[SCM] WebKit Debian packaging branch, debian/unstable, updated. debian/1.1.15-1-40151-g37bb677

[hyatt]

The following commit has been merged in the debian/unstable branch:
commit 8e9813e74c9b1fc4868992570c3e1f322bb30601
Author: hyatt <hyatt at 268f45cc-cd09-0410-ab3c-d52691b4dbfc>
Date:   Mon Oct 21 08:16:13 2002 +0000

    	Fix for margin problems on aintitcool.com.  There was a subtlety
    	to WinIE's quirk.  For <h1>-<h6> only, they don't collapse the
    	bottom margin for table cells (they still collapse the top
    	margin).  I just love emulating the odd behavior of WinIE.
    	Wheee.
    
            * khtml/rendering/render_flow.cpp:
            (RenderFlow::layoutBlockChildren):
    
    
    git-svn-id: http://svn.webkit.org/repository/webkit/trunk@2392 268f45cc-cd09-0410-ab3c-d52691b4dbfc

diff --git a/WebCore/ChangeLog-2002-12-03 b/WebCore/ChangeLog-2002-12-03
index f5ecccd..fdeaa60 100644
--- a/WebCore/ChangeLog-2002-12-03
+++ b/WebCore/ChangeLog-2002-12-03
@@ -1,3 +1,14 @@
+2002-10-21  David Hyatt  <hyatt at apple.com>
+
+	Fix for margin problems on aintitcool.com.  There was a subtlety
+	to WinIE's quirk.  For <h1>-<h6> only, they don't collapse the
+	bottom margin for table cells (they still collapse the top
+	margin).  I just love emulating the odd behavior of WinIE.
+	Wheee.
+	
+        * khtml/rendering/render_flow.cpp:
+        (RenderFlow::layoutBlockChildren):
+
 2002-10-20  Darin Adler  <darin at apple.com>
 
 	Redid my fix. It was just an ordering thing. The image had to already be
diff --git a/WebCore/ChangeLog-2003-10-25 b/WebCore/ChangeLog-2003-10-25
index f5ecccd..fdeaa60 100644
--- a/WebCore/ChangeLog-2003-10-25
+++ b/WebCore/ChangeLog-2003-10-25
@@ -1,3 +1,14 @@
+2002-10-21  David Hyatt  <hyatt at apple.com>
+
+	Fix for margin problems on aintitcool.com.  There was a subtlety
+	to WinIE's quirk.  For <h1>-<h6> only, they don't collapse the
+	bottom margin for table cells (they still collapse the top
+	margin).  I just love emulating the odd behavior of WinIE.
+	Wheee.
+	
+        * khtml/rendering/render_flow.cpp:
+        (RenderFlow::layoutBlockChildren):
+
 2002-10-20  Darin Adler  <darin at apple.com>
 
 	Redid my fix. It was just an ordering thing. The image had to already be
diff --git a/WebCore/ChangeLog-2005-08-23 b/WebCore/ChangeLog-2005-08-23
index f5ecccd..fdeaa60 100644
--- a/WebCore/ChangeLog-2005-08-23
+++ b/WebCore/ChangeLog-2005-08-23
@@ -1,3 +1,14 @@
+2002-10-21  David Hyatt  <hyatt at apple.com>
+
+	Fix for margin problems on aintitcool.com.  There was a subtlety
+	to WinIE's quirk.  For <h1>-<h6> only, they don't collapse the
+	bottom margin for table cells (they still collapse the top
+	margin).  I just love emulating the odd behavior of WinIE.
+	Wheee.
+	
+        * khtml/rendering/render_flow.cpp:
+        (RenderFlow::layoutBlockChildren):
+
 2002-10-20  Darin Adler  <darin at apple.com>
 
 	Redid my fix. It was just an ordering thing. The image had to already be
diff --git a/WebCore/khtml/rendering/render_flow.cpp b/WebCore/khtml/rendering/render_flow.cpp
index 093b9af..0ced634 100644
--- a/WebCore/khtml/rendering/render_flow.cpp
+++ b/WebCore/khtml/rendering/render_flow.cpp
@@ -37,6 +37,8 @@
 #include "xml/dom_nodeimpl.h"
 #include "xml/dom_docimpl.h"
 #include "khtmlview.h"
+#include "htmltags.h"
+
 using namespace DOM;
 using namespace khtml;
 
@@ -361,9 +363,17 @@ void RenderFlow::layoutBlockChildren( bool relayoutChildren )
     // self-collapsing blocks.
     bool topMarginContributor = true;
     
+    // The last non-collapsed child we encountered.      
+    // We will avoid throwing away the bottom margins if the last non-collapsed child encountered is
+    // an H1-H6 element. -dwh
+    RenderObject* lastNonCollapsedChild = 0;
+    
     // These flags track the previous maximal positive and negative margins.
     int prevPosMargin = maxTopMargin(true);
     int prevNegMargin = maxTopMargin(false);
+    
+    // Whether or not we encountered an element with clear set that actually had to
+    // be pushed down below a float.
     int clearOccurred = false;
     
     int oldPosMargin = prevPosMargin;
@@ -463,6 +473,7 @@ void RenderFlow::layoutBlockChildren( bool relayoutChildren )
                     ypos = m_height + prevPosMargin - prevNegMargin;
             }
             else {
+                lastNonCollapsedChild = child;
                 if (!topMarginContributor || 
                      (!canCollapseWithChildren && 
                       (element()->getDocument()->parseMode() == DocumentImpl::Strict ||
@@ -558,11 +569,24 @@ void RenderFlow::layoutBlockChildren( bool relayoutChildren )
     // with it.
     if (canCollapseWithChildren && !autoHeight)
         canCollapseWithChildren = false;
-        
+    
+    // XXX This is a gross hack. Basically if the last child is marginless, we will (in quirks
+    // mode) allow the collapsed bottom margin to be added to the table cell.  This is a sleazy
+    // way of keeping <td><p>Foo</p></td> from showing a bottom margin but still allowing
+    // <td><font><h3>Foo</h3></font></td> to show a bottom margin.  
+    // A better fix (since we only cared about H1-H6 in the first place would be to recur
+    // into the last non-collapsed children until you don't find one any more, and if that
+    // innermost child was an H1-H6, use it. -dwh
+    int marginBottom = -1;
+    if (lastNonCollapsedChild)
+        marginBottom = lastNonCollapsedChild->marginBottom();
+    
     // If we can't collapse with children then go ahead and add in the bottom margins.
     if (!canCollapseWithChildren && !topMarginContributor &&
         (element()->getDocument()->parseMode() == DocumentImpl::Strict ||
-         !isTableCell()))
+         !isTableCell() || (lastNonCollapsedChild && lastNonCollapsedChild->element() &&
+           lastNonCollapsedChild->element()->id() >= ID_H1 && 
+           lastNonCollapsedChild->element()->id() <= ID_H6) || (marginBottom == 0)))
         m_height += prevPosMargin - prevNegMargin;
         
     m_height += toAdd;

-- 
WebKit Debian packaging