[mythtv-users] is mythtv smart enough to do this(overlap/back-to-back) with recordings?

[chris at cpr.homelinux.net]

On Sat, Jul 29, 2006 at 07:56:01PM +1200, Steve Hodge wrote:
> On 7/29/06, Michael T. Dean <mtdean at thirdcontact.com> wrote:
> > So, once someone installs a second capture card, we're not talking about
> > a single tuner writing all of both recordings, anymore.  If the system
> > includes one hardware encoder and one software encoder or one encoder
> > (hardware or software) and one digital tuner (or, worse, one of all
> > three), how can data from two different types of encoders get written to
> > the same file?
> Who is suggesting it should be? If two adjacent programs are being
> recorded on two different tuners then everything's fine. What's your
> point exactly?

As I understand the argument, the scenario is as follows:

pre/post padding: 2 minutes
show 1: 1700-1730, channel 38, priority 5
show 2: 1700-1730, channel 40, priority 1
show 3: 1730-1800, channel 51, priority 5
show 4: 1730-1800, channel 38, priority 1
tuner 1: high quality
tuner 2: low quality

The issue is what to do between 1728 and 1732 when the channel 38 
switches from the higher priority recording to the lower priority 
recording.  Do we force channel 38 to stay on one tuner for the 
whole hour so that we can duplicate the 4 minutes of overlap and 
create two monolithic recordings?  If so, which tuner do we use?  
Would the answer be different if the shows had different priorities 
favouring one half hour or the other?  What if we had a rediculous 
number of sequential shows on channel 38?

The simple solution is to say "this is too complicated" and post a 
conflict.  Shows #2 and #4 both get dropped.  Show #1 gets tuner #1 
and show #2 gets tuner #2 and both record for the full 34 minutes.

A slightly more enlightened solution says "I can't give you the 
pre/post padding, but if none of these shows can be rescheduled 
then I'll do a hard swap between the tuners at 1732."  (I chose the 
late switch on the assumption that it's better to miss the start of 
a show than the end of a show.)

Another variation would be to truncate show #2 so that #3 can be 
recorded in its entirety on tuner #2 (the question is whether a 
complete show is better than a high-quality show or vice-versa), in 
which case tuner #1 doesn't switch and the feed can be duplicated 
from 1728 through 1732 so that show #4 has a pre-pad.  Everybody 
wins here except show #2.

A *REALLY* fuzzy solution would split tuner #1 at 1728 and then 
swap tuners at 1731.  The 3-minute fragment from tuner #1 would 
then be transcoded to match the output of tuner #2 and the parts 
that make up show #4 assembled during post-processing.  In this 
case, Shows #1 and #4 are complete, show #2 loses half the 
post-pad, and show #3 starts a minute late.

> Again, that's not the implementation most people here are talking
> about. We are talking about a single tuner writing two adjacent
> recordings to two files and including the overlap in both files.

That's what *you* are talking about.  Dean (?) is saying that 
although your simple example has a simple solution, that solution 
doesn't scale.  The more constraints you put on a situation ("two 
shows on the same channel on the same tuner") the less practical 
the solution becomes.