[Pkg-utopia-maintainers] dbus-sesssion questions

[Simon McVittie]

On Mon, 24 Aug 2026 at 12:04:06 +0100, Mark Hindley wrote:
>Currently dbus-user-session
>
>  Depends: dbus-daemon (= 1.16.2-5+b1) | dbus-broker, dbus-session-bus-common (>= 1.16.2-5), libpam-systemd, systemd
>
>To make this work, I think this would need changing to
>
>  Depends: dbus-daemon (= 1.16.2-5+b1) | dbus-broker, dbus-session-bus-common (>= 1.16.2-5), libpam-systemd | openrc-dbus-user-session | other-dbus-user-session(s)

Yes, if you want to rely on dbus-user-session's copy of 
20dbus_xdg-runtime, we'd have to go this route. Please let me know when 
you have a good list of other D-Bus user session implementations - I'd 
be happy to add them in experimental to start with, if you're not sure 
they're all ready in unstable.

Or you could rename openrc-dbus-user-session's copy to something not in 
dbus' namespace, like perhaps 20openrc-dbus-user-session, and rely on 
the fact that they're both idempotent, so if both dbus-user-session and 
openrc-dbus-user-session are installed, the duplication is harmless. 
That would allow dropping the Conflicts without necessarily adding a 
Depends.

>i.e also drop explicit Depends: systemd and rely on the fact that libpam-systemd
>already Depends: systemd for the systemd case.

Yes, I think that would be fine. I don't remember why we have/had the 
separate Depends: systemd, but I see that back in 2015, I originally 
added Depends: systemd instead of the correct Depends: libpam-systemd - 
I don't think it was obvious at that stage that libpam-systemd is the 
right thing to depend on in packages that need a working `systemd 
--user`.

The Recommends on systemd-sysv would also need to turn into 
systemd-sysv|other-dbus-user-session(s), or we could just remove it. It 
was useful to have this Recommends when sysvinit was the default init 
system, but it isn't really helpful now that (a) we have the init 
metapackage forcing installation of some sort of init system, and (b) 
systemd-sysv is the distro-level default anyway.

>Am I correctly interpreting what you mean/intend?

Yes, I think so.

     smcv