summaryrefslogtreecommitdiff
path: root/applets
diff options
context:
space:
mode:
Diffstat (limited to 'applets')
-rw-r--r--applets/notification_area/libstatus-notifier-watcher/gf-sn-watcher-v0.c14
1 files changed, 14 insertions, 0 deletions
diff --git a/applets/notification_area/libstatus-notifier-watcher/gf-sn-watcher-v0.c b/applets/notification_area/libstatus-notifier-watcher/gf-sn-watcher-v0.c
index 8e82c10f..3d3a97c7 100644
--- a/applets/notification_area/libstatus-notifier-watcher/gf-sn-watcher-v0.c
+++ b/applets/notification_area/libstatus-notifier-watcher/gf-sn-watcher-v0.c
@@ -284,10 +284,24 @@ gf_sn_watcher_v0_handle_register_item (GfSnWatcherV0Gen *object,
if (watch != NULL)
{
+ /* the specification doesn't explicitly state what should happen when
+ * trying to register the same item again, so it would make sense to
+ * forbid it. Unfortunately libappindicator tries re-registering pretty
+ * often, and even falls back to System Tray if it fails.
+ * So in practice we need to be forgiving and pretend it's OK. */
+#if 0
g_dbus_method_invocation_return_error (invocation, G_DBUS_ERROR,
G_DBUS_ERROR_INVALID_ARGS,
"Status Notifier Item with bus name '%s' and object path '%s' is already registered",
bus_name, object_path);
+#else
+ g_warning ("Status Notifier Item with bus name '%s' and object path '%s' is already registered",
+ bus_name, object_path);
+ /* FIXME: is it OK to simply ignore the request instead of removing the
+ * old one and adding the new one? I don't see the problem as
+ * they are identical, but...? */
+ gf_sn_watcher_v0_gen_complete_register_item (object, invocation);
+#endif
return TRUE;
}