[mythtv] More scheduling scheduler

[Max Barry]

>  > Date: Mon, 17 Apr 2006 20:30:24 +1000
>> From: Paul Andreassen <paulx[at]andreassen.com.au>
> 
>> > SUGGESTION #3: In "Setup," soft pads should be specified in minutes, not
>> > seconds. Soft pads are "just in case" buffers, for which it's not
>> > important to specify a time that is precise to the second.
> 
>> I believed some people would only require a small pad. Some set top box take
>> 10 seconds to change channels and remove their osd.
> 
> Certainly when I was trying to do things like this on a TiVo, I wished
> I had 30-second granularity (instead of just 1-minute) to try to split
> the difference for shows I knew would run only a few seconds long or
> start only a few seconds early---a full minute was too long for one
> and would chop the beginning off the next, etc.

IMHO this is *not* a situation that calls for a soft buffer. The key
word is "knew" -- if you *know* a show varies predictably from the guide
data, you should correct it with a hard buffer. Because if you use a
soft one, it might be dropped, and then you miss some of your show.

Soft buffers are "just in case" -- insurance against unpredictable time
shifts. If you try to use them to capture time that you *know* is part
of the program, sooner or later you'll miss some.

But I do agree it would be nice if hard buffers could be configured
per-second. I also tend to think that a 1-minute conflict (or a
10-second one) shouldn't prevent the rest of your 1-hour show from being
recorded. But that's another issue. :)

> By the way---if we wind up in a situation where doing the right thing
> is contingent on "don't bump something to a another tuner" vs "please
> bump to record everything I asked for", I'd like to put in a bid for
> making this configurable. Since all of my tuners are identical, I'm
> in the "please use as many tuners as you'd like" camp, but I know lots
> of people have a rather heterogeneous set and would be in the "please
> don't swap between tuners" camp. I don't know if the current tuner
> priorities are sufficient for making this distinction or not.

My guess (please correct me if I'm wrong, anyone who knows better) is
that as a first pass we'll end up with a new soft pad system that does
bump to idle tuners by default (and can't be made to do otherwise), plus
the old pre-roll/post-roll that doesn't (and also can't be made to do
otherwise). So users will have a kind of choice, in that they can use
the old pre-roll/post-roll or the new softpad.

But in the longer-term, the exact "hardness" of softpad -- i.e. whether
MythTV moves recordings to later recordings to preserve the soft
padding, and whether it uses idle tuners -- could hopefully be
user-configured, too. And this might even remove the need for the old
pre-roll/post-roll.

> P.S. Exactly how many ways -are- there of adding a buffer to the
> beginning or end of a recording, in either the existing mechanism
> or the one being discussed? I'm trying to come up with better
> terminology, but it would probably help to start with a really
> clear listing of what each one does... (I think better terminology
> would also make the discussion and the code easier to follow. Even
> saying "type 1 padding" vs "type 2 padding" might be better, if there
> was a single clear and unambiguous table of what those meant...") 

There are three ways of adding buffers: two current plus one proposed.

1. Hard padding: This is recording-specific. You go in, you say "End
late by 3 minutes" or whatever, and that 3 minutes is then considered an
essential part of the show. MythTV will record it or throw a conflict.

You can't currently set a global default value for these hard pads, but
I think one might be in the works.

2. Pre-roll/post-roll (also known as "overrecord"): A global, very soft
buffer. It's applied to all programs, but only if it can be done so
without affecting anything else. That is, MythTV won't move reassign the
recording to an earlier/later showing, or use an idle tuner.

3. (Proposed) Softpad: A global, mostly-soft buffer. It's applied to all
programs, and recordings will be moved to earlier/later showings, or
assigned to an idle tuner, if that helps preserve the buffer. It will be
dropped rather than cause a conflict.

Hope this helps,

Max.