Re: [patch] merge from java-gui-branch

[Andrew Haley <aph at redhat dot com>]

Mark Wielaard writes:
 > On Fri, 2004-11-26 at 05:49, Mark Wielaard wrote:
 > > On Fri, 2004-11-26 at 05:34, graydon hoare wrote:
 > > > except that some of the examples do odd -- well frankly "impossible" -- 
 > > > things like jump to functions which aren't actually called anywhere. so 
 > > > we think these bugs relate more to the recent BC ABI merge than the 
 > > > actual gui changes
 > > 
 > > I need to recheck with a clean tree without this patch applied.
 > 
 > Checked and it fails in the same way.
 > And the workaround below also makes it work again.
 > 
 > > But the following change made both the GNU Classpath awt demo and the
 > > mauve visual test engine work just fine for me (without this patch they
 > > crash whenever you try to do something with menus):
 > > 
 > > --- gnu/java/awt/peer/gtk/GtkMenuItemPeer.java  2 Sep 2004 05:10:11 -0000       1.5
 > > +++ gnu/java/awt/peer/gtk/GtkMenuItemPeer.java  26 Nov 2004 04:47:09 -0000
 > > @@ -74,8 +97,10 @@
 > >        }
 > >      else // parent instanceof Menu
 > >        {
 > > -       ((GtkMenuPeer)((Menu)parent).getPeer ()).addItem (this,
 > > -                                                         item.getShortcut ());
 > > +       Menu m = (Menu) parent;
 > > +       GtkMenuPeer gmp = (GtkMenuPeer) m.getPeer();
 > > +       MenuShortcut ms = item.getShortcut();
 > > +       gmp.addItem(this, ms);
 > >        }
 > >    }
 > > 
 > > The patch should not change the semantics of the code. But obviously it
 > > does...
 > 
 > So it seems that with or without the gui merge patch
 > gnu.java.awt.peer.gtk.GtkMenuItemPeer.setParent(MenuItem) is
 > miscompiled...

Is there some way to make a test case from this?

Thanks,
Andrew.