[mythtv] 'stable' tag proposal

[Brad Templeton]

On Mon, Mar 21, 2005 at 08:37:15PM -0500, J. Donavan Stanley wrote:
> something. Hell CVS Myth is a heck of a lot more stable than many
> "stable" branches of other projects I've seen.

I think that the reason for greater frustration is that people find
themselves depending on the tool more than others, and the failures are
thus more "catastrophic."   Now in one sense, it's a bit laughable to
think of missing a TV show as catastrophic, but hear me out.

There are lots of ways tools fail.   For tools where you are doing work
that needs to be saved, most of the time their failures mean you lose a
little bit of work, and consider the tool moderately unstable.  It's very
rare for a failure to lose something entirely, and if such failures are
possible the project is considered highly unstable.  A mail system
that dropped your mail on the ground would be considered unusable by most
people.

So a PVR's curse is that one of its failure modes is to miss recording
a favourite show entirely.   Sometimes this can be recovered by a rerun,
or by bittorrent, but not alwyas.   Something the user expected is "lost"
and they get much more upset.    Conversely a video player like mplayer
might crash half the time and not bother people nearly so much.  The
program is much more unstable but no harm but having to restart comes
out of it.  (Especially if it included a way to remember where it was,
which mplayer doesn't have but could consider having, be handy as myth's
video player too!)   Even an unstable X server is only considered a huge
curse because it possibly kills other programs and makes them lose work.
(though the better programs never lose work even when killed by random
power outages.)

This is a great challenge on programs that work with capturing real
time data and real time events.  Their bar is higher, they must work
much harder on robustness to get the same impression of stability in
the view of the public.