Merge lp:~alan-griffiths/compiz-compizconfig-gconf/Bug-953214 into lp:compiz-compizconfig-gconf

[Alan Griffiths]

Seems to fix lp:953214

I'm not too sure about this API - so I hope someone who understands gtk can review. But I've seen no ill-effects and no more SEG faults.

[Daniel van Vugt (vanvugt)]

Looks quite sane.

review: Approve

[Mikkel Kamstrup Erlandsen (kamstrup)]

g_main_loop_is_running(NULL) looks odd to me. The code for that function is:

gboolean
g_main_loop_is_running (GMainLoop *loop)
{
  g_return_val_if_fail (loop != NULL, FALSE);
  g_return_val_if_fail (g_atomic_int_get (&loop->ref_count) > 0, FALSE);

  return loop->is_running;
}

So this statement always returns FALSE. So you've effectively just removed the processEvents(0) call.

=== modified file 'src/gconf.c'
--- src/gconf.c	2011-09-09 09:33:52 +0000
+++ src/gconf.c	2012-03-15 16:04:24 +0000
@@ -1958,7 +1958,11 @@
 static Bool
 finiBackend (CCSContext * context)
 {
-    processEvents (0);
+
+    if (g_main_loop_is_running(NULL))
+    {
+	processEvents (0);
+    }
 
     gconf_client_clear_cache (client);
     finiClient ();
@@ -1972,7 +1976,10 @@
     gconf_engine_unref (conf);
     conf = NULL;
 
-    processEvents (0);
+    if (g_main_loop_is_running(NULL))
+    {
+	processEvents (0);
+    }
     return TRUE;
 }