Fix crash when text is contained inside a table-colgroup.

[WebKit-https.git] / WebCore / ChangeLog-2005-08-23

2004-08-19  David Hyatt  <hyatt@apple.com>

        Fix crash when text is contained inside a table-colgroup.
        
        Reviewed by john

        * khtml/xml/dom_textimpl.cpp:
        (TextImpl::rendererIsNeeded):

2004-08-19  Trey Matteson  <trey@apple.com>

        Unexpected errors hit while finding word boundaries, leading to crash.

        Reviewed by Ken.

        * kwq/KWQTextUtilities.cpp:
        (KWQFindWordBoundary):  Don't call UCFindTextBreak with edge cases it thinks
        are param errors, and pass correct mask for forward case.  Also fix off-by-one
        crashers in fallback code.

2004-08-19  David Hyatt  <hyatt@apple.com>

        Make XSLT imports/includes work.  This code has to be turned off until the newer version of libxslt is
        available.
        
        Reviewed by kocienda

        * khtml/css/css_ruleimpl.cpp:
        (CSSImportRuleImpl::init):
        * khtml/xml/dom_docimpl.cpp:
        (DocumentImpl::DocumentImpl):
        (DocumentImpl::~DocumentImpl):
        (DocumentImpl::applyXSLTransform):
        * khtml/xml/dom_docimpl.h:
        (DOM::DocumentImpl::setTransformSource):
        (DOM::DocumentImpl::transformSource):
        * khtml/xml/dom_xmlimpl.cpp:
        (DOM::ProcessingInstructionImpl::checkStyleSheet):
        * khtml/xml/xml_tokenizer.cpp:
        (khtml::matchFunc):
        (khtml::openFunc):
        (khtml::readFunc):
        (khtml::writeFunc):
        (khtml::createQStringParser):
        (khtml::XMLTokenizer::setTransformSource):
        * khtml/xsl/xsl_stylesheetimpl.cpp:
        (DOM::XSLStyleSheetImpl::XSLStyleSheetImpl):
        (DOM::XSLStyleSheetImpl::~XSLStyleSheetImpl):
        (DOM::XSLStyleSheetImpl::isLoading):
        (DOM::XSLStyleSheetImpl::clearDocuments):
        (DOM::XSLStyleSheetImpl::parseString):
        (DOM::XSLStyleSheetImpl::loadChildSheets):
        (DOM::XSLStyleSheetImpl::loadChildSheet):
        (DOM::XSLImportRuleImpl::parentStyleSheet):
        (DOM::XSLStyleSheetImpl::compileStyleSheet):
        (DOM::XSLStyleSheetImpl::locateStylesheetSubResource):
        (DOM::XSLImportRuleImpl::XSLImportRuleImpl):
        (DOM::XSLImportRuleImpl::~XSLImportRuleImpl):
        (DOM::XSLImportRuleImpl::setStyleSheet):
        (DOM::XSLImportRuleImpl::isLoading):
        (DOM::XSLImportRuleImpl::loadSheet):
        * khtml/xsl/xsl_stylesheetimpl.h:
        (DOM::XSLStyleSheetImpl::setOwnerDocument):
        (DOM::XSLStyleSheetImpl::setDocument):
        (DOM::XSLStyleSheetImpl::markAsProcessed):
        (DOM::XSLStyleSheetImpl::processed):
        (DOM::XSLImportRuleImpl::href):
        (DOM::XSLImportRuleImpl::styleSheet):
        (DOM::XSLImportRuleImpl::isImportRule):
        * khtml/xsl/xslt_processorimpl.cpp:
        (DOM::m_sourceDocument):
        (DOM::stylesheetLoadFunc):
        (DOM::XSLTProcessorImpl::transformDocument):
        * khtml/xsl/xslt_processorimpl.h:

=== Safari-158 ===

2004-08-18  Trey Matteson  <trey@apple.com>

        3765958 - downstreamPosition() can hit infinite loop when at end of doc
        
        The problem was that I had a position that was after the maximum position in the text
        node, because it was the old caret position before a backspace was processed.  Later
        I happened to call downstream() on that position, and hit the bug.  Fix is to consider
        a position past the end point of its node if it is *greater than* or equal to its max offset.
                        
        Reviewed by Ken.

        * khtml/xml/dom_positioniterator.cpp:
        (DOM::PositionIterator::atEnd):

2004-08-18  David Hyatt  <hyatt@apple.com>

        Make the header overlap the footer in the zero-line case.  Changes to Emerson's template will ensure the
        header draws over the footer.
        
        Reviewed by darin

        * khtml/rendering/render_block.cpp:
        (khtml::getHeightForLineCount):

2004-08-18  Richard Williamson   <rjw@apple.com>

        Replace horrible pollForAppletInView: with new
        webPlugInGetApplet.  The details of how the applet instance
        is provided now belong to the Java team.  Yeh.
                
        Reviewed by Chris.
        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::getAppletInstanceForView):
        * kwq/WebCoreBridge.h:

2004-08-18  Chris Blumenberg  <cblu@apple.com>

        Fixed: <rdar://problem/3692199> 8A146: Safari crashes in toHTMLWithOptions, selection with no renderer (various sites)

        Reviewed by trey.

        * khtml/xml/dom2_rangeimpl.cpp:
        (DOM::RangeImpl::toHTML): renamed, don't assume that nodes of the range had renderers, use the common ancestor of the range as the root
        * khtml/xml/dom2_rangeimpl.h:
        * khtml/xml/dom_nodeimpl.cpp:
        (NodeImpl::recursive_toHTML): renamed, removed code that determines whether to include the root in the HTML, leave this up to the caller
        * khtml/xml/dom_nodeimpl.h:
        * kwq/WebCoreBridge.mm:
        (-[WebCoreBridge markupStringFromNode:nodes:]): call renamed methods
        (-[WebCoreBridge markupStringFromRange:nodes:]): ditto

2004-08-18  Ken Kocienda  <kocienda@apple.com>

        Reviewed by me

        * khtml/css/css_valueimpl.cpp:
        (CSSStyleDeclarationImpl::copy): Roll back silly last minute change that broke this code.
        Note to self: read code before making changes to it.

2004-08-18  Ken Kocienda  <kocienda@apple.com>

        Coded by Darin and Ken

        * khtml/css/css_computedstyle.cpp: Added CopyProperties static array. This contains
        the properties we implement that we also want to copy in the new
        CSSComputedStyleDeclarationImpl::copy described below.
        (DOM::CSSComputedStyleDeclarationImpl::copy): New function. In this class, copies the
        computed values of all the properties listed in CopyProperties. In essence, this makes
        a freeze-dired version of a computed style.
        (DOM::CSSComputedStyleDeclarationImpl::diff): Removes every property from the passed-in 
        CSSStyleDeclarationImpl that is also in the computed style.
        * khtml/css/css_computedstyle.h:
        * khtml/css/css_valueimpl.cpp:
        (CSSStyleDeclarationImpl::copy): New function. In this class, the copy operation is
        straightforward. Returns a copy that will be unchanged when the original changes.
        * khtml/css/css_valueimpl.h:
        (DOM::CSSStyleDeclarationImpl::values): Added a accessor suitable for use when the 
        CSSStyleDeclarationImpl is const.

2004-08-17  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Darin.

        <rdar://problem/3703768> CrashTracer: ...50 crashes at com.apple.WebCore: KHTMLPart::xmlDocImpl const + 0
        
        * khtml/khtmlview.cpp:
        (KHTMLView::viewportMouseMoveEvent): Add a nil check and an
        assertion for m_part being null. It seems impossible for this to
        happen, so we want to debug it ourselves, but in the meantime,
        let's try to avoid causing crashes for our users.

2004-08-17  David Hyatt  <hyatt@apple.com>

        Fix the line truncation function for Emerson so that at the far left setting of the slider, only the header
        is visible.
        
        Reviewed by darin

        * khtml/rendering/render_block.cpp:
        (khtml::getHeightForLineCount):
        * khtml/rendering/render_flexbox.cpp:
        (khtml::RenderFlexibleBox::layoutVerticalBox):

2004-08-17  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Hyatt

        Rewrite of the command that deletes a selection. I deleted great 
        big swaths of bug-ridden code to accomplish this and replaced it
        with code that is much cleaner and smarter.
        
        Also, renamed equivalentUpstreamPosition and equivalentDownstreamPosition to
        upstream to downstream, respectively.
        
        Added a couple of new helper methods.
        
        * khtml/editing/htmlediting.cpp: DeleteCollapsibleWhitespaceCommand and
        RemoveNodeAndPruneCommand now obsolete. A huge win.
        * khtml/editing/htmlediting.h: Ditto.
        * khtml/editing/htmlediting_impl.cpp:
        (khtml::debugPosition): Fix printf which had a placeholder, but no argument passed in the varargs.
        (khtml::CompositeEditCommandImpl::deleteUnrenderedText): New helper. Much simplified and cleaner
        version of 
        (khtml::ApplyStyleCommandImpl::doApply): upstream/downstream name change
        (khtml::ApplyStyleCommandImpl::nodeFullySelected):  upstream/downstream name change
        (khtml::DeleteSelectionCommandImpl::doApply):  upstream/downstream name change
        (khtml::DeleteTextCommandImpl::DeleteTextCommandImpl): Add an assert to check that the
        passed offset is less than the length of the text node.
        (khtml::InputNewlineCommandImpl::insertNodeAfterPosition): upstream/downstream name change
        (khtml::InputNewlineCommandImpl::insertNodeBeforePosition): upstream/downstream name change
        (khtml::InputNewlineCommandImpl::doApply): upstream/downstream name change
        (khtml::InputTextCommandImpl::prepareForTextInsertion): upstream/downstream name change
        (khtml::InputTextCommandImpl::execute): upstream/downstream name change
        (khtml::InputTextCommandImpl::insertSpace): upstream/downstream name change
        (khtml::ReplaceSelectionCommandImpl::doApply): upstream/downstream name change
        (khtml::TypingCommandImpl::issueCommandForDeleteKey): upstream/downstream name change
        (khtml::TypingCommandImpl::deleteKeyPressed):
        * khtml/editing/htmlediting_impl.h:
        * khtml/xml/dom_position.cpp:
        (DOM::Position::previousWordBoundary):
        (DOM::Position::nextWordBoundary):
        (DOM::Position::upstream):
        (DOM::Position::downstream):
        (DOM::Position::inRenderedText): Add null check.
        (DOM::Position::isRenderedCharacter): New helper.
        (DOM::isWS): New helper in this file.
        (DOM::Position::leadingWhitespacePosition): New helper. Factored out from htmlediting_impl.cpp.
        (DOM::Position::trailingWhitespacePosition): Ditto.
        (DOM::Position::debugPosition): Add null check.
        * khtml/xml/dom_position.h:
        * khtml/xml/dom_selection.cpp:
        (DOM::Selection::toRange): upstream/downstream name change
        (DOM::Selection::validate): upstream/downstream name change
        (DOM::Selection::debugPosition): upstream/downstream name change
        * layout-tests/editing/deleting/delete-block-contents-003-expected.txt: Updated tests with new expected results.
        * layout-tests/editing/deleting/delete-contiguous-ws-001-expected.txt: Ditto.
        * layout-tests/editing/deleting/delete-selection-001-expected.txt: Ditto.
        * layout-tests/editing/deleting/delete-trailing-ws-001-expected.txt: Ditto.
        * layout-tests/editing/inserting/insert-br-case1-expected.txt: Ditto.
        * layout-tests/editing/inserting/insert-br-case2-expected.txt: Ditto.
        * layout-tests/editing/style/style-3681552-fix-002-expected.txt: Ditto.

2004-08-17  Trey Matteson  <trey@apple.com>

        Various spelling fixes.

        Reviewed by Ken.

        * khtml/khtml_part.cpp:
        (KHTMLPart::setSelection): No misspellings in the spelling code comments
        * khtml/xml/dom_docimpl.cpp:
        (DocumentImpl::removeMarker):  Repaint if doc changes.  Sometimes the markers
        were not being erased when you clicked in a word.
        * khtml/xml/dom_position.cpp:
        (DOM::Position::previousWordBoundary):  Small optimization.  Bail after first
        try if the second try will not come out any different.
        (DOM::Position::nextWordBoundary):  Ditto
        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::markMisspellingsInSelection):  Comment.

2004-08-17  Darin Adler  <darin@apple.com>

        Reviewed by Maciej.

        - fixed <rdar://problem/3689700> crash loading page; stoxx.com (works in IE and Firefox)

        * khtml/khtml_part.h: Make completeURL public.
        * kwq/KWQKJavaAppletWidget.mm: (KJavaAppletWidget::KJavaAppletWidget): Complete the base URL
        before passing it across the bridge. This sidesteps the crashing bug in CFURL, filed as
        '<rdar://problem/3764632> CFURLCreateAbsoluteURLWithBytes crashes if passed the string "../.."'
        and also is obviously correct behavior that may fix other sites too.

        - fixed <rdar://problem/3547725> Crashes at csuohio.edu, list box vs. mouse event problem (Spoof No Fix)

        * kwq/KWQListBox.mm:
        (QListBox::~QListBox): Nil out the pointer from the KWQTableView back to the widget by calling
        a new detach method.
        (-[KWQTableView detach]): Set the pointer to the QListBox to 0. Also set the delegate and data
        source to nil, so we don't need nil checks in delegate and data source methods.
        (-[KWQTableView mouseDown:]): Add nil check.
        (-[KWQTableView keyDown:]): Add nil check.
        (-[KWQTableView keyUp:]): Add nil check.
        (-[KWQTableView becomeFirstResponder]): Add nil check.
        (-[KWQTableView resignFirstResponder]): Add nil check.
        (-[KWQTableView canBecomeKeyView]): Add nil check.
        (-[KWQTableView tableViewSelectionDidChange:]): Add nil checks, even though this is a delegate
        callback, to handle cases where calls to JavaScript result in the QListBox going away partway
        through.
        (-[KWQTableView drawRow:clipRect:]): Add nil check.
        (-[KWQTableView _accessibilityTableCell:tableColumn:]): Add nil check.

2004-08-17  Trey Matteson  <trey@apple.com>

        Fix ASSERT in spelling marker management.

        Reviewed by John.

        * khtml/xml/dom_docimpl.cpp:
        (DocumentImpl::shiftMarkers):  Use assert instead of ASSERT.
        Tweak test to allow for a start position of 0.

2004-08-17  Trey Matteson  <trey@apple.com>

        3764147 - failure of subframe to load leaves links in parent doc broken

        Reviewed by Maciej.

        * khtml/khtml_part.cpp:
        (KHTMLPart::childBegin):  New method to mark part as not complete.
        (KHTMLPart::processObjectRequest):  Mark child part imcomplete, so if we
        later get a failure on load it won't think it's already complete and do nothing.
        * khtml/khtml_part.h:

=== Safari-157 ===

2004-08-16  David Hyatt  <hyatt@apple.com>

        Fix the #define.  I had it all backwards.
        
        * WebCorePrefix.h:

2004-08-16  David Hyatt  <hyatt@apple.com>

        Land initial support for XSLT using xml-stylesheet PIs.
        
        * WebCore.pbproj/project.pbxproj:
        * khtml/khtml_part.cpp:
        (KHTMLPart::replaceDocImpl):
        * khtml/khtml_part.h:
        * khtml/khtmlview.h:
        * khtml/xml/dom_docimpl.cpp:
        (DocumentImpl::DocumentImpl):
        (DocumentImpl::~DocumentImpl):
        (DocumentImpl::recalcStyleSelector):
        (DocumentImpl::applyXSLTransform):
        (DocumentImpl::setTransformSourceDocument):
        * khtml/xml/dom_docimpl.h:
        (DOM::DocumentImpl::setTransformSource):
        (DOM::DocumentImpl::transformSource):
        (DOM::DocumentImpl::transformSourceDocument):
        * khtml/xml/dom_xmlimpl.cpp:
        (DOM::ProcessingInstructionImpl::checkStyleSheet):
        * khtml/xml/dom_xmlimpl.h:
        (DOM::ProcessingInstructionImpl::isXSL):
        * khtml/xml/xml_tokenizer.cpp:
        (khtml::matchFunc):
        (khtml::openFunc):
        (khtml::createQStringParser):
        (khtml::XMLTokenizer::processingInstruction):
        (khtml::XMLTokenizer::finish):
        (khtml::XMLTokenizer::setTransformSource):
        * khtml/xml/xml_tokenizer.h:
        (khtml::Tokenizer::setTransformSource):
        * khtml/xml/xsl_stylesheetimpl.cpp: Removed.
        * khtml/xml/xsl_stylesheetimpl.h: Removed.
        * khtml/xsl/xsl_stylesheetimpl.cpp: Added.
        (DOM::XSLStyleSheetImpl::XSLStyleSheetImpl):
        (DOM::XSLStyleSheetImpl::~XSLStyleSheetImpl):
        (DOM::XSLStyleSheetImpl::isLoading):
        (DOM::XSLStyleSheetImpl::checkLoaded):
        (DOM::XSLStyleSheetImpl::docLoader):
        (DOM::XSLStyleSheetImpl::parseString):
        * khtml/xsl/xsl_stylesheetimpl.h: Added.
        (DOM::XSLStyleSheetImpl::isXSLStyleSheet):
        (DOM::XSLStyleSheetImpl::type):
        (DOM::XSLStyleSheetImpl::ownerDocument):
        (DOM::XSLStyleSheetImpl::document):
        (DOM::XSLStyleSheetImpl::clearDocument):
        * khtml/xsl/xslt_processorimpl.cpp: Added.
        (DOM::m_sourceDocument):
        (DOM::XSLTProcessorImpl::~XSLTProcessorImpl):
        (DOM::XSLTProcessorImpl::transformDocument):
        (DOM::bufferWrite):
        (DOM::XSLTProcessorImpl::addToResult):
        (DOM::XSLTProcessorImpl::documentFromXMLDocPtr):
        * khtml/xsl/xslt_processorimpl.h: Added.

