[mythtv-users] scheduler settling with conflicts where they are resolvable

[Brian J. Murrell]

On 11-09-08 11:13 PM, Ronald Frazier wrote:
> 
> But you see, it DIDN'T give up too soon from it's point of view.

Of course.  That's obvious.  But that doesn't make it any less a bug.

> So you think choosing channel 84-* instead of 109-* will fix it?

I don't just think it.  I manually overrode the scheduler and it did fix
it and it was "best fit".

> But
> the thing is, myth may be looking at those channels, seeing a tie, and
> (since one of them has to win the tiebreaker) deciding that 109-* is
> slightly higher priority.

How can it be "higher priority" if it was a tie.  Either it's a tie or
it's not.  If what you mean is that since it was a tie the scheduler had
to just choose one to try to fit the schedule, then I would agree with you.

But since that choice of "tie-breaker" left conflicts it should try
breaking the tie in the other direction.  Since it's a *tie* as you
describe it, which way it's broken is arbitrary and since the choice it
made left a schedule with conflicts, it has to try the other channel to
see if it can leave a schedule without conflicts.

Ultimately, breaking a tie one way and leaving conflicts where breaking
it the other way is a bug, IMHO.

> I'm not saying it's the right or ideal way to handle this. Indeed, it
> would be better if the scheduler could someday be redesigned to better
> understand that when there's a tie it shouldn't show preference to one
> arrangement at the expense of missing another recording.

Exactly.

> However, I'm
> just trying to help you understand why (I think) myth did what it
> did....what it was "thinking" when it made that less-than-optimal
> decision.

And that's fair.  What I was trying to do was demonstrate that there is
a bug in the scheduler, not so much looking for an explanation why the
bug is there, although I sincerely appreciate your efforts.

This bug seems to be worse than just not optimally scheduling though.

At the time, that the scheduler gave up and left a schedule with
conflicts, there were more tuners available to record those conflict and
it just didn't even bother trying to schedule on them -- not that I
would consider using additional tuners to be optimal considering a good
fit can be achieved on the single digital multirec tuner.

This gets even more interesting, FWIW.  If I remove the manual override
and let the scheduler decide, and quit leaving conflicts and available
tuners, there are many more conflicts elsewhere in the schedule, again,
leading me to believe that the scheduler is giving up way to quickly for
some unknown (read: bug) reason.  Even with the override and the
schedule 98% good, I am still left with a conflict while there are idle
tuners still available.

b.


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 262 bytes
Desc: OpenPGP digital signature
Url : http://www.mythtv.org/pipermail/mythtv-users/attachments/20110909/1e4628c4/attachment.bin