Re: the 'at-boot' package method: a change of direction

I don't have time to reply to all of Jonathan's message, but
I thought I would heave in one comment...

(For those who don't know, Jonathan made very significant
contributions to the initial Arusha ideas, before he
abandoned Glasgow for the bright city lights...)

> ... : perhaps we should jump back a moment and ask
> ourselves whether a 'package' is the right unit of
> abstraction for worrying about starting and stopping
> daemons. Perhaps we need a new class of object, like
> 'services' which encapsulates this.

One possibility is that our 'packages' really *are*
'services'.  It may be slightly weird to think that when you
type

    ark package reveal emacs--20.7

you are really setting up an "Emacs service", rather than
"installing an Emacs package", but the idea makes sense.

This *may* turn out to be an Arusha win...  Most other
systems in our domain are host-based: the goal is to get one
or more *hosts* to a known state.  We instead approach the
question with site-at-a-time thinking, "What services do I
need for my site?"  It's probably six-of-one, half-dozen-of-
the-other, but our approach *may* get us higher up the
abstraction tree more quickly.

Will

PS: Jonathan, I have *not* cut another ARK2 snapshot this
weekend, simply because the motorcycle is still in small
pieces on the garage floor and barely runs at all :-)  Yell
if I'm holding you back...