Machine id option to deploy

[William Reade]

On Mon, 2013-04-08 at 21:11 -0300, Gustavo Niemeyer wrote:
> On Mon, Apr 8, 2013 at 8:49 PM, William Reade
> > Time passes; the service constraints change, and the unit of the other
> > service is removed; the unit is now on a machine it "shouldn't" be on,
> > and there's no clear way to determine why.
> 
> This is misleading, as it suggests that it's not important to use
> language that represents intention in a sane way because we can't
> change past events. The reason to use language that represents
> intention is so we can *preserve intention* across future events, and
> extract it, move it over, enrich it, and encode it into pleasant
> textual descriptions.

I clearly misled you, because I agree completely. I believe that
encoding intent allows for very rich and valuable behaviour, and
reproducible stacks, and so on; but I think that these advantages will
be compelling enough to automatically discourage use of --force-machine
in any but the direst need. It's an escape hatch, essentially: the
smarter and safer juju gets, the less it'll be wanted, but right now we
need it to keep our precious test pilots alive and reasonably happy ;).

Cheers
William