2004-08-16  Maciej Stachowiak  <mjs@apple.com>

        Fix build.

        * khtml/khtml_part.cpp:
        (KHTMLPart::isImmediateRedirectPending):

2004-08-16  Richard Williamson   <rjw@apple.com>

        Fixed <rdar://problem/3704339> Context2D forces integer positions in drawImage

        Use floats instead of ints to draw images.

        Reviewed by Ken.

        * khtml/ecma/kjs_html.cpp:
        (KJS::Context2DFunction::tryCall):
        * kwq/KWQPainter.h:
        * kwq/KWQPainter.mm:
        (QPainter::drawPixmap):
        (QPainter::drawFloatPixmap):

2004-08-13  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Darin.

        - fixed <rdar://problem/3752509> Pop up windows not showing up within SAP's BW Module (changing location.href on new window created by window.open)
        
        (actually the previous fix for this bug was mostly correct, but
        this additional change is needed to avoid the regression in
        <rdar://problem/3751025> REGRESSION: website rejects Safari 125.9
        as "need to upgrade to IE 6", but didn't reject 125.8

        So if merging for a software update, make sure to include both
        this and the previous fix.
        
        * khtml/html/htmltokenizer.cpp:
        (khtml::HTMLTokenizer::write): When there is an immediate reidrect pending,
        make sure to stop tokenizing, because we need to make sure no further
        script tags are processed beyond the one that triggered the redirect.
        * khtml/khtml_part.cpp:
        (KHTMLPart::isImmediateRedirectPending): New method to allow
        checking if a redirect is pending.
        * khtml/khtml_part.h:

2004-08-15  David Hyatt  <hyatt@apple.com>

        - fixed <rdar://problem/3760508> REGRESSION (154-155): No text in textarea

        Fix the blank textarea problem by ensuring that any change to a <textarea>'s DOM children causes the form control
        to resync with the DOM.  This behavior matches WinIE.
        
        Reviewed by kocienda

        * khtml/html/html_formimpl.cpp:
        (HTMLTextAreaElementImpl::childrenChanged):
        * khtml/html/html_formimpl.h:

2004-08-13  Trey Matteson  <trey@apple.com>

        3761794 Slider doesn't call onmouseup handler

        Reviewed by Hyatt

        * khtml/rendering/render_form.cpp:
        (RenderSlider::RenderSlider):  Listen for signal.
        (RenderSlider::slotClicked):  Pass to superclass.
        * khtml/rendering/render_form.h:
        * kwq/KWQSlider.h:
        * kwq/KWQSlider.mm:
        (-[KWQSlider mouseDown:]):  Generate mouseUp and clicked events,
        since AK consumes the mouseUp event in a modal tracking loop.
        (QSlider::QSlider):  Make signal.
        (QSlider::clicked):  Send signal.

2004-08-12  Trey Matteson  <trey@apple.com>

        3761329 - query result links all dead in ingrammicro.com (sometimes)
        3761328 - links in some docs dead when doc is loaded from WebArchive

        Make sure to get part to completed state when end is called,
        even if we have no doc.  See WebKit changelog for more info.

        Reviewed by Richard and Darin.

        * khtml/khtml_part.cpp:
        (KHTMLPart::end):

2004-08-13  Trey Matteson  <trey@apple.com>

        3761098 - red dotted underline for misspelled words shows up in drag image

        Reviewed by Ken

        * khtml/rendering/render_text.cpp:
        (RenderText::paint):  Don't draw misspelling when creating selection image.

2004-08-12  Richard Williamson   <rjw@apple.com>

        Bring npruntime.h and friends closer to compliance with
        latest spec.

        Reviewed by Maciej.

        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::windowScriptNPObject):

2004-08-12  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Richard.

        <rdar://problem/3245706> URLs with backslashes instead of slashes work on WinIE; should work on Safari (SAP)
        <rdar://problem/3506429> <BASE> tag containing backslash is breaking images with absolute URLs
        
        * kwq/KWQKURL.mm:
        (substituteBackslashes): Helper method
        (KURL::KURL): If the URL contains any backslashes, substitute all
        that appear before the query or fragment.

2004-08-12  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Chris

        Fix for this bug:
        
        <rdar://problem/3761014> command-down-arrow takes you to start of document instead of end of document

        * khtml/xml/dom_selection.cpp:
        (DOM::Selection::modifyExtendingRightForward): Code used to assume, incorrectly, that index 1 of the
        document element was beyond the last node in the document. But this is not true, since the document
        element is the HTML element (generally). Instead, move to the index equal to the number of children
        of the document element. This puts us past everything.
        (DOM::Selection::modifyMovingRightForward): Ditto.

2004-08-12  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Trey

        Fix for this bug:
        
        <rdar://problem/3695446> shift-down-arrow on last line of editable text should select to end of document
        
        Detect when current position is on first or last line and move to the
        start or end of that line, respectively.

        * khtml/xml/dom_position.cpp:
        (DOM::Position::previousLinePosition)
        (DOM::Position::nextLinePosition)

=== Safari-156 ===

2004-08-12  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Darin

        Some consolidation in style application code.

        * khtml/editing/htmlediting_impl.cpp:
        (khtml::StyleChange::StyleChange): Made this a full-on class and added a couple of 
        members and a constructors to make a StyleChange from a CSSStyleDeclarationImpl, as
        well as from a CSSStyleDeclarationImpl and a Position.
        (khtml::StyleChange::init): Common init function for StyleChange constructors.
        (khtml::StyleChange::currentlyHasStyle): Moved this here from ApplyStyleCommandImpl.
        (khtml::CompositeEditCommandImpl::applyTypingStyle): Tweak to adjust to new StyleChange
        interface.
        (khtml::ApplyStyleCommandImpl::applyStyleIfNeeded): Ditto.
        * khtml/editing/htmlediting_impl.h:
        (khtml::StyleChange::StyleChange):
        (khtml::StyleChange::cssStyle): New accessor.
        (khtml::StyleChange::applyBold): Ditto.
        (khtml::StyleChange::applyItalic): Ditto.
        * khtml/xml/dom_position.cpp:
        (DOM::Position::computedStyle): New helper.
        * khtml/xml/dom_position.h: Ditto.

2004-08-12  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Darin

        Fix for this bug:
        <rdar://problem/3751098> HTML email has one set of SPAN tags per character in the message
        
        Progress on this bug:
        <rdar://problem/3755562> Typing styles do not use same tag application conventions as font and color panel

        * khtml/editing/htmlediting_impl.cpp:
        (khtml::CompositeEditCommandImpl::applyTypingStyle): Name changed from createTypingStyleElement.
        Also, interface changed to take the node to which the typing style is to be applied. 
        This makes it easier to apply what may be up to three levels of nested tags to get the 
        desired style (<B>, <I>, and <SPAN STYLE="">).
        Also, Borrow some of the style change smarts from ApplyStyleCommandImpl to use bold and
        italic tags for applying styles when that is apprpriate. This creates on opportunity to
        factor the code to do this so that this function and the ApplyStyleCommandImpl class can
        share the implementation. I will follow up with a change to do that after landing this
        change. Some future code factoring could be done here to bring together some similar code
        into one place.
        (khtml::ApplyStyleCommandImpl::applyStyleIfNeeded): Add comment about code factoring work.
        (khtml::ApplyStyleCommandImpl::computeStyleChange): StyleChange struct no longer a member of the
        ApplyStyleCommandImpl class. CompositeEditCommandImpl needs it now in its applyTypingStyle()
        function.
        (khtml::InputNewlineCommandImpl::doApply): Pass along node to style to applyTypingStyle.
        (khtml::InputTextCommandImpl::prepareForTextInsertion): Ditto.
        * khtml/editing/htmlediting_impl.h:
        (khtml::StyleChange::StyleChange): Pull this struct out of ApplyStyleCommandImpl so 
        CompositeEditCommandImpl can use it.
        * khtml/khtml_part.cpp:
        (KHTMLPart::notifySelectionChanged): Always clear typing style when the selection
        changes, not only when closing typing. This fixes 3751098.
        
        These three tests actually had results that treated the buggy behavior as correct!
        
        * layout-tests/editing/style/style-3681552-fix-001-expected.txt
        * layout-tests/editing/style/style-3681552-fix-002-expected.txt
        * layout-tests/editing/style/typing-style-002-expected.txt

2004-08-12  Darin Adler  <darin@apple.com>

        Reviewed by Ken.

        - fixed <rdar://problem/3740485> Repro crash involving replacing content that includes form field

        * kwq/KWQLineEdit.mm: (QLineEdit::selectAll): Since this function calls selectText: which has a side
        effect of making the text field be first responder, call to bridge first to make it first responder.
        The bridge version lets WebHTMLView know we are changing the responder, avoiding some unpleasantness
        because it sets the "changing focus programmatically" flag. Without that flag set, we were getting
        an additional setFocusNode(0) call, which is unnecessary and incorrect.

2004-08-12  Darin Adler  <darin@apple.com>

        Reviewed by Ken.

        - fixed <rdar://problem/3758756> copying text selected with down arrow results in all text to end of document

        * khtml/xml/dom2_rangeimpl.h: Make startNode and pastEndNode public.
        * khtml/xml/dom2_rangeimpl.cpp: (DOM::RangeImpl::pastEndNode): Fix bug where this would return
        one node too far in the case where the end container was not a text node.

        * khtml/xml/dom_nodeimpl.cpp:
        (NodeImpl::recursive_toHTMLWithOptions): Rewrite loop, using startNode and pastEndNode,
        to fix bug where it would run past the end node, including too many nodes in the generated
        HTML. Nice side benefit: easier to read the code.

2004-08-11  Chris Blumenberg  <cblu@apple.com>

        Fixed: <rdar://problem/3758216> PARENTAL: buttons on parental controls page only work once

        Reviewed by john.

        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::submitForm): prevent a form from being submitted more than once only if it uses a scheme of http or https

2004-08-11  Darin Adler  <darin@apple.com>

        Reviewed by Ken.

        - fixed <rdar://problem/3715878> 8A162: connect.apple.com password field showed in cleartext

        * kwq/KWQTextField.mm: (-[KWQSecureTextField textDidEndEditing:]):
        Enhanced workaround for shifting focus from one secure text field to another so that it works
        even for the case of shifting focus from a secure text field back to itself.

2004-08-11  Ken Kocienda  <kocienda@apple.com>

        Reviewed by me

        Missed adding this file before.

        * kwq/KWQTextUtilities.mm: Added.
        (KWQFindNextWordFromIndex):

2004-08-11  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Trey

        Efficiency improvements on string manipulations in these two new function
        implementations. Use the versions of QString append/prepend that take
        (QChar *c, uint length) instead of creating new strings each time.

        * khtml/xml/dom_position.cpp:
        (DOM::Position::previousWordPosition): Changed, as described above.
        (DOM::Position::nextWordPosition): Ditto.

2004-08-11  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Darin

        Fix for this bug:
        <rdar://problem/3675812> Moving a word at a time does not use the correct conception of "word"
        
        I have implemented versions of previousWordPosition and nextWordPosition that are now
        different than previousWordBoundary and nextWordBoundary. The behavior of the new
        functions attempts to match what Cocoa does as closely as it can. Let the bug filing begin! 

        * WebCore.pbproj/project.pbxproj: Added KWQTextUtilities.mm
        * khtml/misc/helper.cpp:
        (khtml::nextWordFromIndex): Glue to call through to KWQFindNextWordFromIndex.
        * khtml/misc/helper.h: Declare the function above.
        * khtml/misc/khtml_text_operations.cpp: Added SimplifiedBackwardsTextIterator class.
        (khtml::SimplifiedBackwardsTextIterator::SimplifiedBackwardsTextIterator): New
        (khtml::SimplifiedBackwardsTextIterator::advance): Ditto.
        (khtml::SimplifiedBackwardsTextIterator::handleTextNode): Ditto.
        (khtml::SimplifiedBackwardsTextIterator::handleReplacedElement): Ditto.
        (khtml::SimplifiedBackwardsTextIterator::handleNonTextNode): Ditto.
        (khtml::SimplifiedBackwardsTextIterator::exitNode): Ditto.
        (khtml::SimplifiedBackwardsTextIterator::emitCharacter): Ditto.
        (khtml::SimplifiedBackwardsTextIterator::range): Ditto.
        * khtml/misc/khtml_text_operations.h:
        (khtml::SimplifiedBackwardsTextIterator::atEnd): Ditto.
        (khtml::SimplifiedBackwardsTextIterator::length): Ditto.
        (khtml::SimplifiedBackwardsTextIterator::characters): Ditto.
        * khtml/xml/dom_position.cpp:
        (DOM::Position::previousWordBoundary): Updated to gather appropriate text and call through to
        AppKit to perform the same calculations NSText uses.
        (DOM::Position::nextWordBoundary): Ditto.
        (DOM::Position::previousWordPosition): Unrelated change to fix case where the function could get "stuck".
        (DOM::Position::nextWordPosition): Ditto
        (DOM::Position::equivalentDeepPosition): Changed to look backwards if the position's offset is equal
        to the number of child nodes it has. This handles more cases correctly, like when the position is
        gives as one beyond the end of a document element's last child.
        * kwq/KWQTextUtilities.h: Declared KWQFindNextWordFromIndex.
        * kwq/KWQTextUtilities.mm: Added.
        (KWQFindNextWordFromIndex): New function.

2004-08-11  Ken Kocienda  <kocienda@apple.com>

        Reviewed by John

        Fix for this bug:
        
        <rdar://problem/3732702> crash in CSSComputedStyleDeclarationImpl running devtools.com editing sample code

        * khtml/khtml_part.cpp:
        (KHTMLPart::selectionComputedStyle): Added a null check.

2004-08-10  Darin Adler  <darin@apple.com>

        Reviewed by Trey.

        - fixed <rdar://problem/3710123> Loading iframe that replaces content in the parent document crashes Safari

        I fixed three problems:

            1) script interpreter destroyed while it was interpreting scripts, caused random havoc
            2) code trying to get to view after view was detached from part, caused nil-deref
            3) signals sent to parent after child was no longer in the parent's frames list, caused nil-deref

        Now the test page works fine. Hope the real sites do too.

        * khtml/khtml_part.h: Add connectChild and disconnectChild helper functions (private).
        * khtml/khtml_part.cpp:
        (KHTMLPart::clear): Call disconnectChild on each frame as we detach it (see below).
        (KHTMLPart::end): Ref the part at the start, and deref the part at the end, of this function.
        Otherwise, we can end up destroying the part, and hence the interpreter, inside a script that
        the interpreter itself is running.
        (KHTMLPart::slotFinishedParsing): Add another check for a nil m_view, after the call to
        checkCompleted.
        (KHTMLPart::checkCompleted): Remove bogus if statement with empty body.
        (KHTMLPart::processObjectRequest): Call disconnectChild to disconnect the child <-> parent signals of the
        old child that the new one is replacing, and connectChild to connect the signals (nicer factoring).
        (KHTMLPart::slotChildCompleted): Fixed up a confusing boolean if/expression to be simpler. Not related to
        the bug fix, but an earlier version of the fix had changes in this function.
        (KHTMLPart::connectChild): Added. Connects the appropriate signals for a child frame.
        (KHTMLPart::disconnectChild): Added. Disconnects the same signals that connectChild connects.

        * kwq/KWQKHTMLPart.mm: (KHTMLPart::frameDetached): Added a call to disconnectChild before removing the
        child from the frames list.

2004-08-09  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Trey.

        WebCore part of:

        - made basic marked text highlighting work to complete basic level of <rdar://problem/3704359> input method support not yet implemented for HTML editing

        * kwq/WebCoreBridge.mm:
        (-[WebCoreBridge setMarkedDOMRange:]): Added this new call to support storing
        a marked range in WebCore. The provided DOMRange must start and end in the same
        node, which must be a text node.
        (-[WebCoreBridge markedDOMRange]): New call to get the marked range. 
        (-[WebCoreBridge clearMarkedDOMRange]): New call to clear the marked range.
        * kwq/WebCoreBridge.h: Prototype new methods.
        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::markedRange): Implementation of WebCore call above.
        (KWQKHTMLPart::setMarkedRange): Implementation of WebCore call above -
        store the marked range, and repaint new and old nodes if needed.
        (KWQKHTMLPart::clear): Clear marked range.
        * kwq/KWQKHTMLPart.h: Prototype new methods.
        * khtml/rendering/render_text.cpp:
        (InlineTextBox::paintMarkedTextBackground): New method to paint the background
        for marked text, modeled on paintSelection.
        (RenderText::paint): Optionally handle painting marked text
        background as well as selection background in the marked text
        pass.
        * khtml/rendering/render_text.h: Prototype new method.

2004-08-10  Darin Adler  <darin@apple.com>

        Reviewed by Dave.

        - switch PCRE to do UTF-16 directly instead of converting to/from UTF-8 for speed

        * kwq/KWQRegExp.mm:
        (QRegExp::KWQRegExpPrivate::compile): Null-terminate the pattern and pass it.
        (QRegExp::match): Use the 16-bit string directly, no need to convert to UTF-8.

2004-08-10  Darin Adler  <darin@apple.com>

        Reviewed by Ken.

        - minor cleanup

        * khtml/editing/jsediting.cpp: Capitalize command names to match Windows.
        The dictionary lookup is case insensitive.

        * kwq/KWQFoundationExtras.h: Remove inaccurate comment.

2004-08-10  Trey Matteson  <trey@apple.com>

        3757094 - crash spell checking after a paste

        Reviewed by Ken

        * khtml/rendering/render_text.cpp:
        (InlineTextBox::paintSelection):  Add nil check.

