wrappercamerabinsrc: use the drain query instead of flush

The flush might clear any pending captures that are still in
queues being processed while the drain query will wait for those
to be properly done

diff --git a/gst/camerabin2/gstwrappercamerabinsrc.c b/gst/camerabin2/gstwrappercamerabinsrc.c
index 24f44f39aa51b57d028c15e22b89e64c2f81d58e..5c08d0eb9b815a4e52cbc94a35b6901ac48fafe5 100644 (file)
--- a/gst/camerabin2/gstwrappercamerabinsrc.c
+++ b/gst/camerabin2/gstwrappercamerabinsrc.c
@@ -833,18 +833,16 @@ start_image_capture (GstWrapperCameraBinSrc * self)
 
   GST_DEBUG_OBJECT (self, "Starting image capture");
 
-  /* FIXME - V4L2 source will not close the device until all buffers have came
-   * back. Flushing the pipeline, will ensure it's properly closed, and that
-   * setting it back to PLAYING will work. This is more a workaround then a
-   * solution to buffer reclaiming. */
+  /* V4L2 source will not close the device until all buffers have came
+   * back. Draining the pipeline, will ensure it's properly closed, and that
+   * setting it back to PLAYING will work. */
   pad = gst_element_get_static_pad (self->src_vid_src, "src");
   if (self->image_renegotiate) {
 
     peer = gst_pad_get_peer (pad);
     gst_object_unref (pad);
-    gst_pad_send_event (peer, gst_event_new_flush_start ());
+    gst_pad_query (peer, gst_query_new_drain ());
     gst_element_set_state (self->src_vid_src, GST_STATE_READY);
-    gst_pad_send_event (peer, gst_event_new_flush_stop (TRUE));
     gst_object_unref (peer);
 
     /* clean capsfilter caps so they don't interfere here */