I know some people are going to point and laugh because I'm running an
old version.  Personally, I don't care.  But I'm explaining why, here,
because I want this patch to be considered seriously regardless, even
though I won't have a tested version for trunk before the feature
freeze window closes.  And I'm doing it here because having this
hashed out on the list is just a distraction.

Note that, once 0.22 comes out, I'll probably transition to it.
Eventually.  But not today.

So why am I running such an old version, namely 0.18.1?  Because it
works for me:
(a) This system is used as part of a research project.  It's in
    production.  There is a ton of scaffolding written around it.
    Disturbing that could disturb a lot of other things.
(b) I only ever record SD, never HD, and display it using a PVR-350
    on an actual interlaced, analog, SD CRT.  Later versions tend to
    want to speak OpenGL and use widescreen themes.  Now that 0.22
    will support VDPAU, and hence perhaps deinterlacers that look as
    good as my native interlaced, I might start running it.  But only
    if it still supports my existing SDTV setup, of course, if I can
    find a card that can output interlaced and looks as good.
(c) Upgrading MythTV is very tricky for a production system, since:
    (1) You can't ever go backwards.
    (2) You need real hardware to test with.
    (3) Some bugs are intermittent and won't manifest immediately.
    This means, by the time you know you're in trouble, you're
    screwed, unless you'd like to dump everything Myth's recorded
    since the upgrade.  Or unless you have a completely separate set
    of hardware (including tuners! and cable boxes!) to run your test
    with.  So upgrading is only worth the pain and potential hassles
    if it offers clear advantages.  For what -I- do, so far, 0.18.1's
    been good enough.  Sometime soon, I'll jump to 0.22 when I can do
    so without disturbing my production setup and then test it very
    thoroughly before cutting over.  That day has not yet arrived.
