[mythtv] [PATCH] pcHDTV: Enable seeking in recordings, trim unused data from stream

[Edward Wildgoose]

> The following patch fixes a couple problems with pcHDTV support, and  
> provides a framework for a few more enhancements.

You should definitely take a look at the DVB code, especially the latest changes that Kenneth has written.  I think you may be duplicating a lot of work here?

> The two major benefits of this patch:
> * Enables seeking in recordings.  Seeking in live TV was working OK,  
> but the old code was not reliably writing keyframes to the  
> recordedmarkup table (maybe 1 in 10 minutes) so seeking in stuff you  
> recorded was pretty busted.

Also needs doing with DVB mpeg

> Note that size could be reduced  
> even further by recoding all the way to an MPEG2 system stream, which I  
> have not done.

This is done in the DVB code using some similar code to the original ts2ps program which comes with one of the dvb tools packages.

I wonder if there is still some work which needs doing to this though because timestamps seem to be a bit wonky...?

> This second change also works around a bug where playback would use all  
> audio packets regardless of whether they belong to our stream or not,  
> resulting in garbled audio on stations with >1 subprogram.

We have a "pid" extractor which just grabs selected pids.  This allows extracting the ac3, or dts, or analogue audio, etc.

> There are also comments indicating areas for future work.  Two big  
> areas:
> * decode the PAT (Program Association Table) data, allowing Myth to  
> tune to different subprograms of a channel (right now it picks the  
> first).  We would want to store each subchannel as a separate channel  
> in the database, with either a new field or an encoding of the freqid  
> to indicate which subchannel it represents.
> * decode the PSIP (Program and System Information Protocol) data, which  
> carries guide data provided by the broadcaster.  Coverage from station  
> to station and city to city seems to be spotty, but it could be  
> valuable, particularly for where HDTV differs from normal analog  
> listings.

It doesn't work well for me (DVB-T), but Kenneth has some decent looking code to do something similar for DVB.  I'm assuming that the HDTV is likely to be quite similar (seeing how these things work).

Why not drop the guy a line, have a peek at his latest code and see if we can't consolidate some of it?  It would sure be very useful if bug fixes in one were effectively fixing problems in the other as well!

Thanks  

Ed W