[Bug 754222] adaptivedemux: Timestamping of multi-period streams is not correct

[GStreamer (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=754222

--- Comment #12 from Sebastian Dröge (slomo) <slomo at coaxion.net> ---
(In reply to Thiago Sousa Santos from comment #11)
> Review of attachment 310519 [details] [review]:
> 
> ::: gst-libs/gst/adaptivedemux/gstadaptivedemux.c
> @@ +830,3 @@
>        stream->segment.base =
> +          gst_segment_to_running_time (&demux->segment, GST_FORMAT_TIME,
> +          period_start);
> 
> Suppose you have a dash mpd with 2 periods, P0 and P1 with 10s each. No
> offsets.
> 
> The user does a flushing seek to 7s, that lands within P0 and the segment
> will have: base=0, start=7s - 0.
> It plays P0 until the end and moves to P1, now the segment will have:
> base=10s, start=0.

No, it will have base=3s because the running time is calculated based on the
demuxer segment. The demuxer segment will be start=7s,base=0s which leads to a
running time of 3s at P1's start time (10s).

I tested this case btw, but with presentation time offsets.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.