[mythtv-users] Missing --passthrough option in mythtranscode

[Paul Gardiner]

On 06/06/2011 13:01, Jean-Yves Avenard wrote:
> Hi
>
> On 6 June 2011 20:41, Paul Gardiner<lists at glidos.net>  wrote:
>
>> Thanks for reply. Are not timestamps lost in fifo mode in any case? I
>> assume in fifo mode, mythtranscode has to make sure it generates exactly
>> the right number of audio sample to match the number of frames it puts
>> out, dropping from or inserting to one of the other to keep sync, in the
>> case there is a temporary fault in the stream. That must be possible
>> because presumably a player has to do the same thing, and in fifo
>> mode, mythtranscode is more or less playing the movie to the pipes. I
>> don't know much about ac3 etc., but imagined the same was possible
>> by counting packets.
>
> In Fifo mode you end up with two separates, independent stream: one
> video, one audio...
>
> The player, on the contrary keeps also the timestamps so it can
> recover when there's a fault in the stream etc..
>
> I don't know how fifo can do that.

Sorry, I'm not describing what I mean well. I'll have another go.

If the player is able to recover by using timestamps, then rather
than feed the display buffer and the audio system, it could
feed video frames and audio samples into two pipes, ensuring always
to maintain the correct rate for both (at times having to throw away
or simulate data to keep the rates correct). And if the player
could do it then so could mythtranscode. Timestamps are needed only
to deal with slow drift and breaks in the stream (I assume). Provided
at the point in the system at which timestamps are lost, correct
rates for the two streams are ensured, loss of sync should be avoided
(other than the slow drift thing).

Maybe I'm missing something.

Cheers,
	Paul.