elogind 246.9-1+debian1

[Adam Borowski]

On Mon, Dec 21, 2020 at 07:01:52PM +0000, Mark Hindley wrote:
> On Mon, Dec 21, 2020 at 07:02:37PM +0100, Adam Borowski wrote:
> > Could you please expand these entries?  Neither bug is "please update to the
> > new version".
> > 
> > I'd suggest something like:
> >     * Merge upstream v246.9
> >       + Support ABI of systemd-logind 246 (Closes: #968379).
> >       + Log to syslog rather than kernel log (Closes: #964006).
> 
> OK.

As you've apparently gone to sleep, I've applied this change and uploaded.

> > >   * Change rootlibdir to /lib.
> > 
> > Could you please provide a rationale for this?  Booting the actual system
> > from separate /usr is long since unsupported (must be one fs or mount from
> > initrd) thus early-boot rationales no longer apply; there's a migration off
> > /lib; moving stuff around can cause problems with usrmerge (which is bad IMO
> > but currently the default).
> 
> I originally made the change[1] in response to Devuan #482[2]. Yes, the poster
> was using lvm without an initrd. But it fixed his issue without causing any
> regressions that I am aware of. Although I accept usrmerge is not the default on
> Devuan.

There are two distinct things here:
* dropping support for split / vs /usr without initrd
  + which is something with a good rationale
* symlinking /bin /sbin /lib to /usr/*
  + which is causes damage and is considered unsupported by dpkg's
    maintainers, yet is pushed as default by some folks.

Moving stuff between /usr/lib and /lib is one of many things that can fail
with the latter.  I tested the upgrade and it worked for me, but I'm wary.

> It also mirrors the systemd setup which (still?) has libsystemd0 in /lib.

Ha ha.

> Is emptying /lib a target for bullseye?

Nope.

As using /lib reduces the diff from upstream, I've uploaded as-is for now.


Meow!
-- 
⢀⣴⠾⠻⢶⣦⠀ .--[ Makefile ]
⣾⠁⢠⠒⠀⣿⡁ # beware of races
⢿⡄⠘⠷⠚⠋⠀ all: pillage burn
⠈⠳⣄⠀⠀⠀⠀ `----