2004-08-09  Trey Matteson  <trey@apple.com>

        3756195 - spell checking leaves misspelling marker behind after bad word is deleted
        ... and other follow-on spell check fixes

        Reviewed by Ken

        * khtml/rendering/render_text.cpp:
        (InlineTextBox::paintMarker):  Close inspection shows we were drawing the 
        misspelling marker one pixel lower than AK, and one pixel outside the selection
        rect we draw, in the case of Times-16.  So move it up one.  Still not an exact
        match for AK, but less bad.
        * khtml/xml/dom_docimpl.cpp:
        (DocumentImpl::removeMarker):  Track whether we make any changes, so we only
        repaint if something actually changed.
        (DocumentImpl::removeAllMarkers):  New utility.
        (DocumentImpl::removeAllMarkers):  Use clear() instead of (errant)
        hand-rolled loop to empty array.
        (DocumentImpl::shiftMarkers):  Track whether we make any changes, so we only
        repaint if something actually changed.
        * khtml/xml/dom_docimpl.h:
        * khtml/xml/dom_textimpl.cpp:
        (CharacterDataImpl::deleteData):  Along with shifting existing markers around,
        remove any markers in the deleted range.  Fixes 3756195.
        (CharacterDataImpl::replaceData):  Ditto for the replaced range.
        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::updateSpellChecking): comment

2004-08-08  Trey Matteson  <trey@apple.com>

        3745023 - Safari crashes trying to access anchor while downloading

        I bet this is behind a few other crashers as well.  In this bug the start of the
        download leaves a KWQPageState hanging around, and when that is freed it damages
        the part and view.  If you're still using that page, you're dead.

        The fix is to properly invalidate the PageState when we receive an error before
        reaching WebFrameCommitted state.  Normally this happens when a page is reheated
        from the PageState, but in this case we never manage to leave the page to begin
        with, although we've already created the PageState.

        Other errors besides the synthetic one download generates would have caused similar
        crashing.  Another example would be clicking on a second link before the load
        caused by clicking on the first link reached committed state.

        Reviewed by Richard

        * kwq/WebCoreBridge.h:
        * kwq/WebCoreBridge.mm:
        (-[WebCoreBridge didNotOpenURL:pageCache:]):  Invalidate the pageCache state
        when a load doesn't get off the ground.

2004-08-06  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Maciej

        Finish off spellchecking support to HTML editing. Includes work to
        enable continuous spellchecking.

        * khtml/editing/htmlediting_impl.cpp:
        (khtml::EditCommandImpl::markMisspellingsInSelection): Basically, a one-liner convenience to
        make the call over to the KWQKHTMLPart.
        (khtml::ReplaceSelectionCommandImpl::doApply): Did some rearranging of code so that the
        inserted content can be spell-checked. The function is basically the same, except for
        the addition of calls to markMisspellingsInSelection.
        (khtml::TypingCommandImpl::markMisspellingsAfterTyping): New function. Takes a look at the
        selection that results after typing and determines whether it needs to spellcheck. 
        Since the word containing the current selection is never marked, this does a check to
        see if typing made a new word that is not in the current selection. Basically, you
        get this by being at the end of a word and typing a space.
        (khtml::TypingCommandImpl::typingAddedToOpenCommand): Call markMisspellingsAfterTyping.
        * khtml/editing/htmlediting_impl.h: Add new function declarations.
         * khtml/khtml_part.cpp:
        (KHTMLPart::setSelection): Since spell checks are updated when the selection changes, 
        and every selection change passes through here, this is a good place to put the call 
        to the spellchecker.
        * khtml/rendering/render_text.cpp:
        (InlineTextBox::paintMarker): Remove temporary misspelling line drawing code. Replace with
        call that does AppKit-style drawing. Fix up some comments.
        * khtml/xml/dom_docimpl.cpp:
        (DocumentImpl::addMarker): Repaint the node that had the marker added. This makes it show 
        up on setting it.
        (DocumentImpl::removeMarker): Ditto.
        (DocumentImpl::removeAllMarkers): New function. Convenience for clearing all markers.
        Used when not in continuous spellchecking mode.
        (DocumentImpl::shiftMarkers): Moves markers in response to changes in a node's contents.
        This shifts the marker offsets by a given amount. This keeps the markers in the right
        place when a user types in a node with markers already set on it.
        * khtml/xml/dom_docimpl.h: Added new functions. Removed unnecessary enum qualifier from some
        declarations.
        * khtml/xml/dom_position.cpp:
        (DOM::Position::previousWordBoundary): This function was susceptible to endless loops...and
        needlessly so. Basically, if the current position is at a word boundary, run the code again
        to find the previous word boundary.
        (DOM::Position::nextWordBoundary): Same as above, but for next word boundary.
        * khtml/xml/dom_textimpl.cpp:
        (CharacterDataImpl::setData): Call shiftMarkers to update markers when this node changes.
        (CharacterDataImpl::insertData): Ditto.
        (CharacterDataImpl::deleteData): Ditto.
        (CharacterDataImpl::replaceData): Ditto.
        * kwq/KWQKHTMLPart.h:
        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::advanceToNextMisspelling):
        (KWQKHTMLPart::markMisspellingsInSelection):
        (KWQKHTMLPart::updateSpellChecking):
        (KWQKHTMLPart::respondToChangedSelection):
        * kwq/KWQPainter.h:
        * kwq/KWQPainter.mm:
        (QPainter::drawLineForMisspelling): New function. Call over to WebKit to do the drawing.
        * kwq/WebCoreBridge.h:
        * kwq/WebCoreBridge.mm:
        (-[WebCoreBridge alterCurrentSelection:direction:granularity:]): Pass markMisspellings flag to
        setSelection call. 
        * kwq/WebCoreTextRenderer.h:

=== Safari-155 ===

2004-08-05  David Hyatt  <hyatt@apple.com>

        Fix for 3752542, stack overflow that crashes Safari at www.dr.dk.  This bug is a regression caused by a fix that
        attempted to repair <caption> behavior to make it behave like Panther.  This fix was incorrect, and in addition
        even our <caption> behavior on Panther was incorrect.

        The patch that fixes this bug also makes <caption> handling work when <caption>s are contained inside a <td>, a <tr>,
        a <th>, or various table section tags (<tbody>, <tfoot>, <thead>).  The <caption> is pulled out and inserted just before
        the relevant ancestor table section.  This behavior matches other browsers.
        
        Reviewed by mjs

        * khtml/html/htmlparser.cpp:
        (KHTMLParser::insertNode):

2004-08-04  David Hyatt  <hyatt@apple.com>

        The top-level XSL sheet is now loaded and shows up in the activity window.  It is not yet parsed.
        
        Reviewed by kocienda

        * WebCore.pbproj/project.pbxproj:
        * khtml/css/css_base.h:
        (DOM::StyleBaseImpl::isXSLStyleSheet):
        * khtml/css/css_stylesheetimpl.cpp:
        (CSSStyleSheetImpl::CSSStyleSheetImpl):
        * khtml/css/css_stylesheetimpl.h:
        (DOM::StyleSheetImpl::isLoading):
        * khtml/misc/loader.cpp:
        * khtml/xml/dom_xmlimpl.cpp:
        (DOM::ProcessingInstructionImpl::ProcessingInstructionImpl):
        (DOM::ProcessingInstructionImpl::checkStyleSheet):
        (DOM::ProcessingInstructionImpl::sheet):
        (DOM::ProcessingInstructionImpl::isLoading):
        (DOM::ProcessingInstructionImpl::setStyleSheet):
        * khtml/xml/dom_xmlimpl.h:

2004-08-04  David Hyatt  <hyatt@apple.com>

        Add XSL Stylesheets to the WebCore cache.
        
        Reviewed by kocienda

        * khtml/misc/loader.cpp:
        (CachedXSLStyleSheet::CachedXSLStyleSheet):
        (CachedXSLStyleSheet::ref):
        (CachedXSLStyleSheet::deref):
        (CachedXSLStyleSheet::data):
        (CachedXSLStyleSheet::checkNotify):
        (CachedXSLStyleSheet::error):
        (DocLoader::requestXSLStyleSheet):
        (Cache::requestXSLStyleSheet):
        (Cache::getStatistics):
        * khtml/misc/loader.h:
        (khtml::CachedObject::):
        (khtml::CachedXSLStyleSheet::sheet):
        (khtml::CachedXSLStyleSheet::schedule):

2004-08-03  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Hyatt

        Changes to improve our handling of object tags during editing, including
        fixing this bug:
        
        <rdar://problem/3744533> Problem editing <OBJECT> elements displayed by WebPlugIns 

        Part of the fix for this bug involves removing the close() function from render objects.
        It was decided that this code was no longer needed, as the work it did could be moved
        to other, more modern, places.

        Fixed this bug:
        
        <rdar://problem/3748537> crash due to nil node passed into parentNode in computeTypingStyle deleting text

        Finally, I filed and fixed this bug I discovered while fixing the one above:
        
        <rdar://problem/3749338> Select-all + delete leaves editing view without blinking caret

        * khtml/editing/htmlediting_impl.cpp:
        (khtml::DeleteSelectionCommandImpl::computeTypingStyle): Fix for 3748537. Put in some more null
        checks. Bail, returning 0 for typing style if any null checks yield a null.
        (khtml::DeleteSelectionCommandImpl::doApply): Fix for 3749338. A removeNodeAndPrune call may wind
        up deleting the node where we calculated that we wanted to put the selection after deleting.
        If this happens, move this ending selection to a sensible alternative.
        (khtml::RemoveNodeAndPruneCommandImpl::doApply): Call previousNodeConsideringAtomicNodes instead 
        of traversePreviousNode when doing the prune.
        * khtml/html/html_objectimpl.cpp:
        (HTMLObjectElementImpl::attach): Part of the fix for 3744533.
        (HTMLObjectElementImpl::recalcStyle): Change old strcmp check for type of renderer to be a call
        to canRenderImageType.
        (HTMLObjectElementImpl::childrenChanged): New function. Helps to keep object tags up to date 
        as their children change. This also will help to make object tags respond properly to having their
        params changed by DOM calls.
        * khtml/html/html_objectimpl.h: Cosmetic change.
        * khtml/html/htmlparser.cpp:
        (KHTMLParser::insertNode): Remove call to obsolete closeRenderer() function.
        (KHTMLParser::popOneBlock): Ditto.
        * khtml/rendering/render_block.cpp:
        (khtml::RenderBlock::makeChildrenNonInline): Remove call to obsolete close() function.
        * khtml/rendering/render_container.cpp:
        (RenderContainer::updatePseudoChild): Ditto.
        * khtml/rendering/render_form.cpp: Remove obsolete close() function.
        * khtml/rendering/render_form.h: Ditto.
        * khtml/rendering/render_frames.cpp: Ditto.
        * khtml/rendering/render_frames.h: Ditto.
        * khtml/rendering/render_inline.cpp:
        (RenderInline::splitFlow): Remove calls to obsolete close() function.
        * khtml/rendering/render_object.h: Remove obsolete close() function.
        * khtml/rendering/render_table.cpp:
        (RenderTableCell::layout): Ditto.
        * khtml/rendering/render_table.h: Ditto.
        * khtml/xml/dom_docimpl.cpp:
        (DocumentImpl::closeInternal): Remove call to obsolete close() function.
        * khtml/xml/dom_nodeimpl.cpp: Remove obsolete m_rendererNeedsClose initialization.
        (NodeImpl::NodeImpl): Remove obsolete closeRenderer() function.
        (NodeImpl::attach): Remove call to obsolete close() function.
        (NodeImpl::isAtomicNode): New function. Helps to fix 3744533. Determines if a node should
        be treated as an atomic node for the purposes of editing.
        (NodeImpl::previousNodeConsideringAtomicNodes): New helper to traverse tree taking atomic nodes
        into account.
        (NodeImpl::nextNodeConsideringAtomicNodes): Ditto.
        (NodeImpl::previousLeafNode): Now calls nextNodeConsideringAtomicNodes to iterate. This helps to 
        prevent deleting PARAM tag portions of object tags erroneously.
        (NodeImpl::nextLeafNode): Ditto.
        * khtml/xml/dom_nodeimpl.h: Removed m_rendererNeedsClose bit. Added declarations for new functions.
        * khtml/xml/dom_position.cpp:
        (DOM::Position::equivalentDeepPosition): Now takes atomic nodes into account as it drills down into the
        tree.
        * khtml/xml/xml_tokenizer.cpp:
        (khtml::XMLTokenizer::endElement): Remove call to obsolete closeRenderer() function.

2004-08-03  David Hyatt  <hyatt@apple.com>

        Add the deprecated text/xsl MIME type (introduced by Internet Explorer 5) as an acceptable MIME type for XML
        documents.
        
        Reviewed by john

        * khtml/ecma/xmlhttprequest.cpp:
        (KJS::XMLHttpRequest::getValueProperty):
        * khtml/khtml_part.cpp:
        (KHTMLPart::begin):
        * khtml/misc/loader.cpp:
        (CachedXBLDocument::CachedXBLDocument):

2004-08-03  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Ken.

        - remove assertions that asserted text is non-empty, just avoid
        doing anything for the empty case; this is needed because input
        methods like to insert empty text in various cases.

        * khtml/editing/htmlediting_impl.cpp:
        (khtml::InsertTextCommandImpl::InsertTextCommandImpl):
        (khtml::InsertTextCommandImpl::doApply):
        (khtml::InsertTextCommandImpl::doUnapply):

2004-08-03  Darin Adler  <darin@apple.com>

        Reviewed by Ken.

        - fixed <rdar://problem/3740937> ER: A way to turn a DOMRange into text (equivalent of -innerText)

        * khtml/xml/dom2_rangeimpl.h: Added text function. Like innerText, but on a range.
        * khtml/xml/dom2_rangeimpl.cpp: (DOM::RangeImpl::text): Added.

        * kwq/DOM.mm: (-[DOMRange _text]): Added. Calls DOM::RangeImpl::text.
        * kwq/DOMPrivate.h: Added.

        * WebCore.pbproj/project.pbxproj: Added DOMPrivate.h, an internal header (private in WebKit, internal here).

2004-08-02  John Sullivan  <sullivan@apple.com>

        Reviewed by Darin.
        
        WebCore part of fix for <rdar://problem/3631868> NSToolbar adoption: 
        Tab key should cycle around toolbar and page content

        * kwq/WebCoreBridge.mm:
        (-[WebCoreBridge nextKeyViewInsideWebFrameViews]):
        Made this method start looking from the current focus node. This won't
        affect any existing callers because there were no existing callers.
        (-[WebCoreBridge previousKeyViewInsideWebFrameViews]):
        ditto

2004-08-02  David Hyatt  <hyatt@apple.com>

        Create a #define for XSLT support that at the moment will only be enabled on Tiger.
        
        Reviewed by kocienda

        * WebCorePrefix.h:

2004-08-02  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Hyatt

        Fix for this bug:
        
        <rdar://problem/3747945> Deleting replaced element can cause crash

        * khtml/xml/dom_selection.cpp:
        (DOM::Selection::layoutCaret): The crux of the problem is that
        the caret drawing code tries to draw the caret at the image
        offset after it has been removed from the document. So, make
        sure the start node for the selection is in the document before
        trying to lay out the caret using that node.

2004-07-31  Ken Kocienda  <kocienda@apple.com>

        Reviewed by John

        <rdar://problem/3745498> HTMLCompose: Can't edit new empty message (and typing eventually crashes)
        <rdar://problem/3746408> HTMLCompose: cannot type in message body of new messages (and tabbing causes a crash.)
        
        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::setDisplaysWithFocusAttributes): Add in some smarts to look around for
        a good place to put the caret if focus is turning "on", the part is contentEditable,
        and it does not currently have a selection. This has the effect of flashing the caret 
        in a contentEditable view automatically without requiring the programmer to set a 
        selection explicitly. This also fixes the bug listed above. Both are cases where the
        bad behavior happened since the editing code tried to process input without an active 
        selection. I always took the extra step of setting the selection explicitly when
        opening new windows in programs like Blot, but it seems silly to require developers to
        do this in general. With this patch, they no longer need to.

2004-07-30  Ken Kocienda  <kocienda@apple.com>

        Reviewed by John

        Fix for this bug:
        
        <rdar://problem/3745498> HTMLCompose: Can't edit new empty message (and typing eventually crashes)

        * khtml/xml/dom_selection.cpp:
        (DOM::Selection::layoutCaret): Don't bail when the start position of the selection is not in rendered
        content. The validate() bottleneck function will do all it can to make sure that the start is
        moved to rendered content if at all possible before this function runs. If no rendered position can
        be found, like in the case of an empty body element, we still want to flash the caret there.
        (DOM::Selection::validate): Look for rendered positions for the base and extent, but save off the
        base before making this move, and set the selection to the enclosing block flow element of this
        original base if no rendered positions can be found. This has the effect of flashing the caret 
        someplace in an editable block, even if it does not contain any rendered content.

2004-07-30  Ken Kocienda  <kocienda@apple.com>

        Reviewed by John

        Fix a large number of editing layout test regressions.

        * khtml/misc/khtml_text_operations.cpp:
        (khtml::TextIterator::TextIterator): Use a new way to compute the end offset of a node for
        purposes of text iteration. The end offset is either child count of a node with children,
        or the maxOffset() of a node that does not.
        * khtml/xml/dom_nodeimpl.cpp: Remove maxOffset() implementation from NodeBaseImpl. This 
        conflicted with the usage of this function that is needed for editing.
        * khtml/xml/dom_nodeimpl.h: Ditto.

2004-07-30  Trey Matteson  <trey@apple.com>

        Next steps for spell checking:  We have data structures for the marked pieces of
        text and the smarts to draw them (although at this point, they just get a green
        underline instead of using the real AppKit pattern.)

        Note we don't call this code outside of development, since at this point I suspect
        it could be made to crash by mixing spelling and editing. 

        Reviewed by Ken.

        * khtml/rendering/render_text.cpp:
        (InlineTextBox::paintMarker):  Paint one marker's intersection with a text run
        (RenderText::paint):  Find intersections of all markers and runs, call paintMarker
        * khtml/rendering/render_text.h:
        * khtml/xml/dom_docimpl.cpp:
        (DocumentImpl::addMarker):  Adds a marker to a DOM Range.
        (DocumentImpl::removeMarker):  Removes a marker from a DOM Range.
        (DocumentImpl::addMarker):  Adds a marker to a single DOM Node, merging previous
        markers as needed.
        (DocumentImpl::removeMarker):  Removes a marker from a single DOM Node, breaking
        up previous markers as needed.
        (DocumentImpl::markersForNode):  Return markers for a Node.
        * khtml/xml/dom_docimpl.h:
        (DOM::DocumentMarker::):  New marker struct
        (DOM::DocumentMarker::operator == ):
        (DOM::DocumentMarker::operator != ):
        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::advanceToNextMisspelling):  Mark misspelled ranges when we find them.
        Debug only for now.

