[mythtv] AC3 + bigdvb3.5 + australia + transport.c

[Tim Davies]

> Folks,
> 
> Attahced is a new version of the AC3 bigdvb patch.
> 
> 1. It contains a fully functioning transport.c that requires no mpeg.c
> hack,
> finally! It is a combintation of Marcus' changes and mine.
> 

Yay!  It works.  Well I did make a *small* change...

> 2. I have also started implementing/fixing audio track selection through
> the
> +/- keys. I think it might actually work for TS recording, but haven't
> tested
> it. This should also allow toggling of languages where avialable, however
> I
> have no streams to test it against here in oz.

Works on the TS.  Switches between the AC3 track and the MPEG track
perfectly.

> 
> 3. Included is an attempt to fix the audio change problems in the TS
> recoding,
> again it is not tested but has a good chance of working. Tim can you test
> it?
> 

Yeah, I found the real problem.  I just moved the incCurrentAudio stream
down a bit, after where it determines that we aren't looking at a packet
from the wantedAudioStream.  I should have seen that earlier...

And I removed your multiple packet checker! :-o

> 4. There is still some code in transport.c that I have ifdef'd out.
> Everything
> I have tested works without it and if it is left in then there is audio
> stuttering.
> 

The TS works whether this is in or not.


It's interesting to note, that when recording the TS my machine (Athlon XP
3200+) has enough grunt to do 1080i (with deinterlace, and then some).  But
if I don't record the TS it doesn't quite make it.

I still get jerky video when changing to the 576p channels.  Just an
annoyance really.  And only when recording the TS.  It doesn't carry on to
the subsequent channel change anymore.  Just exit and "Watch TV" again, and
it is fine.

And the timing thing.  It is all screwed up when the PCR is sent as a stream
on its own.  No probs if the PCR stream is the video stream.  I don't know
if it means anything, but in the recordedmarkup table the mark (number of
frames) is incorrect but the offset (bytes in file) is correct.

So is PICTURE_START not being determined properly?  Mpegts.c doesn't seem to
be doing much with the PCR at all.


Tim.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: oz-ac3-bigdvb3.5v4.diff
Type: application/octet-stream
Size: 24308 bytes
Desc: not available
Url : http://mythtv.org/pipermail/mythtv-dev/attachments/20050113/d1888a7b/oz-ac3-bigdvb3.5v4.obj