baseparse: properly track upstream timestamps

... rather than with a delay.
This commit is contained in:
Mark Nauwelaerts 2011-01-11 15:24:02 +01:00 committed by Tim-Philipp Müller
parent 6df2a01762
commit 054621a58c

View file

@ -2106,6 +2106,14 @@ gst_base_parse_chain (GstPad * pad, GstBuffer * buffer)
break;
}
/* move along with upstream timestamp (if any),
* but interpolate in between */
timestamp = gst_adapter_prev_timestamp (parse->adapter, NULL);
if (GST_CLOCK_TIME_IS_VALID (timestamp) &&
(parse->priv->prev_ts != timestamp)) {
parse->priv->prev_ts = parse->priv->next_ts = timestamp;
}
/* FIXME: Would it be more efficient to make a subbuffer instead? */
outbuf = gst_adapter_take_buffer (parse->adapter, fsize);
outbuf = gst_buffer_make_metadata_writable (outbuf);
@ -2114,14 +2122,7 @@ gst_base_parse_chain (GstPad * pad, GstBuffer * buffer)
GST_BUFFER_OFFSET (outbuf) = parse->priv->offset;
parse->priv->offset += fsize;
GST_BUFFER_TIMESTAMP (outbuf) = GST_CLOCK_TIME_NONE;
/* move along with upstream timestamp (if any),
* but interpolate in between */
timestamp = gst_adapter_prev_timestamp (parse->adapter, NULL);
if (GST_CLOCK_TIME_IS_VALID (timestamp) &&
(parse->priv->prev_ts != timestamp)) {
parse->priv->prev_ts = parse->priv->next_ts = timestamp;
}
GST_BUFFER_DURATION (outbuf) = GST_CLOCK_TIME_NONE;
frame->buffer = outbuf;
ret = gst_base_parse_handle_and_push_frame (parse, bclass, frame);