2004-07-29  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Darin.

        <rdar://problem/3745808> Seed: WebKit: Table's caption broken
        
        * khtml/html/htmlparser.cpp:
        (KHTMLParser::insertNode): When a caption appears in an illegal
        place in a table, pop blocks until we hit a place where it's
        allowed.

=== Safari-154 ===

2004-07-29  Darin Adler  <darin@apple.com>

        Reviewed by Ken.

        - added constants for motion across entire document for use in operations like
          move to beginning of document

        * khtml/xml/dom_selection.h: Added DOCUMENT to ETextGranularity.
        * khtml/xml/dom_selection.cpp:
        (DOM::Selection::modifyExtendingRightForward): Added case for DOCUMENT.
        (DOM::Selection::modifyMovingRightForward): Added case for DOCUMENT.
        (DOM::Selection::modifyExtendingLeftBackward): Added case for DOCUMENT.
        (DOM::Selection::modifyMovingLeftBackward): Added case for DOCUMENT.
        (DOM::Selection::validate): Changed if statements into switch statement,
        added case for DOCUMENT.

        * kwq/WebCoreBridge.h: Added WebCoreBridge to WebSelectionGranularity.

2004-07-28  Trey Matteson  <trey@apple.com>

        Small refinement of last checkin.  The text iterators now return reasonable values
        for range() when atEnd, which gets rid of special cases and pitfalls in the client.

        Reviewed by Ken.

        * khtml/misc/khtml_text_operations.cpp:
        (khtml::TextIterator::TextIterator):  Set endOffset to a valid value instead of
        using LONG_MAX.
        (khtml::TextIterator::range):  Implement when atEnd.
        (khtml::CharacterIterator::range):  Don't massage result when atEnd.
        (khtml::WordAwareIterator::WordAwareIterator):  Now that this is fixed, we no
        longer need this special case.
        (khtml::WordAwareIterator::advance):  Always set range, even when atEnd, now that
        TextIterator::range() always works.
        * khtml/xml/dom_nodeimpl.cpp:
        (NodeImpl::childNodeCount):  Made const
        (NodeBaseImpl::maxOffset):  Add missing implementation
        * khtml/xml/dom_nodeimpl.h:
        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::advanceToNextMisspelling):  Get rid of special cases, always call
        range on our iterator even when it is atEnd.

2004-07-28  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Trey.

        <rdar://problem/3678534> controls on page at http://help.sap.com don't work properly

        To fix this I implemented the various JavaScript BarInfo properties in JavaScript.
        
        * khtml/ecma/kjs_window.cpp:
        (Window::Window):
        (Window::locationbar):
        (Window::menubar):
        (Window::personalbar):
        (Window::statusbar):
        (Window::toolbar):
        (Window::scrollbars):
        (Window::mark):
        (Window::get):
        (SelectionFunc::tryCall):
        (BarInfo::BarInfo):
        (BarInfo::~BarInfo):
        (BarInfo::get):
        (BarInfo::put):
        * khtml/ecma/kjs_window.h:
        (KJS::Window::):
        (KJS::BarInfo::):
        (KJS::BarInfo::part):
        (KJS::BarInfo::classInfo):
        * khtml/ecma/kjs_window.lut.h:
        (KJS::):
        * kwq/KWQKHTMLPart.h:
        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::locationbarVisible):
        (KWQKHTMLPart::menubarVisible):
        (KWQKHTMLPart::personalbarVisible):
        (KWQKHTMLPart::scrollbarsVisible):
        (KWQKHTMLPart::statusbarVisible):
        (KWQKHTMLPart::toolbarVisible):

2004-07-28  Ken Kocienda  <kocienda@apple.com>

        Reviewed by me
        
        Added more layout tests.

        * layout-tests/editing/style/style-3681552-fix-001-expected.txt: Added.
        * layout-tests/editing/style/style-3681552-fix-001.html: Added.
        * layout-tests/editing/style/style-3681552-fix-002-expected.txt: Added.
        * layout-tests/editing/style/style-3681552-fix-002.html: Added.
        * layout-tests/editing/style/style-3690704-fix-expected.txt: Added.
        * layout-tests/editing/style/style-3690704-fix.html: Added.
        * layout-tests/editing/style/typing-style-001-expected.txt: Added.
        * layout-tests/editing/style/typing-style-001.html: Added.
        * layout-tests/editing/style/typing-style-002-expected.txt: Added.
        * layout-tests/editing/style/typing-style-002.html: Added.

2004-07-28  Ken Kocienda  <kocienda@apple.com>

        Reviewed by me
        
        Added some layout tests.

        * layout-tests/editing/selection/select-all-001-expected.txt: Added.
        * layout-tests/editing/selection/select-all-001.html: Added.
        * layout-tests/editing/selection/select-all-002-expected.txt: Added.
        * layout-tests/editing/selection/select-all-002.html: Added.
        * layout-tests/editing/selection/select-all-003-expected.txt: Added.
        * layout-tests/editing/selection/select-all-003.html: Added.

2004-07-28  Trey Matteson  <trey@apple.com>

        Spellchecking, Part I.  Basic spellcheck is working.  Spelling panel is hooked up.

        At this point, no special marking of misspellings, no grammar check, no context
        menu integration, no "check continually" mode.

        Much of the TextIterator and CharacterIterator interface got published outside
        of khtml_text_operations.cpp, with a little API rationalizing.

        Reviewed by Ken.

        * khtml/misc/khtml_text_operations.cpp:
        (khtml::TextIterator::range):  Name changes.
        (khtml::CharacterIterator::CharacterIterator):  Ditto.
        (khtml::CharacterIterator::range):  Ditto.
        (khtml::CharacterIterator::advance):  Ditto.
        (khtml::CharacterIterator::string):  New method to consume chars into a string.
        (khtml::WordAwareIterator::WordAwareIterator):  New class that iterates over
        the text respecting word boundaries.
        (khtml::WordAwareIterator::advance):
        (khtml::WordAwareIterator::length):
        (khtml::WordAwareIterator::characters):
        (khtml::plainText):  Name changes.
        (khtml::findPlainText):  Ditto.

        API moved from cpp to header file.
        * khtml/misc/khtml_text_operations.h:
        (khtml::TextIterator::atEnd):
        (khtml::TextIterator::length):
        (khtml::TextIterator::characters):
        (khtml::CharacterIterator::atBreak):
        (khtml::CharacterIterator::atEnd):
        (khtml::CharacterIterator::length):
        (khtml::CharacterIterator::characters):
        (khtml::CharacterIterator::characterOffset):
        (khtml::WordAwareIterator::atEnd):
        (khtml::WordAwareIterator::range):

        * khtml/xml/dom_position.cpp:
        (DOM::Position::previousWordBoundary):  New name for the old routine.  This routines semantics
        match the current behavior of this code.
        (DOM::Position::nextWordBoundary):  Ditto.
        (DOM::Position::previousWordPosition):  Call old code with the new name.  When we fix
        word advancement, this routine will have its own impl.
        (DOM::Position::nextWordPosition):  Ditto.
        * khtml/xml/dom_position.h:
        * khtml/xml/dom_selection.h:
        (DOM::Selection::rangeStart):  New convenience methods
        (DOM::Selection::rangeEnd):
        * kwq/KWQKHTMLPart.h:
        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::findString):  Ensure we use range-compatible positions.
        (KWQKHTMLPart::advanceToNextMisspelling):  Brand new.
        * kwq/WebCoreBridge.h:
        * kwq/WebCoreBridge.mm:
        (-[WebCoreBridge advanceToNextMisspelling]):  Typical bridge glue.

2004-07-28  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Maciej
        
        Fixed this bug:
        
        <rdar://problem/3690704> marking partly bold text italic across blocks wrongly extends the bold section

        * khtml/editing/htmlediting_impl.cpp:
        (khtml::ApplyStyleCommandImpl::doApply): Fixed loop that gathers up groups of nodes to
        pass off to the function that applies styles. While this is meant to be conservative in
        order to leave the DOM well-formed in all cases, it contained an error where the set
        of nodes to be styled together erroneously could include a node that was meant for 
        the next set. Fixed.
        (khtml::ApplyStyleCommandImpl::applyStyleIfNeeded): Unrelated change to remove DOM:: scope
        resolution qualifier. Unneeded here.

2004-07-28  Darin Adler  <darin@apple.com>

        Reviewed by Trey.

        - fixed <rdar://problem/3658471> REGRESSION: Node.appendChild( ) fails when parent already contains that child

        * khtml/xml/dom_nodeimpl.cpp: (NodeImpl::isAncestor): Restore the original meaning of this function.
        It returns true if the parameter is an ancestor of this, but had been changed to return true if this
        is an ancestor of the parameter. However, we do retain one change we made at the same time, which is
        that it does not consider a node an ancestor of itself.

        * khtml/editing/htmlediting_impl.cpp: (khtml::ApplyStyleCommandImpl::nodeFullySelected):
        * khtml/xml/dom2_traversalimpl.cpp: (DOM::NodeIteratorImpl::notifyBeforeNodeRemoval):
        Reverse parameters for callers who wanted the new meaning of isAncestor, with care to not use it in
        any cases where the pointer might be 0.

        * khtml/xml/dom_nodeimpl.h: Added const to the parameter to make things more symmetric and allow the
        new uses to all compile.

2004-07-28  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Darin

        Fix for this bug:
        
        <rdar://problem/3681552> html editing needs to preserve typing font when replacing selection

        * khtml/css/css_valueimpl.cpp:
        (CSSStyleDeclarationImpl::merge): New helper that merges styles together. Helpful for
        dealing with typing styles.
        * khtml/css/css_valueimpl.h:
        * khtml/editing/htmlediting_impl.cpp:
        (khtml::DeleteSelectionCommandImpl::computeTypingStyle): New helper that updates the
        typing style based on the current selection. This is a convenient bottleneck for all
        the code that needs to worry about typing style.
        (khtml::DeleteSelectionCommandImpl::doApply): Update typing style before doing the delete.
        (khtml::InputNewlineCommandImpl::doApply): No need to redeclare exceptionCode local.
        Improve comments. 
        (khtml::InputTextCommandImpl::prepareForTextInsertion): Remove unneeded complication when 
        figuring out where to insert style node. Not even sure what I was trying to do here, but
        it does not seem to be needed any more.
        (khtml::InputTextCommandImpl::execute): Remove unneeded comment.
        * khtml/editing/htmlediting_impl.h: Declare new computeTypingStyle helper.
        * khtml/khtml_part.cpp:
        (KHTMLPart::setCaretVisible): Call selectionLayoutChanged instead of notifySelectionChanged
        in this function. The selection did not change simply by calling this function, but it
        does need a layout.
        (KHTMLPart::notifySelectionChanged): Treat clearing the typing style much like closing typing,
        instead of clearing it unconditionally.
        (KHTMLPart::applyStyle): In the case where the current selection is a caret, merge the
        style being applied with any current typing style that already exists.

2004-07-28  Darin Adler  <darin@apple.com>

        Reviewed by Ken.

        - fixed <rdar://problem/3711080> REGRESSION (125-146): form fields have shrunk by 1 pixel, and now clip descenders

        * kwq/KWQLineEdit.mm: (QLineEdit::sizeForCharacterWidth):
        Compute the size without ever calling cellSize. After talking with Andrew Platzer,
        we decided that hard-coding the size of the borders of a text field was the best
        way for WebCore to handle this.

2004-07-28  Darin Adler  <darin@apple.com>

        Reviewed by Ken.

        - fixed <rdar://problem/3743204> REGRESSION: crash at startribune.com in InlineBox::root()

        * khtml/rendering/render_flow.cpp: (RenderFlow::dirtyLinesFromChangedChild):
        Added a check for nil.

2004-07-27  Darin Adler  <darin@apple.com>

        Reviewed by Trey.

        - fixed <rdar://problem/3743138> opaque colors come out of getComputedStyle as "rgba", transparent ones as "rgb"

        * khtml/css/css_valueimpl.cpp: (CSSPrimitiveValueImpl::cssText): Use "rgb" rather than
        "rgba" when alpha is 0xFF, not when alpha is 0.

2004-07-27  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Hyatt
        
        Fixed the problem where BR elements on lines by themselves don't paint when selected.

        * khtml/khtml_part.cpp:
        (KHTMLPart::selectAll): Related fix. Include BR's when figuring out what "all" is.
        * khtml/rendering/render_br.cpp:
        (RenderBR::paint): New function implementation. Teach BR's how to paint selections.
        * khtml/rendering/render_br.h:
        * khtml/rendering/render_text.cpp: 
        (InlineTextBox::paintSelection): Fix some geometry calculations to be more readable (i.e. don't
        reuse function argument as a local). Also, improve the logic for determining whether to 
        extend the selection to block boundaries.
        (RenderText::paint): Remove extendSelection argument from paintSelection. The logic to figure out
        extensions is now fully contained in paintSelection.
        * khtml/rendering/render_text.h:

2004-07-27  David Hyatt  <hyatt@apple.com>

        Fix for 3665211 (again).  Make sure not to leave child line boxes pointing to deleted ancestor line boxes.
        
        Reviewed by kocienda

        * khtml/rendering/render_block.cpp:
        (khtml::RenderBlock::removeChild):

2004-07-26  Chris Blumenberg  <cblu@apple.com>

        Fixed: WebView scrolls to the top after making an editing change via drag & drop

        Fixed by Ken, reviewed by me.

        * kwq/WebCoreBridge.mm:
        (-[WebCoreBridge ensureCaretVisible]): do nothing if the selection is not a caret

=== Safari-153 ===

2004-07-26  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by John.

        <rdar://problem/3740855> REGRESSION: Flash incorrectly positioned at macromedia.com
        
        * khtml/html/htmltokenizer.cpp:
        (khtml::HTMLTokenizer::scriptExecution): Do the prepending-src
        trick as for inline script execution.

        Added new layout tests covering this and other recent tokenizer
        bugs.
        
        * layout-tests/fast/tokenizer/001-expected.txt: Added.
        * layout-tests/fast/tokenizer/001.html: Added.
        * layout-tests/fast/tokenizer/002-expected.txt: Added.
        * layout-tests/fast/tokenizer/002.html: Added.
        * layout-tests/fast/tokenizer/003-expected.txt: Added.
        * layout-tests/fast/tokenizer/003.html: Added.
        * layout-tests/fast/tokenizer/resources/003-script.js: Added.

2004-07-23  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Trey

        Fix for this bug:
        
        <rdar://problem/3738920> Caret blinks in inactive window

        As part of the fix, I cleaned up the way we handle special drawing that needs
        to be done in the HTML view that is first responder in the key window (e.g the
        drawing of text selection highlight and caret blinking).

        * khtml/khtml_part.cpp: Removed setCaretVisible function. Updating caret visibility
        is now done in setDisplaysWithFocusAttributes, described below.
        * khtml/khtml_part.h: Ditto.
        * khtml/khtmlpart_p.h:
        (KHTMLPartPrivate::KHTMLPartPrivate): m_caretVisible now defaults to false. This prevents
        the caret from blinking when an app is started from the command line and remains in the
        background.
        * khtml/khtmlview.cpp:
        (KHTMLView::focusInEvent): Remove caret code from here. Not needed.
        (KHTMLView::focusOutEvent): Ditto.
        * kwq/KWQKHTMLPart.h:
        (KWQKHTMLPart::displaysWithFocusAttributes): New accessor to return cached value to other code
        in WebCore.
        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::KWQKHTMLPart):
        (KWQKHTMLPart::setDisplaysWithFocusAttributes): Renamed from setShowsFirstResponder and modified
        to do the display updating in a clearer way....and now has comments!
        * kwq/WebCoreBridge.h: Remove several obsolete functions that use to try to do the work of the new
        setDisplaysWithFocusAttributes function (and did so less well).
        * kwq/WebCoreBridge.mm:
        (-[WebCoreBridge drawRect:]): Now calls displaysWithFocusAttributes to figure out how to draw the
        text selection background.
        (-[WebCoreBridge setDisplaysWithFocusAttributes:]): Calls through to setDisplaysWithFocusAttributes in 
        KWQKHTMLPart.
        (-[WebCoreBridge selectionColor]): Flipped the value of this ternary expression to accommodate the
        change from usesInactiveTextBackgroundColor to setDisplaysWithFocusAttributes.
        (-[WebCoreBridge setCaretVisible:]): Removed.

2004-07-23  John Sullivan  <sullivan@apple.com>

        Reviewed by Maciej.

        - fixed <rdar://problem/3691569> REGRESSION (142): cmd-shift-clicking on a link 
        now also extends selection (even if there wasn't one before)

        * khtml/khtml_part.cpp:
        (KHTMLPart::handleMousePressEventSingleClick):
        Uh, duh. Got a ! wrong last time somehow, so it did even more the wrong thing
        than ever. I did test this before, so my best guess is an accidental undo or
        something before committing.

2004-07-23  David Hyatt  <hyatt@apple.com>

        Fix for 3735084 and 3737209, when you can find no line boxes in your previous sibling, the line box to dirty
        should be the *first* line box and not the last.
        
        Reviewed by kocienda

        * khtml/rendering/render_flow.cpp:
        (RenderFlow::dirtyLinesFromChangedChild):

2004-07-23  Ken Kocienda  <kocienda@apple.com>

        Reviewed by John

        * khtml/xml/dom_selection.cpp:
        (DOM::Selection::validate): Not exactly a fix of anything per se, but more of a refinement
        of how selection validation works. The idea is to let the selection validation process
        end up with an empty selection given real DOM position inputs if the validation code
        cannot find a rendered position. Since selection is all about display, this makes sense.
        Currently, the code will allow the selection to be placed in a location that is not
        rendered, and this does nobody any good, as it can lead to bugs like accepting key input
        in a div that is set to display:none.

