EText BadCursor bug (Was Re: [Evolution] more bugs...)

[Damon Chaplin <damon helixcode com>]

Federico Mena Quintero wrote:

> There is still a pesky BadCursor bug that I'm trying to hunt down, but
> it is driving me nuts.

I've figured this out. It goes like this:

 o EText is being edited when something else grabs the focus.
 o the canvas sends the EText a focus_out event.
 o the EDayView text event signal handler is called first, and it sends
   the updated iCalendar object to the server. The server sends the
   obj_updated signal back and the EDayView destroys the current EText
   and creates a new one.
 o Now the EText focus_out class handler is run, but the EText is now
   destroyed, and has destroyed its cursors. But it doesn't check for
   this so it tries to set the cursor -> BadCursor


This patch seems to fix it. It sets the cursors to NULL after destroying
them, to avoid any possible BadCursor.

It also returns from the event class function if the EText item is
destroyed.

Damon


--- e-text.c.orig       Sat Sep 23 17:03:30 2000
+++ e-text.c    Sat Sep 23 19:07:18 2000
@@ -1545,7 +1545,9 @@
        text->gc = NULL;
 
        gdk_cursor_destroy (text->i_cursor);
+       text->i_cursor = NULL;
        gdk_cursor_destroy (text->default_cursor);
+       text->default_cursor = NULL;
 
        if (parent_class->unrealize)
                (* parent_class->unrealize) (item);
@@ -2566,6 +2568,9 @@
        ETextEventProcessorEvent e_tep_event;
 
        gint return_val = 0;
+
+       if (GTK_OBJECT_DESTROYED (item))
+               return FALSE;
 
        e_tep_event.type = event->type;
        switch (event->type) {