On Fri, 2014-09-05 at 16:01 -0700, Steve Langasek wrote: > On Fri, Sep 05, 2014 at 11:48:04PM +0100, Ben Hutchings wrote: > > On Fri, 2014-09-05 at 13:52 -0400, Zack Weinberg wrote: > > > Steve Langasek wrote: > > > > > No, that's not the true package relationship. There's no reason that > > > > you should always get this added service by default when you install > > > > a system with non-systemd init that doesn't need logind. Making this > > > > a recommends would be a workaround for bad metadata in the > > > > libpam-systemd package; we should fix that problem at its source the > > > > right way. > > > > I filed bug #746578 against libpam-systemd back in May; I believe the > > > proposed change (depend on systemd-shim | systemd-sysv rather than the > > > other way around) addresses most if not all of this class of issues. It > > > is currently WONTFIXed. > > [...] > > > It's a bit counter-intuitive to have the default init system second, but > > now that I think about it, I can see that it will do the right thing on > > a jessie installation. > > > Upgrades from wheezy are the problem. Currently, upgrading sysvinit > > should result in installing init and, unless upstart or sysvinit-core is > > already installed, systemd-sysv. But if sysvinit and some rdep of > > libpam-systemd are upgraded at the same time, and the order of > > libpam-systemd's dependencies is switched, APT (or other package > > manager) might consider it preferable to install sysvinit-core and > > systemd-shim. Has this been tested? > > systemd-shim expresses no preference for init system, and is completely > coinstallable with systemd-sysv - and should be a no-op when booting under > systemd because the dbus name is already taken. [...] Sorry, I misread the Breaks field in systemd-shim (the version is higher than the current version of systemd in unstable). If I understand the last changelog entry correctly, systemd-shim can also be built against a later version of systemd and should get the right versioned Breaks. However, systemd-shim/experimental has been built against systemd/unstable on i386, resulting in: Package: systemd-shim Version: 7-2exp1 Breaks: systemd (>= 209) whereas on amd64 (maintainer build) it seems to have been built against systemd/experimental and so has the intended: Package: systemd-shim Version: 7-2exp1 Breaks: systemd (<< 209) Ben. -- Ben Hutchings Experience is directly proportional to the value of equipment destroyed. - Carolyn Scheppner
Attachment:
signature.asc
Description: This is a digitally signed message part