D-Bus User Bus

[Lennart Poettering]

On Wed, 19.05.10 22:06, Havoc Pennington (hp at pobox.com) wrote:

> With the per-user bus you're basically just saying network homedirs
> are unsupported, which is pretty radical. I'm all for radical but know
> what you're getting into...

Well, I don't follow that argument. I'd say the current per-session
design does *not* in fact cover anything but the
one-session-per-user-on-one-machine case. And by moving the session bus
to become a user bus we'd at least cover one case more: the
multiple-local-sessions-by-the-same-user case. But we'd leave out in the
dark the network case, and both schemes do not differ on that in any way.

Or with other words: for the network case nothing would change by my
proposal. The semi-working code in gconfd would continue to
semi-work.

> > Well, I am not too concerned with the display handling stuff being too
> > hard. As I tried to point out multiple times I'd say that already many
> > programs are broken when they are started more than once. With the
> > scheme I proposed we'd just change the default from "broken but claims
> > to work" to "refuses to work".
> 
> Well not all broken is equal. "last wins and overwrites" is a kind of
> minor problem, which some stuff has now. "will not let me log in" /
> "won't start up" is a big problem (if you're affected by it).

Well, as long as this is just about gconf keys. 

I think you should widen the focus here: there are probably more
per-machine background services than per-$HOME background services
around. I can list you more services like rygel, pa, gnome-user-share,
gnome-bluetooth, yadda yadda, that are inherently per-machine than you
can list me services like gconfd that are really per-$HOME. Want to play
that game with me?

> What the X server and dbus-daemon are doing here is like the killall
> at the end of the OS shutdown scripts. Just doing in anything that
> wasn't behaving itself. It works. It worked for the X server to do
> this, and that's why dbus-daemon just copies the same known-working
> model for things that may not connect to X. (Unlike Xlib, providing an
> opt-out, of course.)

We should move to a model where privileged code does that
clean-up. (hint, hint: systemd!) Everything else is just playing games
of manual garbage collection.

Lennart

-- 
Lennart Poettering                        Red Hat, Inc.
lennart [at] poettering [dot] net
http://0pointer.net/lennart/           GnuPG 0x1A015CC4