Elogind v239.1 Devuan/Debian combined source

Ian Campbell ijc at hellion.org.uk
Thu Nov 15 09:14:53 GMT 2018


On Wed, 2018-11-14 at 07:16 +0100, Andreas Messer wrote:
> On Tue, Nov 13, 2018 at 07:47:52PM +0000, Mark Hindley wrote:
> > On Tue, Nov 13, 2018 at 07:29:21PM +0000, Peter Maydell wrote:
> > > So if you hand-hack these PAM files to refer to pam_elogind then
> > > do things work without the symlink? This would at least tell us if
> > > this is all we need to fix in these other packages, or if there are
> > > other places that are using pam_systemd.so directly somehow.
> > 
> > I think I tried that first and it only worked partially. To get the ability to
> > reboot/poweroff from the desktop I think the syslink was required. So I suspect
> > pam_systemd is compiled in. There are quite a few rdepends.
> 
> I consider any application which directly and explicitly loads a pam
> module as broken. This is not the way how it was intended to work because
> the intentation of /etc/pam.d is, that a site admin can configure it, no
> compiled in forced magic.

Indeed, it does seem rather odd for anything outside the pam machinery
itself or something closely related to the management of a pam backend
to be linking directly against a pam_foo.so module (whether the systemd
one or another).

I wonder if the need to do so (assuming there really is one) is an
obscure enough case that a lintian warning would be justified? Even if
it were very low level warning the next (Debian) archive-wide lintian
run would give a nice list of candidates to inspect.

Ian.





More information about the Debian-init-diversity mailing list