[mythtv] Re: Scheduler behavior, why?

Brad Templeton brad+mydev at templetons.com
Sat Feb 12 22:10:27 UTC 2005


On Sat, Feb 12, 2005 at 04:51:10PM -0500, Dan Christensen wrote:
> In the case where the show that would be delayed has higher priority,
> it is less clear whether the user would want this to happen.  So it
> makes sense that this case is controlled by a user option.

This is a hard problem, but one change I would vote for would be
to have resked-higher on by default.   That's an expression, in my
view, of the "hard disk video" philosophy -- that you are no longer
supposed to care very much when a show is on.

As such, if it's a choice between delaying a show's recording (which
you are not supposed to care about) and not getting a show at all,
that strikes me as a no-brainer for the default -- if you buy into the
philosophy.

Of course, no philosophy is always true, there are shows where it makes
a difference when you record them, but perhaps a single scalar priority
number isn't the answer to this problem.   However, I also believe
that there is a lot of virtue in keeping the UI simple, so the single
scalar may already be too complex (since so many things feed into it.)

Designing from scratch, I would start with a basic priority system and
then add an "Unusual events" screen, where the user is shown a list of
scheduling "anomolies" that are new since the last time they looked at
the scheduling anomolies list.  Anomolies would include "A program will
be recorded later than it could be" or "A show will not be recorded."
and anything else.   With a difference between a delay of a couple of
hours and one of several days.

Then the UI would let the user make choices and learn from them.  Over
time, it would learn and there would be no more anomolies, and the
alert that says they are present would leave the menus.

But this remains a hard problem, and different PVRs treat it differently
because nobody has come up with an ideal solution -- other than lots of
tuners.


More information about the mythtv-dev mailing list