Thank you!  This does indeed fix it.

For some reason there's still a "glitch" in the file that caused the
initial A/V to be out of sync.  This "glitch" does not exhibit itself in
xine, so I don't think we can call it a corruption.  Or at the very
least, xine handles it much more gracefully.

Either way, I'm much happier with it now.

Thanks,
Brian

On Mon, 2006-09-11 at 23:32 +0100, Martin Ebourne wrote:
> On Sat, 09 Sep 2006 16:42:36 -0400, Brian C. Huffman wrote:
> > On Sat, 2006-09-09 at 16:35 -0400, Isaac Richards wrote:
> >> Because it works fine for me, and you provided no way for me to reproduce 
> >> things?  Because I'm very, very tired of people completely ignoring the 
> >> TicketHowTo (linked to on the 'new ticket' page) as well as repeated requests 
> >> not to play with the severity/priority/milestone settings?
> >> 
> >> If you want to provide additional debugging, please try finding out what 
> >> revision the problem showed up in.
> >> 
> > 
> > Fair enough.  How much time do I have to find this before .20 comes
> > out?  ;-)
> 
> This sounds quite similar to problems I had which I reported here:
> 
> http://www.gossamer-threads.com/lists/mythtv/dev/204857
> 
> I suggested a possible workaround but unfortunately there wasn't really
> a response to that email and I didn't get round to chasing it up. It's
> definitely broken for some vob and mpg files.
> 
> I've been running with the below patch in my tree ever since then and it
> has worked well for me. If it's considered an acceptable fix I'll submit
> it to trac properly.
> 
> As to the seeking problems that could be a lack of seek table and the
> problem with not using gop by frame. There's a proper ticket for the gop
> issue (trac is disabled at the moment though so can't look for it) and the
> devs want a real fix not a hack (see second patch for the hack) which is
> reasonable. If there was a bit of guidance on what the proper fix should
> include (esp. wrt database upgrade if required) I might code that one up
> myself. Once you've got the gop hack in you need to run 'mythcommflag
> --video <full-path-to-file>' to build the seek table.
> 
> After these two changes all my videos and vobs work well in the
> internal player which I've been using exclusively for ages. I use ffmpeg
> but libmpeg2 works as well (as far as I've tried).
> 
> Cheers,
> 
> Martin.
> 
> (Sorry, can't do attachments in pan)
> 
> mythtv-disable-wraptimecode.patch
> 
> --- libs/libmythtv/NuppelVideoPlayer.cpp        (revision 9953)
> +++ libs/libmythtv/NuppelVideoPlayer.cpp        (working copy)
> @@ -3329,6 +3329,11 @@
>  
>  void NuppelVideoPlayer::WrapTimecode(long long &timecode, TCTypes tc_type) 
>  {
> +    if (!ringBuffer->LiveMode())
> +    {
> +        return;
> +    }
> +
>      if ((tc_type == TC_AUDIO) && (tc_wrap[TC_AUDIO] == LONG_LONG_MIN))
>      {
>          long long newaudio;
> 
> 
> mythtv-gop-byframe.patch
> 
> --- libs/libmythtv/avformatdecoder.cpp  (revision 9953)
> +++ libs/libmythtv/avformatdecoder.cpp  (working copy)
> @@ -859,17 +859,8 @@
>          }
>  
>          // we will not see a position map from db or remote encoder,
> -        // set the gop interval to 15 frames.  if we guess wrong, the
> -        // auto detection will change it.
> -        keyframedist = 15;
> -        positionMapType = MARK_GOP_START;
> -
> -        if (!strcmp(fmt->name, "avi"))
> -        {
> -            // avi keyframes are too irregular
> -            keyframedist = 1;
> -            positionMapType = MARK_GOP_BYFRAME;
> -        }
> +        keyframedist = 1;
> +        positionMapType = MARK_GOP_BYFRAME;
>      }
>  
>      // Don't build a seek index for MythTV files, the user needs to
> 
> _______________________________________________
> mythtv-dev mailing list
> mythtv-dev@mythtv.org
> http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev