Handling the shutdown of the mountee in non-transparent mode

[Sergiu Ivanov]

Hello,

On Sun, Aug 16, 2009 at 11:48:31PM +0200, olafBuddenhagen@gmx.net wrote:
> On Mon, Aug 03, 2009 at 09:31:53PM +0300, Sergiu Ivanov wrote:
> 
> > I'm inclined to think that a non-transparent unionmount should go away
> > when the mountee has been shut down, too, because unionmount makes
> > little sense without the mountee, be a transparent or a
> > non-transparent one.
> 
> Well, depends how you look at it. One way is to say, "as soon as we have
> a mount, we are acting as unionmount, and focus on that."
> 
> But that's not the only option. In the non-transparent case, one could
> also see the mount as just another part of the union. (In fact, it would
> be possible to do several mounts in this case...) And if we look at it
> like that, I'm not sure whether it's best to exit when the mountee (or
> one of the mountees) goes away?...
 
It's not the first time when I think that such treatment of the
non-transparent case is most reasonable.  However, when I stop
focusing on unionmount functionality only, it somehow brings about
thoughts about having *multiple* translator union-mounted.  It seems
to me that we should pay attention to making this possible.

If we do keep to this line of thought, unionfs shouldn't indeed go
away when the mountee has been shut down.

Do I understand it right that you are suggesting to modify the
corresponding patch in such a way that unionfs should go away on
mountee shutdown only in transparent mode?

Regards,
scolobb