[mythtv] UK Freeview "Playback"

[David Matthews]

Martin Long wrote:
> I can see the EITpf being an issue, because of the way tuners are
> preallocated  by the scheduler. The fact that a program could begin and end
> at different to scheduled times could create situations where conflicts
> become apparent in realtime, and need to be resolved in realtime instead of
> at the next scheduler run. Also, at any point the extent of the conflict may
> not be known. For example it may not be able to start recording a higher
> prority recording without stopping the existing recording while it still has
> EITpf.

I suspect that there are two classes of users with different 
requirements for the scheduler.  One group schedules many recordings and 
wants the scheduler to avoid conflicts at the expense of possibly losing 
the end of a programme if it overruns.  They would probably not want to 
use EITpf for scheduling even if it was available for them.  There is 
another group, including me, who schedule only a few recordings and 
would prefer to have complete recordings at the risk of a greater chance 
of conflict.

The current scheduler algorithm which tries to minimise the number of 
tuners used caters better for the first group.  I think there is a case 
for a modified algorithm to cater for the second group.  This would 
allocate tuners to recordings dynamically.  The algorithm might work 
something like this.
Start looking for EITpf shortly before the recording is due to start.
Start recording when the show is "present" or at the scheduled time 
depending on how reliable EITpf is.
Continue recording on the current tuner until the show is complete even 
if it overruns.
If a second recording is due to start while the first is recording 
allocate it on a different tuner if that is available otherwise delay 
recording until the first finishes.  It is probably better to lose the 
beginning of a show rather than miss the end of one.  Higher priority 
recordings should probably pre-empt lower priority since the user has 
explicitly said that by setting a priority.

Where EITpf is not available or not reliable soft padding should be 
added to the beginning/end of timings possibly resulting in two 
different tuners being used for back-to-back recordings.  Only if there 
is no tuner available and the beginning of a recording would otherwise 
be lost should the padding be discarded.  Choosing to treat padding in 
this way is different from the current algorithm but from the number of 
times this comes up I think this is how many people expect it to work.

The existing algorithm should be kept for those who are more concerned 
about conflicts but something along the above lines will be needed if 
the precise end time of a show is not known in advance.

> ITV may play their usual dirty tactics of delaying switching the
> signals until after adverts / well into the next timeslot. 

But if they delay it too long wouldn't that have the effect of losing
the beginning of the next programme?  They are more likely to switch
early rather than late since people may sit through the ads before
watching the show they want but will stop watching as soon as the
credits roll.

Once Freeview Playback recorders become generally available the 
broadcasters will want to ensure that EITpf is accurate if that is being 
used to control recording.  They will need to compete with Sky+ and 
there will be an incentive to ensure that people don't lose the 
beginning or end of their favourite show.

David