2004-07-23  Ken Kocienda  <kocienda@apple.com>

        Reviewed by me

        Updated layout tests with new expected results.

        * layout-tests/editing/execCommand/boldSelection-expected.txt
        * layout-tests/editing/selection/extend-by-character-001-expected.txt
        * layout-tests/editing/selection/extend-by-character-003-expected.txt
        * layout-tests/editing/selection/unrendered-005-expected.txt

2004-07-22  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Kevin.

        <rdar://problem/3725467> REGRESSION(140-142) nothing displayed at http://Ye-Olde-Movies.tripod.com
        <rdar://problem/3599494> REGRESSION (100-125): Frame does not refresh until user clicks on browser.
        
        * khtml/rendering/render_frames.cpp:
        (RenderFrameSet::layout): When returning early because there are
        no children, make sure to mark self as no longer needing
        layout. Otherwise, when children are added, we'll fail to
        propagate childrenNeedLayout to our own parent! This is almost
        certainly a long-standing bug that was just masked by our old
        layout tomfoolery.

2004-07-22  Kevin Decker  <kdecker@apple.com>

        Reviewed by Darin

        Fixed <rdar://problem/3682340> (error console does not include source urls or line numbers of event exceptions).

        * khtml/ecma/kjs_events.cpp:
        (JSLazyEventListener::JSLazyEventListener):
        (JSLazyEventListener::parseCode):
        * khtml/ecma/kjs_events.h:
        * khtml/ecma/kjs_proxy.cpp:
        (KJSProxyImpl::createHTMLEventHandler):
        * khtml/ecma/kjs_window.cpp:
        (Window::getJSLazyEventListener):
        * khtml/ecma/kjs_window.h:

2004-07-22  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Hyatt

        Fix for this bug:

        <rdar://problem/3724344> Bolding and unbolding creates extraneous tags

        * khtml/editing/htmlediting_impl.cpp:
        (khtml::ApplyStyleCommandImpl::doApply): Move the start of the selection upstream
        before calling removeStyle. This makes sure we remove all styles that could apply to the 
        selection, and not just ones in from the start position of the selection passed to us. 
        This fixes the bug.
        * khtml/xml/dom_selection.cpp:
        (DOM::Selection::validate): Related fix to "constrain" the selection to be the 
        smallest equivalent range of nodes, in effect making a "canonical" version of the
        selection. While this is not strictly necessary to fix the bug, it is a step I have been 
        wanting to take this step for a long time, and some recent improvements made it 
        possible for me to do now in just two lines of code.

2004-07-22  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Hyatt

        Fix for this bug:
        
        <rdar://problem/3711264> difficult (impossible?) to get cursor in an editable webview containing only tags

        * khtml/rendering/render_container.cpp:
        (RenderContainer::positionForCoordinates): Don't assume you can pass off
        the check to a first child if there is one, since the child can be an element
        we do not want to place the caret in, like a table row with no cells (the case
        in the bug above). So now, we iterate of the renderer's children looking for
        the closest one, but only consider those renderers which either have children
        themselves, or are render block flows or are render inlines. 

2004-07-22  Darin Adler  <darin@apple.com>

        - remove a bunch of now-unused code

        * ForwardingHeaders/qxml.h: Removed.
        * kwq/KWQXmlAttributes.h: Removed.
        * kwq/KWQXmlAttributes.mm: Removed.
        * kwq/KWQXmlDefaultHandler.h: Removed.
        * kwq/KWQXmlDefaultHandler.mm: Removed.
        * kwq/KWQXmlSimpleReader.h: Removed.
        * kwq/KWQXmlSimpleReader.mm: Removed.

2004-07-22  Darin Adler  <darin@apple.com>

        Reviewed by Dave.

        - got rid of QXml classes and changed XML parsing to use libxml directly

        This is the first step toward more direct use of libxml and libxslt.
        Dave is planning to build on this to implement XSLT and to improve our handling
        of XML documents (faster and more feature complete, including DTDs).

        * khtml/html/html_elementimpl.cpp:
        (HTMLElementImpl::createContextualFragment):
        * khtml/html/htmltokenizer.cpp:
        (khtml::HTMLTokenizer::HTMLTokenizer):
        * khtml/html/htmltokenizer.h:
        * khtml/xml/dom_docimpl.cpp:
        (DocumentImpl::createTokenizer):
        (DocumentImpl::open):
        * khtml/xml/dom_docimpl.h:
        (DOM::DocumentImpl::tokenizer):
        * khtml/xml/dom_xmlimpl.cpp:
        (DOM::ProcessingInstructionImpl::checkStyleSheet):
        * khtml/xml/dom_xmlimpl.h:
        * khtml/xml/xml_tokenizer.h:
        * khtml/xml/xml_tokenizer.cpp: Redid this all to use libxml directly.

        * WebCore.pbproj/project.pbxproj: Removed a bunch of files.

2004-07-21  Trey Matteson  <trey@apple.com>

        DHTML dragging should use UTI for MIME-pboard type conversion

        Reviewed by Ken

        * kwq/KWQClipboard.mm:
        (cocoaTypeFromMIMEType):  Use UTI, except for cases we need to hardwire.  Only cut off
        MIME args following a semi-colon for text/plain.
        (MIMETypeFromCocoaType): Use UTI, except for cases we need to hardwire.  
        (KWQClipboard::types):  Filter out ancient NSAsciiPboardType so DHTML never sees this cruft.

2004-07-21  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Trey

        Remove handling of HTML editing key events from WebCore.
        This now happens in WebKit if an event passes all the
        way through WebCore without being handled by the DOM.

        In each case below, function were removed, as their
        only purpose was to handle editing key events.

        * khtml/xml/dom_elementimpl.cpp
        * khtml/xml/dom_elementimpl.h
        * kwq/KWQKHTMLPart.h
        * kwq/KWQKHTMLPart.mm
        * kwq/WebCoreBridge.h

2004-07-21  Ken Kocienda  <kocienda@apple.com>

        Reviewed by John

        * khtml/xml/dom_selection.cpp:
        (DOM::Selection::modifyExtendingRightForward): Add LINE_BOUNDARY case to the switch statement in this
        function. Use the startAndEndLineNodesIncludingNode helper which already existed to get the right
        position.
        (DOM::Selection::modifyMovingRightForward): Ditto
        (DOM::Selection::modifyExtendingLeftBackward): Ditto
        (DOM::Selection::modifyMovingLeftBackward): Ditto
        * khtml/xml/dom_selection.h: 
        (DOM::Selection::): Add LINE_BOUNDARY constant to ETextGranularity enum. This
        specifies a new kind of movement that we need to implement the "move-to beggining/end of line"
        behavior which AppKit binds to cmd+left/right arrow keys.
        * kwq/WebCoreBridge.h: Add WebSelectToLineBoundary constant. This matches

2004-07-20  David Hyatt  <hyatt@apple.com>

        Fix for 3714434, user stylesheet is always parsed in strict mode, when it should honor the document's setting.
        
        Reviewed by john

        * khtml/css/cssstyleselector.cpp:
        (khtml::CSSStyleSelector::CSSStyleSelector):

2004-07-20  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Richard

        * khtml/xml/dom_elementimpl.cpp:
        (ElementImpl::defaultEventHandler): No longer check whether
        the command key is modifying the key event. This check is
        now done elsewhere in the code. See the WebKit checkin that
        added the _web_keyBindingManagerHasBinding method to 
        WebNSEventExtras.

2004-07-20  Chris Blumenberg  <cblu@apple.com>

        Fixed:
        <rdar://problem/3580589> REGRESSION (1.1-1.2): can't open a new window for an image that has not loaded
        <rdar://problem/3612691> Missing image icons (blue ?) lack context menu

        Reviewed by john.

        * kwq/WebCoreBridge.mm:
        (-[WebCoreBridge elementAtPoint:]): if there is no image, put the image URL on the element anyway

2004-07-20  David Hyatt  <hyatt@apple.com>

        Fix for alt text not working on images.
        
        Reviewed by john

        * khtml/rendering/render_image.cpp:
        (RenderImage::RenderImage):
        (RenderImage::setPixmap):
        (RenderImage::paint):
        (RenderImage::calcReplacedWidth):
        (RenderImage::calcReplacedHeight):

2004-07-20  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Dave.

        <rdar://problem/3720111> REGRESSION(125-152): map fails to load depending on banner ad; weather.com (works in IE and Firefox)
        
        * khtml/html/htmltokenizer.cpp:
        (khtml::HTMLTokenizer::begin): Initialize currentPrependingSrc.
        (khtml::HTMLTokenizer::scriptHandler): Store a pointer to the currently
        active prependingSrc variable on the stack, the better to handle weird cases
        of scripts writing out additional script tags and so forth.
        (khtml::HTMLTokenizer::write): Deal with the above.
        * khtml/html/htmltokenizer.h: Declare new member field.

2004-07-20  David Hyatt  <hyatt@apple.com>

        Fix for layer positioning error that occurs when absolute positioned blocks are inside static positioned overflow:auto
        elements.
        
        Reviewed by john

        * khtml/rendering/render_layer.cpp:
        (RenderLayer::updateLayerPosition):

2004-07-20  Trey Matteson  <trey@apple.com>

        3705624 REGRESSION: can't rearrange photos on homepage.mac.com album

        A subtle problem stemming from some interaction between focus handing and
        event propagation.  We previously made it so when a focus shift fails we
        don't propagate the event to KHTML.  The fix is to get rid of the inverse
        effect, so now even if a focus shift succeeds, we don't propagate to KHTML if
        the page already canceled default behavior.

        In this specific case, the page is cancelling default behavior since it
        does its own dragging, and the bug was our new system dragging was
        kicking in.

        Reviewed by Ken.

        * khtml/khtmlview.cpp:
        (KHTMLView::dispatchMouseEvent):  What he said.

2004-07-20  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Darin

        Fix for this bug:
        
        <rdar://problem/3695240> pasting plain text with newlines in it turns them into spaces

        * kwq/WebCoreBridge.mm:
        (-[WebCoreBridge documentFragmentWithText:]): Refine this function to be smart about converting
        line endings into BR elements.

2004-07-20  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Hyatt

        Fix for this bug:
        
        <rdar://problem/3707505> HTMLCompose: blinking cursor in both an address text field and the message body

        * khtml/khtml_part.cpp:
        (KHTMLPart::setCaretVisible): This method renamed from setSelectionVisible. Since it really only operates
        on the caret, this seems like a better name.
        * khtml/khtml_part.h:
        * khtml/khtmlview.cpp:
        (KHTMLView::focusInEvent): Call setCaretVisible(true)
        (KHTMLView::focusOutEvent): Call setCaretVisible(false)
        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::setShowsFirstResponder): Name change: setSelectionVisible becomes setCaretVisible.
        * kwq/WebCoreBridge.h: Expose setCaretVisible as a bridge method.
        * kwq/WebCoreBridge.mm:
        (-[WebCoreBridge setCaretVisible:]): Ditto.

2004-07-19  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Trey.

        <rdar://problem/3721428> REGRESSION (125.8-146): external javascript statements produce extra garbage character (sina.com)
        
        * khtml/misc/stringit.h:
        (khtml::TokenizerSubstring::TokenizerSubstring): For the apple branch, use the new
        stableUnicode() method to get the unicode pointer.
        * kwq/KWQString.h:
        * kwq/KWQString.mm:
        (QString::detachIfInternal): Reorganize this to be a bit less wacky about refcounts.
        It does not leave around a zombie internal data handle but rather destroys it right away,
        and leaves the object pointing to the new handle (which it can then deref). This makes
        the code more clear.
        (QString::~QString): Simplify.
        (QString::stableUnicode): New method that detaches a copy of the KWQStringData if it
        is internal to a string besides this one. This guarantees that if you get the unicode()
        pointer, it won't go bad so long as this string is still alive.

2004-07-19  David Hyatt  <hyatt@apple.com>

        Fix for 3715117, crash from a bug in removeChildren.  Clean up node removal and fix an n-squared removal
        bug.  Also clean up checks in removeChild and in the dispatch of removedFromDocument mutation events to
        make node removal more efficient.
        
        Reviewed by kocienda

        * khtml/xml/dom_nodeimpl.cpp:

2004-07-19  John Sullivan  <sullivan@apple.com>

        Reviewed by Trey.
        
        - fixed <rdar://problem/3141150> can't undo in form textarea fields

        * kwq/KWQTextArea.mm:
        (-[KWQTextArea _createTextView]):
        call setAllowsUndo:YES. My, wasn't this easy?

2004-07-19  David Hyatt  <hyatt@apple.com>

        Merge @import fix for CSS1 test suite from Stephan Kulow.
        
        Reviewed by hyatt

        * khtml/css/parser.cpp:
        * khtml/css/parser.y:
        * layout-tests/css1/basic/containment-expected.txt:

2004-07-19  David Hyatt  <hyatt@apple.com>

        Fix for 3718697, crash clicking on JS tab at alaskaair.com.
        
        Reviewed by kocienda

        * khtml/rendering/render_flow.cpp:
        (RenderFlow::detach):
        (RenderFlow::dirtyLinesFromChangedChild):
        * khtml/rendering/render_flow.h:
        * khtml/rendering/render_object.cpp:
        (RenderObject::dirtyLinesFromChangedChild):
        * khtml/rendering/render_object.h:

2004-07-16  David Hyatt  <hyatt@apple.com>

        Fix for 3726471, need to ensure that it's safe to do updateRendering from within layout, since isContentEditable
        relies on it.  The fix is to just recalcStyle up front before beginning the layout, so that all of our renderobjects
        will be up-to-date.
        
        Reviewed by john

        * khtml/khtmlview.cpp:
        (KHTMLView::layout):

2004-07-16  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Hyatt

        Fix for this bug:
        
        <rdar://problem/3722153> Random crash while typing (DOM::NodeImpl::parentNode (this=0x0))

        * khtml/editing/htmlediting_impl.cpp:
        (khtml::DeleteSelectionCommandImpl::doApply): This check merely makes the code
        more robust. I do not understand how Grant ended up with null start or end 
        blocks in the case of his bug report, but by adding two null checks, the code is 
        now hardened a bit in case it happens again.

2004-07-16  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Hyatt

        Fix for these bugs:

        <rdar://problem/3730785> Crash when arrow navigation goes to empty table cell
        <rdar://problem/3730790> Caret does not blink when placed in empty table cell

        * khtml/rendering/render_block.cpp:
        (khtml::RenderBlock::paintObject): Caret node's renderer might be this block, so
        don't block the painting of the caret in this case. This was the case in 3730790.
        We had the right geometry for the caret but blocked its painting. 
        * khtml/xml/dom_position.cpp:
        (DOM::Position::previousLinePosition): Rework the logic here. This function asserted
        that the new position we would navigate to had a line box, but empty table cells, for
        instance will not. So, deal with this situation as well. The end result is a 
        function that worked like it did before in cases where the previous line position has
        a line box, but now also will allow navigation to work when it does not.
        (DOM::Position::nextLinePosition): Ditto.

2004-07-16  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Hyatt

        Fix for this bug:
        
        <rdar://problem/3687216> editable inline causes crash when focused

        * khtml/xml/dom_docimpl.cpp:
        (DocumentImpl::relinquishesEditingFocus): Add null check for rootEditableElement.
        (DocumentImpl::acceptsEditingFocus): Ditto.

2004-07-16  David Hyatt  <hyatt@apple.com>

        Fix for 3709337, crash when using display: compact.  Need to make sure that the isCompact() check in bidi.cpp is
        also checking for RenderBlocks, since otherwise you match text nodes (and that's just wrong).
        
        Reviewed by kocienda

        * khtml/rendering/bidi.cpp:
        (khtml::RenderBlock::layoutInlineChildren):

=== Safari-152 ===

2004-07-15  David Hyatt  <hyatt@apple.com>

        Fix for 3625611, images offset on first visit to page.  The method for determining the last clean line was
        flawed and would incorrectly miss dirty lines that occurred after the first clean line.
        
        Reviewed by john

        * khtml/rendering/bidi.cpp:
        (khtml::RenderBlock::determineEndPosition):

2004-07-15  Ken Kocienda  <kocienda@apple.com>

        Reviewed by John

        Fix for this bug:
        <rdar://problem/3587601> reproducible assertion failure in Blot deleting text, then image at Yahoo.com
        
        * khtml/editing/htmlediting_impl.cpp:
        (khtml::DeleteCollapsibleWhitespaceCommandImpl::deleteWhitespace): Relax assertion
        that nodes are text nodes during the walk of eligible nodes that could possibly be
        deleted. Change the assertion to an conditional check for text nodes. We may have 
        unrendered nodes in the mix here (as is the case in the bug where the assertion
        failed on an AREA element), and the conditional check is sufficient to
        skip them while leaving the delete logic we want unperturbed. 

2004-07-14  Ken Kocienda  <kocienda@apple.com>

        Reviewed by John

        * khtml/rendering/render_block.cpp:
        (khtml::RenderBlock::paintObject): Add null check on DOM node before
        asking if it is contentEditable.

2004-07-14  David Hyatt  <hyatt@apple.com>

        Fix for 3595073, setting innerHTML on a <table> should work.
        
        Reviewed by rjw

        * khtml/html/html_elementimpl.cpp:
        (HTMLElementImpl::createContextualFragment):

2004-07-14  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Hyatt
        
        Updated these layout tests as a result of the last patch.
        These results are better.

