[Bug 740575] Fixing DTS in GStreamer

[GStreamer (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=740575

--- Comment #101 from Nicolas Dufresne (stormer) <nicolas.dufresne at collabora.co.uk> ---
So I think I better understand the idea of the signed CTTS in version 1 of that
spec. Here's what it looks like:

For video: CT(0) = DT(0) + CTTS(0)
For audio: CT(0) = DT(0)

Where DT(0) is defined as 0. The problem is that there is no way unless CTTS(0)
is 0 to synchronize these streams. I'm not sure how they came up with that
weird idea, but making CTTS signed may help solving this. It directly breaks
the rule that DT(n) <= CT(n) though. Using that feature would mean that DT(n)
is no longer the decoding time as stated in the spec. Instead it would be
shifted forward a fixed value. The problem is that they forgot that this shift
is hard to later on. The demuxer would have to traverse the samples to find the
lowest negative CTTS value and that before producing anything. For this reason,
I think the implemented method is nice since it covers the old spec (change
that most muxer may have never notice). I think there is way our muxer can
accommodate both muxer behaviour, but I would prefer to just keep moving
forward and this and fix it properly though another bug. (if all CTTS are
positive, it's quite clear the old muxing scheme was used).

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