+2006-09-17 Havoc Pennington <hp@redhat.com>
+
+ * dbus/dbus-transport.c (_dbus_transport_get_is_authenticated):
+ fix so that if unix_user_function is set on windows, we still
+ do the default check that the auth identity matches the bus identity
+
2006-09-16 Havoc Pennington <hp@redhat.com>
* dbus/dbus-transport.c (_dbus_transport_open): modify to delegate
*/
dbus_bool_t
_dbus_transport_get_is_authenticated (DBusTransport *transport)
-{
+{
+ /* We don't want to run unix_user_function on Windows, but it
+ * can exist, which allows application code to just unconditionally
+ * set it and have it only be invoked when appropriate.
+ */
+ dbus_bool_t on_windows = FALSE;
+#ifdef DBUS_WIN
+ on_windows = TRUE;
+#endif
+
if (transport->authenticated)
return TRUE;
else
_dbus_auth_get_identity (transport->auth, &auth_identity);
- if (transport->unix_user_function != NULL)
+ if (transport->unix_user_function != NULL && !on_windows)
{
dbus_bool_t allow;
DBusConnection *connection;
_dbus_verbose ("unlock %s\n", _DBUS_FUNCTION_NAME);
_dbus_connection_unlock (connection);
-#ifdef DBUS_WIN
- /* FIXME this is a bad hack for now because we want to ship 1.0
- * without leaking any weird unix-emulation implementation details
- * to the public API. The correct fix will be to move this
- * unix user function invocation into dbus-transport-unix.c, and
- * have a separate, appropriate function for Windows, if any.
- * On Windows we may only use the session (not system) daemon
- * anyway, so it might not matter.
- *
- * The windows and unix callbacks should both be stored/set
- * in cross-platform code, so apps can unconditionally set
- * them both, but only the platform-appropriate one
- * should ever be invoked.
- */
- allow = TRUE;
-#else
allow = (* unix_user_function) (connection,
auth_identity.uid,
unix_user_data);
-#endif
_dbus_verbose ("lock %s post unix user function\n", _DBUS_FUNCTION_NAME);
_dbus_connection_lock (connection);