        * layout-tests/editing/execCommand/selectAll-expected.txt
        * layout-tests/editing/selection/extend-by-character-006-expected.txt
        * layout-tests/editing/selection/unrendered-001-expected.txt
        * layout-tests/editing/selection/unrendered-003-expected.txt

2004-07-14  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Hyatt

        * khtml/rendering/render_block.cpp: 
        (khtml::RenderBlock::paintObject): Change back to start() from caretPosition().
        Since the policy is now to move the selection to rendered content, if possible,
        when the selection is set, there is no longer any reason to have the additional
        caretPosition() function to store where the caret should be drawn.
        * khtml/xml/dom_docimpl.cpp:
        (DocumentImpl::updateSelection): No longer any need to call closestRenderedPosition here.
        This is done in Selection::validate.
        * khtml/xml/dom_position.cpp:
        (DOM::Position::closestRenderedPosition): Improved algorithm. Now much simpler.
        * khtml/xml/dom_selection.cpp:
        (DOM::Selection::Selection): caretPosition() and m_caretPosition now obsolete.
        (DOM::Selection::init): Ditto.
        (DOM::Selection::modifyExtendingRightForward): No longer any need to call 
        closestRenderedPosition here. This is done in Selection::validate.
        (DOM::Selection::modifyMovingRightForward): Ditto.
        (DOM::Selection::modifyExtendingLeftBackward): Ditto.
        (DOM::Selection::modifyMovingLeftBackward): Ditto.
        (DOM::Selection::layoutCaret): Ditto.
        (DOM::Selection::validate): Add code to move the selection to rendered content if possible.
        * khtml/xml/dom_selection.h: caretPosition() and m_caretPosition now obsolete.

2004-07-14  David Hyatt  <hyatt@apple.com>

        Fix for 3716082, assert when you dynamically remove float or position styles.
        
        Reviewed by kevin

        * khtml/rendering/render_object.cpp:
        (RenderObject::setStyle):

2004-07-14  David Hyatt  <hyatt@apple.com>

        Fix for various table regressions (malumovies.com and cityofheroes.gameamp.com) involving bungling of
        percentage heights.
        
        Reviewed by kocienda

        * khtml/rendering/render_box.cpp:
        (RenderBox::calcPercentageHeight):
        (RenderBox::availableHeightUsing):
        * khtml/rendering/render_table.cpp:
        (RenderTable::layout):
        (RenderTableSection::calcRowHeight):
        (RenderTableSection::layoutRows):
        (RenderTableCell::updateFromElement):
        * khtml/rendering/render_table.h:

2004-07-14  Ken Kocienda  <kocienda@apple.com>

        Reviewed by John

        * khtml/xml/dom_position.cpp:
        (DOM::Position::equivalentDeepPosition): New helper function to improve selection handling.
        Soon, I plan to land some changes that use this new function.
        * khtml/xml/dom_position.h:

2004-07-14  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Vicki

        * WebCore-combined.exp: Added QString.at symbol to tests exports to facilitate debugging
        data formatting change that is to come.
        * WebCore-tests.exp:
        
2004-07-13  John Sullivan  <sullivan@apple.com>

        Reviewed by Ken.
        
        - fixed <rdar://problem/3705500> REGRESSION (125.8-146): Cmd-E on an 
        HTML page puts a trailing space on Find pasteboard

        * khtml/misc/khtml_text_operations.cpp:
        (khtml::TextIterator::handleTextBox):
        In the case where a subrun doesn't extend to the end of the text box,
        the well-commented code was supposed to return from this routine, but
        the wrong variable was being compared. This would send us through another
        pass of the loop, which would end up worrying about collapsed space at the
        end of the current box, even though we only cared about the beginning of the
        current box.

2004-07-12  Richard Williamson   <rjw@apple.com>

        Fixed 3717982.  Implemented navigator.language!

        Reviewed by John.

        * khtml/ecma/kjs_navigator.cpp:
        (Navigator::getValueProperty):

2004-07-12  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Hyatt

        * khtml/xml/dom_position.cpp:
        (DOM::Position::inRenderedContent): I broke selection drawing with my last patch. This 
        function is now used in selection drawing, and it was making an unneeded check to see
        that content was editable to answer whether or not is was rendered. Bad. Fixed.

2004-07-12  David Hyatt  <hyatt@apple.com>

        Fix for 3621138, crash on hrweb.apple.com.  Make sure that in the case where objects get pulled up from merging
        blocks that we just delete all line boxes.
        
        Reviewed by kocienda

        * khtml/rendering/render_block.cpp:
        (khtml::RenderBlock::removeChild):

2004-07-10  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by John.

        <rdar://problem/3706080>: (REGRESSION (125.8-147u): Nested <ul> do not display bullets)
        <rdar://problem/3676376>: (Second level bullets not printed when printing Xcode release notes)
        
        * kwq/KWQPainter.h:
        * kwq/KWQPainter.mm:
        (CGColorFromNSColor):
        (QPainter::drawEllipse):
        (QPainter::setShadow):

2004-07-12  Ken Kocienda  <kocienda@apple.com>

        Reviewed by me
        
        Added new editing-related layout tests

        * layout-tests/editing/deleting/delete-after-span-ws-001-expected.txt: Added.
        * layout-tests/editing/deleting/delete-after-span-ws-001.html: Added.
        * layout-tests/editing/deleting/delete-after-span-ws-002-expected.txt: Added.
        * layout-tests/editing/deleting/delete-after-span-ws-002.html: Added.
        * layout-tests/editing/deleting/delete-after-span-ws-003-expected.txt: Added.
        * layout-tests/editing/deleting/delete-after-span-ws-003.html: Added.
        * layout-tests/editing/selection/unrendered-001-expected.txt: Added.
        * layout-tests/editing/selection/unrendered-001.html: Added.
        * layout-tests/editing/selection/unrendered-002-expected.txt: Added.
        * layout-tests/editing/selection/unrendered-002.html: Added.
        * layout-tests/editing/selection/unrendered-003-expected.txt: Added.
        * layout-tests/editing/selection/unrendered-003.html: Added.
        * layout-tests/editing/selection/unrendered-004-expected.txt: Added.
        * layout-tests/editing/selection/unrendered-004.html: Added.
        * layout-tests/editing/selection/unrendered-005-expected.txt: Added.
        * layout-tests/editing/selection/unrendered-005.html: Added.
        * layout-tests/traversal/node-iterator-008-expected.txt: Added.
        * layout-tests/traversal/node-iterator-008.html: Added.
        * layout-tests/traversal/tree-walker-005-expected.txt: Added.
        * layout-tests/traversal/tree-walker-005.html: Added.

2004-07-12  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Hyatt

        Fixes for these bugs:
        
        <rdar://problem/3723359> Extending then "unextending" selection with arrow keys should draw caret but doesn't
        <rdar://problem/3724626> White-space deletion code deletes wrong character when space follows span
        
        * khtml/editing/htmlediting_impl.cpp:
        (khtml::DeleteSelectionCommandImpl::doApply): Modify special-case white-space deletion code so it runs
        only in the special case. It was throwing its net too widely, catching the case described in 3724626.
        By tightening up the special-case white-space deletion, and allowing the more general-purpose code to run,
        the bug is fixed.
        * khtml/xml/dom_docimpl.cpp:
        (DocumentImpl::updateSelection): Use recently-added closestRenderedPosition helper in Position class
        to figure out the start and end positions for selection drawing.
        * khtml/xml/dom_position.cpp:
        (DOM::Position::equivalentUpstreamPosition): Added code to handle white-space that causes line breaks.
        (DOM::Position::equivalentDownstreamPosition): Ditto.
        (DOM::Position::closestRenderedPosition): Trap empty selections at function entry, return *this.
        (DOM::Position::isFirstRenderedPositionOnLine): Can't be first rendered position on line if not rendered.
        Add check for this.
        (DOM::Position::isLastRenderedPositionOnLine): Ditto, but s/first/last/
        * khtml/xml/dom_selection.cpp:
        (DOM::Selection::validate): A selection is in caret state if the start and end are equal *or* equivalent.
        The equivalence case is new, and fixes 3723359.

2004-07-09  Kevin Decker  <kdecker@apple.com>

        Reviewed by Hyatt.
        
        fixes the width:auto problem in
        <rdar://problem/3698344> REGRESSION (143?-144): macrumors.com tabs are compressed and illegible
        
        * khtml/css/cssstyleselector.cpp:
        (khtml::CSSStyleSelector::applyProperty):

2004-07-09  Ken Kocienda  <kocienda@apple.com>

        Reviewed by me
        
        Updated some layout test results.

        * layout-tests/editing/deleting/delete-image-004-expected.txt:
        * layout-tests/editing/selection/extend-by-character-006-expected.txt:

2004-07-09  Chris Blumenberg  <cblu@apple.com>

        Allowed my change for 3715785 to compile on Jaguar.

        Reviewed by kocienda.

        * kwq/WebCoreBridge.h:
        * kwq/WebCoreBridge.mm:
        (-[WebCoreBridge domain]): new, allows access to the domain without using the DOM API which doesn't exist on Jaguar

2004-07-09  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Hyatt
        
        Some improvements to fix:
        
        <rdar://problem/3723111> Caret not drawn when selection set to unrendered content

        * khtml/editing/htmlediting_impl.cpp:
        (khtml::TypingCommandImpl::issueCommandForDeleteKey): Adjust selection to delete if
        selected position is not rendered.
        * khtml/rendering/render_block.cpp:
        (khtml::RenderBlock::paintObject): Use new caretPosition() function on Selection to
        figure out whether to paint.
        * khtml/xml/dom_position.cpp:
        (DOM::Position::previousCharacterPosition): Now correctly deals with a start
        position that is not rendered.
        (DOM::Position::nextCharacterPosition): Ditto.
        (DOM::Position::closestRenderedPosition): New helper.
        * khtml/xml/dom_position.h:
        (DOM::): Moved in EAffinity from Selection header. Now used in closestRenderedPosition function.
        * khtml/xml/dom_selection.cpp:
        (DOM::Selection::Selection): Added new m_caretPosition member. This is the position of the caret
        after a caret layout. This may be different from start or end if start and end are not rendered. 
        (DOM::Selection::init):
        (DOM::Selection::modifyExtendingRightForward): New helper to clean up modify() and make it more readble.
        (DOM::Selection::modifyMovingRightForward): Ditto.
        (DOM::Selection::modifyExtendingLeftBackward): Ditto.
        (DOM::Selection::modifyMovingLeftBackward): Ditto.
        (DOM::Selection::modify): Use new helpers to make this more readble.
        (DOM::Selection::layoutCaret): Uses new closestRenderedPosition helper to place the caret if in unrendered
        content.
        (DOM::Selection::paintCaret): Remove moveToRenderedContent. obsolete.
        * khtml/xml/dom_selection.h:
        (DOM::Selection::caretPosition): New accessor.
        * kwq/WebCoreBridge.mm:
        (-[WebCoreBridge setSelectedDOMRange:affinity:]): EAffinity no longer a member enum of Selection class.

2004-07-08  David Hyatt  <hyatt@apple.com>

        Fix for the table layout test that failed because of a change in how innerText worked.  We need to do
        updateLayout now when using innerText, since the method has been changed to use line boxes in the render tree
        that might otherwise be out of date.
        
        Reviewed by kocienda

        * khtml/html/html_elementimpl.cpp:
        (HTMLElementImpl::innerText):

2004-07-08  John Sullivan  <sullivan@apple.com>

        Reviewed by Ken.
        
        - fixed <rdar://problem/3691569> REGRESSION (142): cmd-shift-clicking on a link 
        now also extends selection (even if there wasn't one before)

        * khtml/khtml_part.cpp:
        (KHTMLPart::handleMousePressEventSingleClick):
        if there's a URL associated with the event, don't extend the selection

2004-07-08  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Vicki

        Added some helper functions which provide strings to display in the 
        Xcode debugger's variable inspector window. These functions are called
        from the LabyrinthDataFormatter debugger plugin I just checked in to 
        the Labyrinth/Tools directory.
        
        Note that these functions are compiled in on Development builds only.

        * WebCore-combined.exp:
        * WebCore-tests.exp: Export all the formatForDebugger symbols so the
        debugger program can link with them.
        * khtml/xml/dom2_rangeimpl.cpp:
        (DOM::RangeImpl::formatForDebugger):
        * khtml/xml/dom2_rangeimpl.h:
        * khtml/xml/dom_elementimpl.cpp:
        (ElementImpl::formatForDebugger):
        * khtml/xml/dom_elementimpl.h:
        * khtml/xml/dom_nodeimpl.cpp:
        * khtml/xml/dom_nodeimpl.h:
        * khtml/xml/dom_position.cpp:
        (DOM::Position::formatForDebugger):
        * khtml/xml/dom_position.h:
        * khtml/xml/dom_selection.cpp:
        (DOM::Selection::formatForDebugger):
        * khtml/xml/dom_selection.h:
        * khtml/xml/dom_textimpl.cpp:
        (TextImpl::formatForDebugger):
        * khtml/xml/dom_textimpl.h:

2004-07-08  John Sullivan  <sullivan@apple.com>

        Reviewed by Vicki.
        
        - fixed <rdar://problem/3721544> crash increasing font size; 
        entrezeroetun.com (works in IE and Firefox)

        * khtml/rendering/render_block.cpp:
        (khtml::RenderBlock::updateFirstLetter):
        Check for nil originalString() before dereffing

2004-07-08  David Hyatt  <hyatt@apple.com>

        Fix for 3721453, CSS3 initial property caused crashes because the macros were not written correctly.
        
        Reviewed by john

        * khtml/css/cssstyleselector.cpp:

=== Safari-151 ===

2004-07-07  David Hyatt  <hyatt@apple.com>

        Fix for 3712133, crash from first-line pseudo-style use.
        
        Reviewed by kocienda

        * khtml/css/cssstyleselector.cpp:
        (khtml::CSSStyleSelector::styleForElement):
        * khtml/css/cssstyleselector.h:
        * khtml/rendering/render_object.cpp:
        (RenderObject::getPseudoStyle):

2004-07-07  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Hyatt

        Fix for this bug:
        
        <rdar://problem/3716479> calling setInnerHTML during a webViewDidChange delegate call causes a crash
        
        The fix involves some rearrangement of code in TypingCommand and TypingCommandImpl.
        Formerly, new TypingCommands would apply themselves (which was a no-op) and then
        do their action in some code a way different than other commands. This type of command
        application is different than for all other commands since TypingCommands can be coalesced.
        The crash occurred as a result of the "no-op" TypingCommand having the unconsidered
        consequence of causing editing delegate notifications to be sent before the command 
        has actually run. This change takes a small step towards making TypingCommandImpl function like
        other commands, where the command work is done in doApply. This makes the notification
        happen in the right order.

        * khtml/editing/htmlediting.cpp:
        (khtml::TypingCommand::TypingCommand):
        (khtml::TypingCommand::insertText):
        (khtml::TypingCommand::insertNewline):
        (khtml::TypingCommand::deleteKeyPressed):
        * khtml/editing/htmlediting.h:
        (khtml::TypingCommand::):
        * khtml/editing/htmlediting_impl.cpp:
        (khtml::TypingCommandImpl::TypingCommandImpl):
        (khtml::TypingCommandImpl::doApply):
        * khtml/editing/htmlediting_impl.h:

2004-07-06  Ken Kocienda  <kocienda@apple.com>

        Reviewed by me

        * khtml/html/html_tableimpl.cpp:
        (HTMLTableElementImpl::addChild): Added a better comment in the
        code I just checked in a few minutes ago.

2004-07-06  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Hyatt
        
        Fix for this bug:
        
        <rdar://problem/3672377> assertion failure in AppendNodeCommandImpl::doApply 
        due to non-0 exception code trying to insert a DIV markup string
        
        The solution was to revert to the code that was rolled out, and removing
        the child checks from NodeImpl::checkAddChild. However, this time, I added
        code very similar to this check into the code that runs while HTML is 
        being parsed to build up tables. This code relies on child-add failure
        to ensure the proper construction of well-formed tables (as gross as that
        sounds), so the check needs to be retained there. No other code seems to
        be so affected. Layout tests are unchanged by this patch.

        * khtml/html/html_tableimpl.cpp:
        (HTMLTableElementImpl::addChild):
        * khtml/xml/dom_nodeimpl.cpp:
        (NodeImpl::checkAddChild):

2004-07-06  Ken Kocienda  <kocienda@apple.com>

        Reviewed by John
        
        Simple change. I switched the arguments of the appendNode helper function
        and the AppendNodeCommand and AppendNodeCommandImpl classes. The node to 
        insert now comes before the parent node in the argument list. I did this
        to make this function match the convention of others in the HTML editing code.
        This was the only one that was "different" in the way that it ordered arguments.
        As a result, I was always looking to see that I was passing things in the right
        order.
        
        * khtml/editing/htmlediting.cpp:
        (khtml::AppendNodeCommand::AppendNodeCommand):
        (khtml::AppendNodeCommand::appendChild):
        (khtml::AppendNodeCommand::parentNode):
        * khtml/editing/htmlediting.h:
        * khtml/editing/htmlediting_impl.cpp:
        (khtml::CompositeEditCommandImpl::insertNodeAfter):
        (khtml::CompositeEditCommandImpl::insertNodeAt):
        (khtml::CompositeEditCommandImpl::appendNode):
        (khtml::AppendNodeCommandImpl::AppendNodeCommandImpl):
        (khtml::AppendNodeCommandImpl::~AppendNodeCommandImpl):
        (khtml::AppendNodeCommandImpl::doApply):
        (khtml::AppendNodeCommandImpl::doUnapply):
        (khtml::ApplyStyleCommandImpl::surroundNodeRangeWithElement):
        (khtml::DeleteSelectionCommandImpl::doApply):
        (khtml::InputNewlineCommandImpl::insertNodeAfterPosition):
        (khtml::InputNewlineCommandImpl::insertNodeBeforePosition):
        (khtml::InputTextCommandImpl::prepareForTextInsertion):
        * khtml/editing/htmlediting_impl.h:
        (khtml::AppendNodeCommandImpl::parentNode):

2004-07-06  Ken Kocienda  <kocienda@apple.com>

        Reviewed by Trey

        Fixed several problems with traversal classes. For one, NodeIterators treat
        FILTER_REJECT and FILTER_SKIP the same, since it treats the DOM tree as a
        flat collection of nodes free of hierarchy. The code before this change did
        not do this correctly. It sure pays to go back and read the specs. :)
        
        Also, the code to traverse from node to node when filters were applied was
        not working correctly. My first attemmpt to implement this was just plain
        buggy, as I discovered when I tried to write tests for my WWDC talk. I have
        settled on an implementation which is much simpler and worked for all the
        tests I threw at it. 

