Created attachment 15708 [details] [review]
fix memory corruption in the registry

shoot me, stupid thing. This will be fixed when I commit my rewrite of
registry/registrypool/plugin querying code.

CVS HEAD appears to be fixed.  i'm not sure whether 0.6.1 needs a fix
for this problem.  Otherwise, close this bug.

There's no gstregistrypool.c in 0.6.x. There is a gstregistry.c with
the following code:

GList*
gst_registry_pool_plugin_list (void)
{
  GList *result = NULL;
  GList *walk = _gst_registry_pool;
                                                                     
          
  while (walk) {
    GstRegistry *registry = GST_REGISTRY (walk->data);
                                                                     
          
    /* FIXME only include highest priority plugins */
    result = g_list_concat (result, g_list_copy (registry->plugins));
                                                                     
          
    walk = g_list_next (walk);
  }
                                                                     
          
  return g_list_concat (_gst_registry_pool_plugins, result);
}

I'm assuming I need to use g_list_copy() on _gst_registry_pool_plugins
here too?

Yes, you need g_list_concat if the GList is supposed to be freed by
the caller (i don't have 0.6.x source code handy so i can't check). 
g_list_concat does _not_ copy the list.

Created attachment 16221 [details] [review]
for review

fixed in 0.6 branch