01:54:40 +0000 (01:54 +0000)]

        - add a test of RTL to the test for bug 8106

        * fast/forms/input-text-scroll-left-on-blur-expected.png: Updated.
        * fast/forms/input-text-scroll-left-on-blur-expected.txt: Updated.
        * fast/forms/input-text-scroll-left-on-blur-expected.checksum: Updated.
        * fast/forms/input-text-scroll-left-on-blur.html: Updated.


        Reviewed by Maciej.

        - get RTL right for bug http://bugzilla.opendarwin.org/show_bug.cgi?id=8106
          REGRESSION (NativeTextField): New text fields don't scroll to the beginning when losing focus

        * rendering/RenderTextField.cpp: (WebCore::RenderTextField::forwardEvent):
        Scroll to the right if RTL.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13666 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Maciej.
darin [Tue, 4 Apr 2006 01:50:51 +0000 (01:50 +0000)]
    Reviewed by Maciej.

        - http://bugzilla.opendarwin.org/show_bug.cgi?id=8147
          convert derived sources script to a Makefile

        * DerivedSources.make: Added.
        * css/CSSPrimitiveValue.idl: Added.
        * css/Counter.idl: Added.
        * css/StyleSheet.idl: Added.

        * WebCore.xcodeproj/project.pbxproj: Added new generated files, IDLs, and the makefile.
        * bindings/scripts/CodeGeneratorJS.pm:

        * generate-derived-sources: Added license header. Removed most of the script, except for
        a single invocation of make.

        * khtml/ecma/kjs_css.h: Removed DOMStyleSheet, DOMCSSPrimitiveValue,
        CSSPrimitiveValueConstructor, and DOMCounter.
        * khtml/ecma/kjs_css.cpp:
        (KJS::DOMCSSStyleSheet::DOMCSSStyleSheet): Changed to use JSStyleSheet as the base class.
        (KJS::DOMCSSStyleSheet::getOwnPropertySlot): Ditto.
        (KJS::DOMCSSValueProtoFunc::callAsFunction): Added.
        (KJS::toJS): Changed to use JSCSSPrimitiveValue.
        (KJS::DOMRGBColor::getValueProperty): Changed to call toJS instead of making a
        DOMCSSPrimitiveValue directly.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13665 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years agoLayoutTests:
justing [Mon, 3 Apr 2006 23:58:43 +0000 (23:58 +0000)]

        Reviewed by harrison

        REGRESSION (NativeTextField): Drag and drop text within a text input field modifies page
        A drag and drop in DumpRenderTree copies the source, instead of cutting it

        * editing/pasteboard/drag-drop-modifies-page-expected.checksum: Added.
        * editing/pasteboard/drag-drop-modifies-page-expected.png: Added.
        * editing/pasteboard/drag-drop-modifies-page-expected.txt: Added.
        * editing/pasteboard/drag-drop-modifies-page.html: Added.

        * fast/lists/drag-onto-marker.html:
        * fast/lists/drag-onto-marker-expected.txt:
        * fast/lists/drag-onto-marker-expected.png:
        * fast/lists/drag-onto-marker-expected.checksum:


        Reviewed by harrison

        REGRESSION (NativeTextField): Drag and drop text within a text input field modifies page

        The frame's selection is only set after all sub-commands have been
        performed.  When we send the khtmlBeforeTextInsertedEvent to the root
        editable element we were using frame->selection(), which may no longer
        be in the document.

        Had to move the construction of the ReplacementFragment to when the
        replace operation is applied, because endingSelection isn't the
        endingSelection of the last operation when the replace operation
        is constructed.

        * editing/ReplaceSelectionCommand.cpp:
        * editing/ReplaceSelectionCommand.h:


        Reviewed by harrison

        A drag and drop in DumpRenderTree copies the source, instead of cutting it

        Tell the source that the drag is over after the drag is performed, not before.

        * DumpRenderTree/EventSendingController.m:
        (-[EventSendingController mouseUp]):

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13664 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Maciej.
staikos [Mon, 3 Apr 2006 23:49:39 +0000 (23:49 +0000)]
    Reviewed by Maciej.

        Implement a unicode abstraction layer to make JavaScriptCore much more
        easily ported to other platforms without having to take in libicu.  Also
        makes the unicode related code easier to understand.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13663 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Layout test for <rdar://problem/4495644> crash when mousing over
bdakin [Mon, 3 Apr 2006 23:46:21 +0000 (23:46 +0000)]
    Layout test for <rdar://problem/4495644> crash when mousing over
        links at nationalrealestateinvestors.com in

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13662 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Tim Hatcher.
sullivan [Mon, 3 Apr 2006 23:17:17 +0000 (23:17 +0000)]
    Reviewed by Tim Hatcher.

        - re-fixed <rdar://problem/4481198> REGRESSION (TOT): search results list in bookmarks view
        remains when search text is deleted

        * WebView/WebView.m:
        (-[WebView searchFor:direction:caseSensitive:wrap:]):
        I fixed this recently, but then broke it again by adding an early bail-out to this method. So
        now I'm removing the early bail-out.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13661 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Hyatt.
bdakin [Mon, 3 Apr 2006 22:30:07 +0000 (22:30 +0000)]
    Reviewed by Hyatt.

        Fix for <rdar://problem/4495644> crash when mousing over links at
        nationalrealestateinvestors.com in

        This is a fix for a repro crasher where a rootLineBox had a stale
        pointer to a render object.

        * rendering/RenderFlow.cpp:
        (WebCore::RenderFlow::dirtyLinesFromChangedChild): Only break from
        the function because of selfNeedsLayout() if we are not an inline
        flow, because if we are, we will not re-layout before bad things
        can happen.

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13660 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Maciej.
thatcher [Mon, 3 Apr 2006 22:26:18 +0000 (22:26 +0000)]
    Reviewed by Maciej.

        Removing idl files and some scripts from the WebCore target to prevent
        them from being copied into the Resources.

        * WebCore.xcodeproj/project.pbxproj:

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13659 268f45cc-cd09-0410-ab3c-d52691b4dbfc

16 years ago Reviewed by Adele.
thatcher [Mon, 3 Apr 2006 22:06:06 +0000 (22:06 +0000)]
    Reviewed by Adele.

        Fixes <rdar://problem/4498338> JavaScriptCore fails to compile for ppc64
        Other 64 bit build fixes.

        * kjs/collector.cpp:
        (KJS::Collector::markOtherThreadConservatively): test for __DARWIN_UNIX03 and use __r1
        * kjs/dtoa.cpp:
        (Bigint::): cast PRIVATE_mem to unsigned to prevent warning
        * bindings/jni/jni_utility.cpp:
        (KJS::Bindings::getJavaVM): cast jniError to long to prevent format warning
        (KJS::Bindings::getJNIEnv): cast jniError to long to prevent format warning
        * bindings/runtime_root.cpp:
        (KJS::Bindings::addNativeReference): cast CFDictionaryGetValue to unsigned long to prevent warning
        (KJS::Bindings::removeNativeReference): cast CFDictionaryGetValue to unsigned long to prevent warning

git-svn-id: https://svn.webkit.org/repository/webkit/trunk@13658 268f45cc-cd09-0410-ab3c-d52691b4dbfc