*** Bug 52847 has been marked as a duplicate of this bug. ***

I think that Bug 52847 does not really fall under the pattern I have described.
In that case, the link (e.g.
http://slashdot.org/articles/04/02/21/1349221.shtml ) does not point to the top
level of a site. Also, the link matches exactly what is displayed in the
address line of konqueror after clicking it. The problem with this link seems
to be more specific.

However, looking at the left column of Slashdot, for example the link "Books"
(pointing to http://books.slashdot.org ) falls under the pattern I described
above. It shows the same behaviour.

I can reproduce this with 3.2 branch.

This problem still exists after upgrading to 3.2.1.

This problem still exists after upgrading to 3.2.3.

CVS commit by binner: 

Color 'http://foo' links as visited if 'http://foo/' is in history
BUG: 75771


  M +4 -1      cssstyleselector.cpp   1.328


--- kdelibs/khtml/css/cssstyleselector.cpp  #1.327:1.328
@@ -754,5 +754,8 @@ static void checkPseudoState( const CSSS
     }
     //completeURL( attr.string() );
-    pseudoState = KHTMLFactory::vLinks()->contains( u ) ? PseudoVisited : PseudoLink;
+    bool contains = KHTMLFactory::vLinks()->contains( u );
+    if ( !contains && u.contains('/')==2 )
+      contains = KHTMLFactory::vLinks()->contains( u+'/' );
+    pseudoState = contains ? PseudoVisited : PseudoLink;
 }
 

In HTTP, http://example.com, http://example.com/, http://example.com/foo, and
http://example.com/foo/ all represent different entities.  Indeed, it's
possible to make a typical web server such as Apache present different entities
for each URL, although by default many do seem to offer canonicalization (e.g.
redirect to http://example.com/foo/ when you visit http://example.com/foo).

Unlike the file systems we are often used to, in URL space, a "directory" can
can contain both a "file" and "subdirectory" with identical names.

I don't think that browsers should force a file-system-centric view of URL
space by treating http://example.com/foo as visited if http://example.com/foo/
is, or http://example.com as visited for http://example.com/ for that matter. 
The real issue is that people who publish URL's are not careful about their
trailing slashes.

When I click on a link to http://example.com , that link should be marked as
visited afterwards (independently of the question if the server redirects me to
http://example.com/ or not).

Therefore, I think that John Belmonte's comment does not change the fact that
the original behaviour is a bug. (However, he might be right that the fix in
comment #6 is not the correct solution to the problem.)