        * khtml/xml/dom2_traversalimpl.cpp:
        (DOM::NodeIteratorImpl::findNextNode):
        (DOM::NodeIteratorImpl::nextNode):
        (DOM::NodeIteratorImpl::findPreviousNode):
        (DOM::NodeIteratorImpl::previousNode):
        (DOM::TreeWalkerImpl::parentNode):
        (DOM::TreeWalkerImpl::firstChild):
        (DOM::TreeWalkerImpl::lastChild):
        (DOM::TreeWalkerImpl::previousSibling):
        (DOM::TreeWalkerImpl::nextSibling):
        (DOM::TreeWalkerImpl::previousNode):
        (DOM::TreeWalkerImpl::nextNode):
        (DOM::TreeWalkerImpl::ancestorRejected):
        * khtml/xml/dom2_traversalimpl.h:

2004-07-06  Vicki Murley  <vicki@apple.com>

        Reviewed by kocienda.

        - added backColorCommand, foreColorCommand, fontNameCommand, 
        and fontSizeCommand
  
        * layout-tests/editing/editing.js:

2004-07-06  Trey Matteson  <trey@apple.com>

        3716053 - www.theage.com.au has extra back/forward items due to ads

        This turned out to be easily fixed by generalizing the fix to 3438441.  We prevent
        addition to the b/f list not just during an onload event, but during any non-user
        gesture, which includes top level script executing.

        Reviewed by Richard.

        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::openURL):  Only real change - prevent adding to b/f list if not
        a user gesture.
        (KWQKHTMLPart::openURLRequest):  Rename "onLoadEvent" to "userGesture", swap sense
        (KWQKHTMLPart::submitForm):  Ditto
        (KWQKHTMLPart::urlSelected): Ditto
        * kwq/KWQKHTMLPartBrowserExtension.mm:
        (KHTMLPartBrowserExtension::createNewWindow):  Ditto
        * kwq/WebCoreBridge.h:

2004-07-02  Darin Adler  <darin@apple.com>

        Reviewed by Dave.

        - fixed half of <rdar://problem/3709244> utf-8 meta tag not parsed when page title contains angle brackets or if </meta> tag used

        * khtml/misc/decoder.cpp: (Decoder::decode): Allow </meta> tags without deciding we
        are done with the header.

2004-06-30  Trey Matteson  <trey@apple.com>

        Dragging within a web view should be allowed to start when the window isn't key.

        A few months ago, Chris made this work, but it relied on the fact that all dragging
        was done in WebKit.  When WebCore got involved in dragging, it was broken.  Now we
        have a new scheme that gets it working again that properly involves WebCore.

        The general idea is that when AK asks us whether to accept the first mouse and do
        "delayed window ordering", we must consult WC to see if we might start a drag.  In
        addition, instead of these drags in non-active windows being started as a special
        case in WK, they go through the normal WK-WC drag machinery.

        Reviewed by John.

        * khtml/khtml_part.cpp:
        (KHTMLPart::shouldDragAutoNode):  New x,y args.
        * khtml/khtml_part.h:
        * khtml/rendering/render_object.cpp:
        (RenderObject::draggableNode):  Pass through new x,y args.
        * khtml/rendering/render_object.h:
        * kwq/KWQKHTMLPart.h:
        (KWQKHTMLPart::setActivationEventNumber):  New setter.
        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::KWQKHTMLPart):  Init new ivar.
        (KWQKHTMLPart::eventMayStartDrag):  New routine that checks if we might start
        a drag in response to a mouseDown.
        (KWQKHTMLPart::khtmlMouseMoveEvent):  Pass x,y to the routine that finds a draggable
        node.  This eventually gets back up to WK's _mayStartDragAtEventLocation:.
        Delay requirement when dragging the selection now implemented here.
        (KWQKHTMLPart::khtmlMouseReleaseEvent):  Must avoid changing the selection if we
        wind up here as part of the first click in a window (because we started handling
        the click to possible start a drag, but that never came through).
        (KWQKHTMLPart::mouseDown): Save away event timestamp.
        (KWQKHTMLPart::shouldDragAutoNode):  Pass location up to WK instead of the
        most recent event we stashed.
        * kwq/WebCoreBridge.h:
        * kwq/WebCoreBridge.mm:
        (-[WebCoreBridge setActivationEventNumber:]):  Trivial glue.
        (-[WebCoreBridge eventMayStartDrag:]):  Ditto.

2004-06-29  Trey Matteson  <trey@apple.com>

        Need to tighten up JS error checking for requesting drag props
        in the wrong cases.

        Reviewed by John.

        * khtml/ecma/kjs_events.cpp:
        (Clipboard::getValueProperty):  Assert if someone somehow set
        dropEffect or effectAllowed and it's a copy/paste clipboard
        instead of a dragging clipboard.
        (Clipboard::putValue):  Don't let anyone set dropEffect or
        effectAllowed on a copy/paste clipboard.
        (ClipboardProtoFunc::tryCall):  Disallow setting dragImage on
        a copy/paste clipboard.

2004-06-29  Trey Matteson  <trey@apple.com>

        DHTML dragging - source should have access to the operation chosen
        by the destination.

        Reviewed by John

        * kwq/KWQKHTMLPart.h:
        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::dragSourceEndedAt):  Set the destination's operation
        on the clipboard.
        * kwq/WebCoreBridge.mm:
        (-[WebCoreBridge dragExitedWithDraggingInfo:]):  For completeness
        we set the source op for the ondragexit event.
        (-[WebCoreBridge concludeDragForDraggingInfo:]):  Ditto for the
        drop event.
        (-[WebCoreBridge dragSourceEndedAt:operation:]):  Pass through of operation.

2004-06-24  Trey Matteson  <trey@apple.com>

        3710422 - REGRESSION: Safari crashes trying to send onbeforecut event at about://blank

        Simple fix - don't try to send the event to the body element if we have no body element.

        Reviewed by John

        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::dispatchCPPEvent):

2004-06-24  Trey Matteson  <trey@apple.com>

        3704950 drag image in DB ConfigBar has horizontal graphics turd WebCore JavaScript

        When we generate a drag image (or a selection image too, for that matter) we
        translate the CTM using a CG call.  Later, WebImageRenderer adjusts the pattern
        phase based on the CTM of the focused view, which doesn't include our translate.
        So we must inform WebKit about the additional phase adjustment.

        Reviewed by Richard

        * kwq/KWQKHTMLPart.mm:  Tell WebKit about the phase adjustment.
        (KWQKHTMLPart::imageFromRect):
        * kwq/WebCoreGraphicsBridge.h:
        * kwq/WebCoreGraphicsBridge.m:
        (-[WebCoreGraphicsBridge setAdditionalPatternPhase:]):  New routine to receive
        the phase adjustment.

2004-06-24  Trey Matteson  <trey@apple.com>

        3679986 - screenX and screenY are flipped and relative to the bottom left of the WebView, rather than the screen
        3699510 - synthesized click events have bogus screen coords

        Reviewed by Darin.

        * khtml/khtmlview.cpp:
        (KHTMLView::dispatchDragEvent):  Generate screen coords for drag events using new func.
        (KHTMLView::dispatchMouseEvent):   Ditto.
        * khtml/khtmlview.h:
        * khtml/xml/dom_nodeimpl.cpp:
        (NodeImpl::dispatchMouseEvent):  Comment tricky semantics.
        Calc screen coords in apple-specific way.
        * kwq/KWQKHTMLView.mm:
        (KHTMLView::viewportToGlobal):  Passthrough to window widget.
        * kwq/KWQScrollView.mm:
        * kwq/KWQWindowWidget.h:
        * kwq/KWQWindowWidget.mm:
        (KWQWindowWidget::mapToGlobal):  Call former code factored to new method below.
        (KWQWindowWidget::viewportToGlobal):  New method to convert "viewport" (which for us
        really means NSWindow coords) to screen coords.

2004-06-24  Trey Matteson  <trey@apple.com>

        3693420 - onbeforecut and onbeforepaste need real implementaion

        Reviewed by Chris.

        * kwq/KWQKHTMLPart.h:
        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::mayCut):  Dispatch event to DHTML.
        (KWQKHTMLPart::mayCopy):  Ditto
        (KWQKHTMLPart::mayPaste):  Ditto
        (KWQKHTMLPart::tryCut):  No more need to send fake onbefore event
        (KWQKHTMLPart::tryCopy):  Ditto
        (KWQKHTMLPart::tryPaste): Ditto
        * kwq/WebCoreBridge.h:
        * kwq/WebCoreBridge.mm:
        (-[WebCoreBridge mayDHTMLCut]):  Standard glue
        (-[WebCoreBridge mayDHTMLCopy]):  Ditto
        (-[WebCoreBridge mayDHTMLPaste]):  Ditto

2004-06-24  Darin Adler  <darin@apple.com>

        Reviewed by John.

        - fixed <rdar://problem/3709385> Find on page doesn't find a string at the very end of the file

        * khtml/misc/khtml_text_operations.cpp: (khtml::findPlainText): Rearrange loop to avoid an early
        exit once we have all the characters we need, but are at the end of the range we are searching.

        - fixed <rdar://problem/3102271>: (text areas have scroll bars even when they don't need them)
        - fixed <rdar://problem/3665430>: (horizontal scroll bar of text area does not show, even when text is wide in "no wrap" mode)

        * kwq/KWQTextArea.mm:
        (-[KWQTextArea _configureTextViewForWordWrapMode]): Added. Helper method that sets up the
        view for a new word wrap mode.
        (-[KWQTextArea _createTextView]): Moved much of the code inside _configureTextViewForWordWrapMode.
        (-[KWQTextArea _frameSizeChanged]): Added. Method shared by setFrame: and initWithFrame: to
        avoid duplicate code that was there before. The old code also had redundant code to update
        the text container size, but NSText handles that automatically.
        (-[KWQTextArea initWithFrame:]): Set wrap to YES by default, which is the key to fixing bug 3665430.
        Call setAutohidesScrollers:YES, which fixes bug 3102271. Also call the new _frameSizeChanged method.
        (-[KWQTextArea setWordWrap:]): Call _configureTextViewForWordWrapMode instead of trying
        to do the work here. The old version did both too little and too much.
        (-[KWQTextArea setFrame:]): Call _frameSizeChanged instead of trying to do the work here.
        The old version did both too little and too much.

2004-06-24  John Sullivan  <sullivan@apple.com>

        Darin made this change on my machine; I reviewed it.
        
        - fixed <rdar://problem/3698333> Find on page doesn't find a particular string 
        with a newline in the source

        * khtml/misc/khtml_text_operations.cpp:
        (khtml::TextIterator::handleTextBox):
        Clear m_lastTextNodeEndedWithCollapsedSpace after taking it into account.

2004-06-23  Richard Williamson   <rjw@apple.com>

        Implemented changes for latest npruntime.h.

        Reviewed by Chris.

        * kwq/KWQKHTMLPart.h:
        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::KWQKHTMLPart):
        (KWQKHTMLPart::windowScriptNPObject):
        (KWQKHTMLPart::getEmbedInstanceForView):
        * kwq/WebCoreBridge.h:
        * kwq/WebCoreBridge.mm:
        (-[WebCoreBridge windowScriptObject]):
        (-[WebCoreBridge windowScriptNPObject]):

2004-06-22  Richard Williamson   <rjw@apple.com>

        Fixed <rdar://problem/3707162>: accessing embeds[] plug-in interface may crash

        embedInstance was uninitialized.

        Reviewed by Chris.

        * khtml/html/html_objectimpl.cpp:
        (HTMLEmbedElementImpl::HTMLEmbedElementImpl):

2004-06-18  John Sullivan  <sullivan@apple.com>

        Reviewed by Darin
        
        - fixed <rdar://problem/3534851> Pop up windows not showing up within SAP's 
        BW Module (changing location.href on new window created by window.open)

        * khtml/khtml_part.cpp:
        (KHTMLPart::scheduleRedirection):
        allow new redirect to win if delay <= current delay, not just <

=== WebCore-146.1 ===

2004-06-18  Trey Matteson  <trey@apple.com>

        3702053 - DHTML dragging destination can't control the cursor (by setting the drop operation)

        Fallout from security work, but an easy fix.

        Reviewed by Darin

        * kwq/KWQClipboard.mm:
        (KWQClipboard::setDropEffect):  Allow dest side to set dropEffect.

2004-06-18  Darin Adler  <darin@apple.com>

        Reviewed by John.

        - re-fixed <rdar://problem/3701893> show expert preferences notes in xcode causes crash (in EllipsisBox code)

        * khtml/rendering/render_block.cpp: (khtml::RenderBlock::paintEllipsisBoxes):
        Check !isInlineFlow rather than checking hasMarkupTruncation. Otherwise, we break
        plain old non-markup truncation.

2004-06-18  Darin Adler  <darin@apple.com>

        Reviewed by John.

        - fixed <rdar://problem/3701893> show expert preferences notes in xcode causes crash (in EllipsisBox code)

        * khtml/rendering/render_block.cpp: (khtml::RenderBlock::paintEllipsisBoxes):
        Only walk the ellipsis list if hasMarkupTruncation is true.

=== Safari-146 ===

2004-06-17  Richard Williamson   <rjw@apple.com>

        Fixed <rdar://problem/3698867> setting the canvas or parent to display:none and updating the causes a nil-deref

        Ensured that we have a renderer before rendering.

        Reviewed by John.

        * khtml/ecma/kjs_html.cpp:
        (KJS::Context2DFunction::tryCall):

2004-06-17  David Hyatt  <hyatt@apple.com>

        Fix for 3674601, "Read More..." links should trail articles in Emerson.
        
        Reviewed by darin

        * khtml/rendering/bidi.cpp:
        (khtml::RenderBlock::layoutInlineChildren):
        (khtml::RenderBlock::deleteEllipsisLineBoxes):
        (khtml::RenderBlock::checkLinesForTextOverflow):
        * khtml/rendering/render_block.cpp:
        (khtml:::RenderFlow):
        (khtml::RenderBlock::paintEllipsisBoxes):
        (khtml::RenderBlock::nodeAtPoint):
        (khtml::shouldCheckLines):
        (khtml::getLineAtIndex):
        (khtml::getHeightForLineCount):
        (khtml::RenderBlock::lineAtIndex):
        (khtml::RenderBlock::lineCount):
        (khtml::RenderBlock::heightForLineCount):
        (khtml::RenderBlock::clearTruncation):
        * khtml/rendering/render_block.h:
        (khtml::RenderBlock::setHasMarkupTruncation):
        (khtml::RenderBlock::hasMarkupTruncation):
        * khtml/rendering/render_flexbox.cpp:
        (khtml::RenderFlexibleBox::layoutVerticalBox):
        * khtml/rendering/render_line.cpp:
        (khtml::EllipsisBox::m_str):
        (khtml::InlineBox::adjustPosition):
        (khtml::InlineFlowBox::adjustPosition):
        (khtml::InlineFlowBox::clearTruncation):
        (khtml::EllipsisBox::paint):
        (khtml::EllipsisBox::nodeAtPoint):
        (khtml::RootInlineBox::clearTruncation):
        (khtml::RootInlineBox::placeEllipsis):
        (khtml::RootInlineBox::paintEllipsisBox):
        (khtml::RootInlineBox::hitTestEllipsisBox):
        (khtml::RootInlineBox::adjustPosition):
        (khtml::RootInlineBox::childRemoved):
        * khtml/rendering/render_line.h:
        (khtml::InlineBox::clearTruncation):
        * khtml/rendering/render_text.cpp:
        * khtml/rendering/render_text.h:
        (khtml::InlineTextBox::clearTruncation):

2004-06-17  Trey Matteson  <trey@apple.com>

        3698514 - coordinates in ondragstart and ondrag events are wrong

        This part fixes the ondragstart coords.  We salt away the window-based mouseDown
        location, since we need that when we dispatch the ondragstart event.  Previously
        we were errantly using a mouseDown point that had already been converted to view
        coords, and then the dispatch converted it again.

        Reviewed by Maciej.

        * kwq/KWQKHTMLPart.h:
        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::khtmlMouseMoveEvent):  Use window based mouse event coords to dispatch event.
        (KWQKHTMLPart::mouseDown):  Save window based mouse event coords .

2004-06-16  David Hyatt  <hyatt@apple.com>

        Fix for 3596620, implement a subset of CSS3 text truncation for Emerson.
        
        Reviewed by mjs

        * khtml/rendering/bidi.cpp:
        (khtml::RenderBlock::checkLinesForTextOverflow):
        * khtml/rendering/font.cpp:
        (Font::checkSelectionPoint):
        * khtml/rendering/font.h:
        * khtml/rendering/render_block.cpp:
        (khtml::RenderBlock::paintObject):
        (khtml::RenderBlock::paintFloats):
        (khtml::RenderBlock::paintEllipsisBoxes):
        * khtml/rendering/render_block.h:
        * khtml/rendering/render_line.cpp:
        (InlineBox::canAccommodateEllipsis):
        (InlineBox::placeEllipsisBox):
        (InlineFlowBox::paintDecorations):
        (InlineFlowBox::placeEllipsisBox):
        (EllipsisBox::paint):
        (RootInlineBox::placeEllipsis):
        (RootInlineBox::placeEllipsisBox):
        * khtml/rendering/render_line.h:
        (khtml::EllipsisBox::m_str):
        (khtml::RootInlineBox::ellipsisBox):
        * khtml/rendering/render_text.cpp:
        (InlineTextBox::placeEllipsisBox):
        (InlineTextBox::paintDecoration):
        (InlineTextBox::offsetForPosition):
        (RenderText::positionForCoordinates):
        (RenderText::paint):
        * khtml/rendering/render_text.h:
        * kwq/KWQFontMetrics.h:
        * kwq/KWQFontMetrics.mm:
        (QFontMetrics::checkSelectionPoint):
        * kwq/WebCoreTextRenderer.h:

=== WebCore-145.1 ===

2004-06-16  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by John.

