[mythtv-users] Events at the beginning of a recording

[David Engel]

On Thu, Feb 09, 2023 at 10:58:28AM +0100, Jan Ceuleers wrote:
> On 09/02/2023 04:36, David Engel wrote:
> >> But I still might!  Someday.  When I have a better solution to the
> >> misuse.  I've never bought the "my guide data is bad" argument and I
> >> still don't.  In the immortal words of Steve Jobs, "You're doing it
> >> wrong?" :)
> >>
> >> Either you care about catching the beginning or ends or your
> >> recordings or you don't.
> > 
> > Also, how many of you that claim you must have long pre- and
> > post-rolls have actually tried using start early/end late instead?  I
> > suspect most of you would get by just fine.
> 
> Just to state my understanding of your point: I believe you refer to
> specifying pre and post-roll in recording rules, is that right? (Perhaps
> I'm not using the correct terminology here; apologies if so).

We settled on the following terminolgoy long ago, to avoid this
confusion.

Start early and end late are options in every recording rule.  They
are specified in minutes and are intended for cases where the probram
is expceted to or might begin early or run long.  The scheduler will
attempt work around these constrains by moving recordings to later of
lower priority tuners to honor the full recording time.  In cases
where the full recording time can't be honored, the scheduler will
report a conflict so the user can make the best informed decision on
what to do.

Pre foll and post roll, are global options.  They are specified in
seconds and are intended to setup and tead down tuners if the tuner is
not already busy.  They are entirely optional and are not guaranteed
to be used.  This is what I contend is misused by many people as a
scheduling feature.

> If so, the reason why I am now using the global pre-roll feature is
> that, in my understanding, it enables back-to-back recordings on the
> same capture card, whereas with recording rule-specified pre/post-rolls
> two capture cards are needed to record back-to-back programs.

This is what doesn't compute for me.  Users say they absolutely have
to have this feature because their guide data is inaccurate and
without it, they'll miss the beginnings or ends of their shows.  Yet,
there's no guarantee this feature will kick in so they still miss
parts of their shows.  Then they go one step further and say without
it being optional , they can't schedule show back to back which,
again, will miss parts of their shows.

Why not use the necessary padding in the first place?  In the case
where there is a conflict, then you can make the choice of which part
of which show to miss that best fits your needs.  Admittedly, MythTV
doesn't do as good of a job of notifying about conflicts as it should.
I've got an incomplete patch that remedies that.

> The other reason is one of convenience: it is easier to specify
> pre/post-roll periods in only one place rather than having to do so in
> each recording rule. Particularly if a bad experience shows that the
> pre/post-roll periods need to be extended. Please keep in mind that the
> guide data unreliability we face here is universal and unpredictable, so
> the pre/post-roll periods need to be applied to all recordings.

Recording, rule templates already make it easy to apply custom,
defaults to all new rules.  It wouldn't be hard to add the ability to
optionally apply template changes to all rules.

> I used to use recording rule-specified pre/post-rolls but switched to
> the global approach for these reasons.
> 
> BTW (and I don't mean to annoy you), a dream feature sidestepping the
> first reason above would be to add "multirec" capabilities to all
> capture cards, by regarding a regular (currently non-multirec capable)
> tuner as one that has as many "multiplexes" as it has channels it can
> tune to (i.e. 1 channel = 1 multiplex), such that a channel can be
> recorded from multiple times at once. This would enable recording
> back-to-back shows, with each recording containing its own pre/post-roll.

I don't fully understand your point here.  However, I have two
comments that might address it.

First, a former developer added a hack long, long ago (yes, this same
discussion has benn going on nearyly 20 years) to quell the masses.
It intentionally avoids putting incompatible recordings back to back
except when it has to be done to avoid conflicts.

Second, MythTV has had the ability for a couple of years now to create
virtual tuners as needed to support overlapping recordings on the same
channel or multiplex.  It currently only kicks in for hard overlaps
but would be trivial to extend for back to back recordings too.

> >> If you do care, you should use the start early/end late options.
> >> That's what they are for(*).  If there is a conflict, wouldn't you
> >> really rather have the scheduler choose a later showing that woh't get
> >> chopped off?  IMO, that's much better hoping that the tetris blocks
> >> just happen to fall in perfect way by accident.  If there is a
> >> conflict, the scheduler will tell you so that YOU can make the best,
> >> informed decision on what to do.
> 
> I'm afraid I don't understand this point. The scheduler cannot know
> whether a show is going to be chopped off because it can only base its
> decisions on the guide data it has at its disposal. If that guide data
> is unreliable all bets are off.

That's exactly my point from a above about not making sense.  Only you
can make the best decision for your cases.  Why do you want to let
MythTV essentially make a random decision?

David
-- 
David Engel
david at istwok.net