:: [DNG] Fw: Ignorant question: What i…
Góra strony
Delete this message
Reply to this message
Autor: Steve Litt
Data:  
Dla: dng
CC: t.glaser
Temat: [DNG] Fw: Ignorant question: What is the point of removing systemd while keeping elogind?
It appears to me that Thorsten Glaser/ //mirabilos, the guy who argued
for continued sysvinit support from day 1 and was shut down by the PTB
(Powers That Be) has created an empty package, called
logind-considered-harmful, that appears to my package management
ignorant mind to make elogind unnecessary. Judging from conversations
on this list, I think this fact might be of interest to some of you.

SteveT


======================================================
Begin forwarded message:

Date: Sat, 6 Jan 2024 22:31:24 +0100 (CET)
From: Thorsten Glaser <t.glaser@???>
To: Matthias Geiger <werdahias@???>
Cc: debian-init-diversity@???
Subject: Re: Ignorant question: What is the point of removing systemd
while keeping elogind?


On Sat, 6 Jan 2024, Matthias Geiger wrote:

> Unfortunately most programs expect something proving a logind
> functionality.


“most” programs?

> For instance apt links against libsystemd-dev. This means we either
> bite the bullet and use libelogind0 (actually libsystemd0 on a sid
> system), or we can't use that (even procps links against
> libsystemd-dev). You need something to provide this functionality
> (i.e. ABI calls) or those programs won't work. Mark has done some
> great work towards getting consolekit2 into debian so that it might
> become a drop-in replacement later on. At the moment using elogind
> (even if it comes form systemd) is our best option (imho). In the
> long run I hope we can use consolekit2 (haven't tested it yet tbh).
> On my openRC system I use


You seem to be conflating libsystemd0 presence with logind
functionality.

The former can be easily satisfied by, well, libsystemd0 presence. It
basically contains a bunch of functions that are used in the systemd
case but not else, for example logging to journal if present (else
syslog), socket activation things (else normal startup), etc. and of
course the systemd communication (which is obviously not used).

Then, there are some programs which ostensibly require logind. I have
heard GNOME is one of them, but I have yet to see any.

I tried elogind, but it inacceptably changed system functionality, so
now I use an empty package that…

Package: logind-considered-harmful
Provides: logind
Conflicts: logind, default-logind, elogind, libpam-elogind,
libpam-systemd

… and I have yet to see anything breaking. So I think the logind
“requirement” is vastly overstated. Sure, some packages think they
require it…

$ wtf -y logind
- Dependencies on “logind [no version]”:
|libpolkit-qt-1-1:amd64 (= 0.112.0-6tarent1.)
|libpolkit-qt5-1-1:amd64 (= 0.113.0-1)
|policykit-1 (= 0.105-31+deb11u1)
- Recommends on “logind [no version]”:
|openssh-server (= 1:8.4p1-5+deb11u2)
|xserver-xorg-core (= 2:1.20.11-1+deb11u6)

… but that does not seem to apply.

> I think at least having those options in debian is worth something


Definitely. But I think not needing them is also worth something.

> Honestly I don't think it'll change that programs like apt link to
> systemd libraries because that would require a major change in the
> linux ecosystem


But that’s okay? These libraries seem to be mostly inert if systemd
is not used.

bye,
//mirabilos
--
15:41⎜<Lo-lan-do:#fusionforge> Somebody write a testsuite for
helloworld :-)