        <rdar://problem/3697602> REGRESSION (144.2-TOT) Selection highlight does not draw

        * khtml/rendering/render_canvas.cpp:
        (RenderCanvas::setSelection):

=== Safari-145 ===

2004-06-16  Richard Williamson   <rjw@apple.com>

        Fixed 3695730:  Added support for embeds[], much like
        applets[], to allow access to a plugin's exported interface.

        Reviewed by Chris.

        * khtml/dom/html_document.cpp:
        (HTMLDocument::embeds):
        * khtml/dom/html_document.h:
        * khtml/ecma/kjs_dom.cpp:
        (KJS::getRuntimeObject):
        * khtml/ecma/kjs_html.cpp:
        (KJS::HTMLDocument::tryGet):
        (KJS::HTMLElement::tryGet):
        (KJS::HTMLCollection::tryGet):
        (KJS::HTMLCollection::getNamedItems):
        * khtml/ecma/kjs_html.h:
        (KJS::HTMLDocument::):
        * khtml/ecma/kjs_html.lut.h:
        (KJS::):
        * khtml/html/html_miscimpl.cpp:
        (HTMLCollectionImpl::calcLength):
        (HTMLCollectionImpl::getItem):
        (HTMLCollectionImpl::getNamedItem):
        * khtml/html/html_miscimpl.h:
        (DOM::HTMLCollectionImpl::):
        * khtml/html/html_objectimpl.cpp:
        (HTMLEmbedElementImpl::getEmbedInstance):
        * khtml/html/html_objectimpl.h:
        * kwq/KWQKHTMLPart.h:
        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::getEmbedInstanceForView):

2004-06-15  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Dave.

        <rdar://problem/3695907>: (can't enable selection inside parent where it is disabled via CSS)

        * khtml/css/cssparser.cpp:
        (CSSParser::parseValue): 
        * khtml/css/cssstyleselector.cpp:
        (khtml::CSSStyleSelector::applyProperty):
        * khtml/rendering/render_object.cpp:
        (RenderObject::shouldSelect):
        * khtml/rendering/render_style.h:
        (khtml::):
        (khtml::RenderStyle::userSelect):
        (khtml::RenderStyle::setUserSelect):
        (khtml::RenderStyle::initialUserSelect):

2004-06-15  David Hyatt  <hyatt@apple.com>

        Initial impl of the EllipsisBox.  This code just gets the box created and gets it placed vertically.  The next
        stage will be to position the box horizontally as well.
        
        Reviewed by trey

        * khtml/rendering/bidi.cpp:
        (khtml::RenderBlock::checkLinesForTextOverflow):
        * khtml/rendering/render_line.cpp:
        (RootInlineBox::placeEllipsis):
        * khtml/rendering/render_line.h:
        (khtml::EllipsisBox::m_str):

2004-06-15  David Hyatt  <hyatt@apple.com>

        Implement canAccommodateEllipsis.  The basic idea is that everything on a line will allow an ellipsis to be
        drawn on top of it unless it's a replaced element.  Then, if the replaced element overlaps, the ellipsis won't
        be allowed.
        
        Reviewed by john

        * khtml/rendering/bidi.cpp:
        (khtml::RenderBlock::checkLinesForTextOverflow):
        * khtml/rendering/render_line.cpp:
        (InlineBox::closestLeafChildForXPos):
        (InlineBox::canAccommodateEllipsis):
        (InlineFlowBox::canAccommodateEllipsis):
        (RootInlineBox::canAccommodateEllipsis):
        * khtml/rendering/render_line.h:

2004-06-15  Vicki Murley <vicki@apple.com>

        - added a few layout tests, rdar://3694510

        * layout-tests/editing/deleting/delete-image-004-expected.txt: Added.
        * layout-tests/editing/deleting/delete-image-004.html: Added.
        * layout-tests/editing/selection/extend-by-character-006-expected.txt: Added.
        * layout-tests/editing/selection/extend-by-character-006.html: Added.

2004-06-15  Trey Matteson  <trey@apple.com>

        Dragging tweak:  We pass the mouse down coords instead of the
        latest mouse drag coords to the ondragstart event.  This makes it
        easy for the client to figure the correct drag image offset,
        whereas the mouse drag location is next to useless for that.

        Reviewed by Louch

        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::khtmlMouseMoveEvent):

2004-06-15  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Darin.

        <rdar://problem/3685236>: (Safari does not support onselectstart event handler)

        * khtml/ecma/kjs_dom.cpp:
        (DOMNode::getValueProperty):
        (DOMNode::putValue):
        * khtml/ecma/kjs_dom.h:
        (KJS::DOMNode::):
        * khtml/ecma/kjs_dom.lut.h:
        (KJS::):
        * khtml/html/html_elementimpl.cpp:
        (HTMLElementImpl::parseHTMLAttribute):
        * khtml/misc/htmlattrs.c:
        (hash_attr):
        (findAttr):
        * khtml/misc/htmlattrs.h:
        * khtml/misc/htmlattrs.in:
        * khtml/rendering/render_object.cpp:
        (RenderObject::shouldSelect):
        * khtml/xml/dom2_eventsimpl.cpp:
        (EventImpl::typeToId):
        (EventImpl::idToType):
        * khtml/xml/dom2_eventsimpl.h:
        (DOM::EventImpl::):

2004-06-15  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Darin.

        Fix crash with last checkin.

        * khtml/ecma/kjs_binding.cpp:
        (ScriptInterpreter::domObjectsPerDocument): check the same
        property we want to initialize.

2004-06-15  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Darin.

        <rdar://problem/3685309>: (properties not shared for JS wrappers of same DOM object, accessed from different frames)

        * khtml/ecma/kjs_binding.cpp:
        (ScriptInterpreter::domObjects):
        (ScriptInterpreter::domObjectsPerDocument):
        (ScriptInterpreter::ScriptInterpreter):
        (ScriptInterpreter::forgetDOMObject):
        (ScriptInterpreter::getDOMObjectForDocument):
        (ScriptInterpreter::putDOMObjectForDocument):
        (ScriptInterpreter::deleteDOMObjectsForDocument):
        (ScriptInterpreter::mark):
        (ScriptInterpreter::forgetDOMObjectsForDocument):
        (ScriptInterpreter::updateDOMObjectDocument):
        * khtml/ecma/kjs_binding.h:
        (KJS::ScriptInterpreter::getDOMObject):
        (KJS::ScriptInterpreter::putDOMObject):
        (KJS::ScriptInterpreter::deleteDOMObject):

2004-06-15  Darin Adler  <darin@apple.com>

        - rolled out Ken's fix for <rdar://problem/3672377> assertion failure in AppendNodeCommandImpl::doApply due to non-0 exception code trying to insert a DIV markup string
          (it was making most layout tests fail)

        * khtml/xml/dom_nodeimpl.cpp: (NodeImpl::checkAddChild): Rolled check back in.

2004-06-15  David Hyatt  <hyatt@apple.com>

        Add a truncation variable to text run boxes that will eventually be used to know how to cut out some of the glyphs when
        rendering.
        
        Reviewed by darin

        * khtml/rendering/render_text.cpp:
        (RenderText::clearTextOverflowTruncation):
        * khtml/rendering/render_text.h:
        (khtml::InlineTextBox:::InlineRunBox):
        (khtml::InlineTextBox::clearTruncation):

2004-06-14  Darin Adler  <darin@apple.com>

        Reviewed by Maciej.

        - fixed some things for GC that Patrick missed, or that happened after the branch

        * kwq/DOM-CSS.mm:
        (-[DOMStyleSheet finalize]): Added.
        (-[DOMStyleSheetList finalize]): Added.
        (-[DOMCSSStyleSheet finalize]): Added.
        (-[DOMMediaList finalize]): Added.
        (-[DOMCSSRuleList finalize]): Added.
        (-[DOMCSSRule finalize]): Added.
        (-[DOMCSSStyleDeclaration finalize]): Added.
        (-[DOMCSSValue finalize]): Added.
        (-[DOMRGBColor finalize]): Added.
        (-[DOMRect finalize]): Added.
        (-[DOMCounter finalize]): Added.
        * kwq/DOM.mm:
        (-[DOMObject finalize]): Added.
        (-[DOMNode finalize]): Added.
        (-[DOMNamedNodeMap finalize]): Added.
        (-[DOMNodeList finalize]): Added.
        (-[DOMImplementation finalize]): Added.
        (-[DOMRange finalize]): Added.
        (-[DOMNodeFilter finalize]): Added.
        (-[DOMNodeIterator finalize]): Added.
        (-[DOMTreeWalker dealloc]): Removed unneeded nil check.
        (-[DOMTreeWalker finalize]): Added.
        * kwq/DOMHTML.mm:
        (-[DOMHTMLCollection finalize]): Added.
        (-[DOMHTMLOptionsCollection finalize]): Added.
        * kwq/KWQClipboard.mm:
        (KWQClipboard::KWQClipboard): Use KWQRetain instead of retain.
        (KWQClipboard::~KWQClipboard): Use KWQRelease instead of release.
        * kwq/KWQEditCommand.mm:
        (-[KWQEditCommand finalize]): Added.
        * kwq/KWQFont.mm:
        (QFont::~QFont): Use KWQRelease instead of release.
        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::keyEvent): Use KWQRetain instead of retain, and KWQRelease instead of release.
        (KWQKHTMLPart::mouseDown): Use KWQRetain instead of retain, and KWQRelease instead of release.
        (KWQKHTMLPart::mouseDragged): Use KWQRetain instead of retain, and KWQRelease instead of release.
        (KWQKHTMLPart::mouseUp): Use KWQRetain instead of retain, and KWQRelease instead of release.
        (KWQKHTMLPart::mouseMoved): Use KWQRetain instead of retain, and KWQRelease instead of release.
        (KWQKHTMLPart::sendContextMenuEvent): Use KWQRetain instead of retain, and KWQRelease instead of release.
        * kwq/KWQPageState.mm:
        (-[KWQPageState finalize]): Added. Filed <rdar://problem/3694163> about the fact that this is not right.
        * kwq/KWQTimer.mm:
        (-[KWQSingleShotTimerTarget finalize]): Added.
        * kwq/WebCoreBridge.mm:
        (-[WebCoreBridge finalize]): Added. Filed <rdar://problem/3694165> about the fact that this is not right.

2004-06-15  David Hyatt  <hyatt@apple.com>

        Initial work on text truncation.  Working top-down, I think I've made all the changes I will need to bidi.cpp
        and to the block code itself.  The rest of the code can be concentrated in the render_line and render_text files.
        
        Reviewed by darin

        * khtml/rendering/bidi.cpp:
        (khtml::RenderBlock::layoutInlineChildren):
        (khtml::RenderBlock::findNextLineBreak):
        (khtml::RenderBlock::deleteEllipsisLineBoxes):
        (khtml::RenderBlock::checkLinesForTextOverflow):
        * khtml/rendering/render_block.h:
        * khtml/rendering/render_line.cpp:
        (RootInlineBox::detach):
        (RootInlineBox::detachEllipsisBox):
        (RootInlineBox::canAccommodateEllipsis):
        (RootInlineBox::placeEllipsis):
        * khtml/rendering/render_line.h:
        (khtml::RootInlineBox::RootInlineBox):
        * khtml/rendering/render_text.cpp:
        (RenderText::clearTextOverflowTruncation):
        * khtml/rendering/render_text.h:

2004-06-14  Trey Matteson  <trey@apple.com>

        3692690 - REGRESSION: canceling drag from WebView cause link to load
        Just need to take extra care to cancel any click handling in
        all cases once we're committed to doing the drag.

        Reviewed by Darin.

        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::khtmlMouseMoveEvent):  Call invalidateClick once
        we're past the hysteresis point, and definitely going to try
        to start a drag and drop.

2004-06-14  Darin Adler  <darin@apple.com>

        - fixed crash on boot

        * kwq/KWQFoundationExtras.h: (KWQRetainNSRelease):
        Fixed return value of this method; was uninitialized.

2004-06-14  Trey Matteson  <trey@apple.com>

        Dashboard wants access to pastboard data during the drag gesture.

        We address this need be allowing docs that are local files to access the
        pasteboard data during a drag, while other pages can only get at the types
        list until the drop happens.  This is deemed safe because local files already
        have such broad super powers.

        Reviewed by Maciej.

        * kwq/KWQKURL.h:
        * kwq/KWQKURL.mm:
        (KURL::isLocalFile):  Implement this QT method to look for the file scheme.
        * kwq/WebCoreBridge.mm:
        (-[WebCoreBridge dragOperationForDraggingInfo:]):  Use more relaxed clipboard
        security for local file pages.
        (-[WebCoreBridge dragExitedWithDraggingInfo:]):  Ditto.

2004-06-14  Darin Adler  <darin@apple.com>

        Reviewed by me, code changes by Patrick Beard.

        - fixed <rdar://problem/3671507>: (WebKit should adopt GC changes and compile with GC enabled)

        * WebCore.pbproj/project.pbxproj: Added KWQFoundationExtras.h.
        * kwq/KWQFoundationExtras.h: Added.
        (KWQRetain): Cover for CFRetain that is tolerant of nil.
        (KWQRelease): Cover for CFRelease that is tolerant of nil.
        (KWQRetainNSRelease): Combination of a CFRetain and an -[NSObject release] that is tolerant of nil.
        Also includes a declaration of finalize so we can call [super finalize] without warnings on Panther.

        * kwq/DOM.mm:
        (ObjCNodeFilterCondition::ObjCNodeFilterCondition): Use CFRetain instead of retain.
        (ObjCNodeFilterCondition::~ObjCNodeFilterCondition): Use CFRelease instead of release.
        * kwq/KWQComboBox.mm:
        (QComboBox::~QComboBox): Use KWQRelease instead of release.
        (QComboBox::setFont): Use KWQRelease instead of release.
        (QComboBox::labelFont): Use KWQRetain instead of retain.
        * kwq/KWQCursor.mm:
        (QCursor::QCursor): Use KWQRetain instead of retain.
        (QCursor::~QCursor): Use KWQRelease instead of release.
        (QCursor::operator=): More of the same.
        * kwq/KWQFileButton.mm:
        (KWQFileButton::KWQFileButton): Use KWQRetainNSRelease to retain a newly-allocated object.
        (KWQFileButton::~KWQFileButton): Use CFRelease instead of release.
        * kwq/KWQFont.mm:
        (QFont::QFont): Use KWQRetain instead of retain.
        (QFont::operator=): More of the same.
        (QFont::setFamily): Use KWQRelease instead of release.
        (QFont::setFirstFamily): Use KWQRelease instead of release.
        (QFont::setPixelSize): Use KWQRelease instead of release.
        (QFont::setWeight): Use KWQRelease instead of release.
        (QFont::setItalic): Use KWQRelease instead of release.
        (QFont::getNSFont): Use KWQRetain instead of retain.
        * kwq/KWQFontMetrics.mm:
        (QFontMetricsPrivate::~QFontMetricsPrivate): Use KWQRelease instead of release.
        (QFontMetricsPrivate::getRenderer): Use KWQRetain instead of retain.
        (QFontMetricsPrivate::setFont): Use KWQRelease instead of release.
        * kwq/KWQKHTMLPart.mm:
        (KWQKHTMLPart::~KWQKHTMLPart): Use KWQRelease instead of release.
        (KWQKHTMLPart::clearRecordedFormValues): Use KWQRelease instead of release.
        (KWQKHTMLPart::recordFormValue): Use KWQRetainNSRelease and use KWQRetain instead of retain.
        (KWQKHTMLPart::windowScriptObject): Use KWQRetainNSRelease.
        * kwq/KWQKJobClasses.mm:
        (KIO::TransferJobPrivate::TransferJobPrivate): Use KWQRetainNSRelease.
        (KIO::TransferJobPrivate::~TransferJobPrivate): Use KWQRelease instead of release.
        (KIO::TransferJob::setLoader): More of the same.
        * kwq/KWQLoader.mm:
        (KWQRetainResponse): Use KWQRetain instead of retain.
        (KWQReleaseResponse): Use KWQRelease instead of release.
        * kwq/KWQPainter.mm:
        (QPainterPrivate::~QPainterPrivate): Use KWQRelease instead of release.
        (QPainter::_updateRenderer): More of the same.
        (QPainter::initFocusRing): Use KWQRetainNSRelease.
        (QPainter::clearFocusRing): Use KWQRelease instead of release.
        * kwq/KWQPixmap.mm:
        (QPixmap::QPixmap): Use KWQRetain instead of retain.
        (QPixmap::~QPixmap): Use KWQRelease instead of release.
        (QPixmap::receivedData): Use KWQRetain instead of retain.
        (QPixmap::resize): Use KWQRetainNSRelease and use KWQRelease instead of release.
        (QPixmap::operator=): More of the same.
        * kwq/KWQRegion.mm:
        (QRegion::QRegion): Use KWQRetain instead of retain.
        (QRegion::~QRegion): Use KWQRelease instead of release.
        (QRegion::operator=): More of the same.
        * kwq/KWQTimer.mm:
        (QTimer::start): Use KWQRetain instead of retain.
        (QTimer::stop): Use KWQRelease instead of release.
        (QTimer::fire): Use KWQRelease instead of release.
        * kwq/KWQWidget.mm:
        (QWidget::QWidget): Use KWQRetain instead of retain.
        (QWidget::~QWidget): Use KWQRelease instead of release.
        (QWidget::setView): More of the same.
        * kwq/WebCoreSettings.mm: (-[WebCoreSettings finalize]):
        Delete the storage here, as well as in dealloc.

2004-06-14  Maciej Stachowiak  <mjs@apple.com>

        Reviewed by Trey.

        <rdar://problem/3693818>: (Safari should use CG calls for circle drawing for better performance)

        * kwq/KWQPainter.m