remove systemd while debian runs on systemd

[Axel Beckert]

Dear Sabrina,

thanks for updating the subject.

On Mon, Apr 04, 2022 at 10:51:00AM +0200, Sabrina wrote:
> I can be transparent with my toughs as they are not that superficial but
> seemingly not obvious;

Not obvious at all. (But that itself doesn't seem to have been
obvious though, obviously. ;-)

> On 4/4/22 10:10, Axel Beckert wrote:
> > On Mon, Apr 04, 2022 at 08:10:51AM +0200, Sabrina wrote:
> > > lol. when i installed the package sysvinit-core recently, apt would
> > > deinstall systemd immediatley.

This sounded to me as if you tried that for the very first time.

> > Of course. That's expected and what most people, who do that, actually
> > want.
> 
> Before the last version upgrade you could easily install the sysV packages,
> do your config, reboot and then remove systemd.

This information (i.e. that you did switching from systemd to sysvinit
in the past and that you consider this to be a change of behaviour)
was clearly missing in your previous mail.

What's still missing: On which Debian release did you do that? Stable?
Oldstable? Testing? Unstable? If you want to talk about changed
dependencies, that's a minimum.

> When I read that apt will remove systemd, I naturally thought;  well how
> nice is that! It must remove systemd after I reboot. What a service to
> init-diversity!

Is this meant to be ironic?

> I don't know what most people want, but I enjoyed it how it used to be;
> Reboot and then remove systemd _after_ the system runs with
> sysvinit*.

This should be still possible. AFAIK there's in no release of Debian
(nor in Sid or Testing) a package relation which forbids that.

But without further information (which Debian release, which packages
were installed beforehand, etc.), I can only expect that this solely
depends on the fact if any other package than systemd-sysv (with which
sysvinit-core conflicts) depends on systemd or not. A typical example for
such a package would be systemd-timesyncd or libpam-systemd.

But if systemd-sysv is the only package which depends on systemd and
systemd is marked as automatically installed through dependencies, the
behaviour you've described is to be expected despite being suboptimal.

So I currently don't expect any change in packaging to be the cause of
the behaviour you've seen but solely the combination of installed
packages at the time of the switch.

> Removing systemd while the system runs on systemd is quite a  surprising
> logic..

Correct. Except that this cannot be expressed properly in dependencies
and hence is usually the sysadmin's responsibility. All non-service
packages whose removal should be done while they're they're currently
running (e.g. systemd, linux-image-*) need to check this on their own.
At least the linux-image-* packages do that. Not sure about systemd.

> > Why do you full-quote spam? Please stop doing that.
> 
> Thank you for the comment: Because I was and am annoyed about all the spam
> in my inbox,

Great. Now you have just trained our all our spam filters that some
spam on this list actually might not be spam. Thanks!

> coming in trough that mailing list.

IMHO spam filtering is primarily a local responsibility because it
partially is a very individual job (some newsletters are wanted by
some while others consider it spam, etc.). It's the job of either of
your Mail Delivery Agent (MDA; e.g. procmail) or your Mail User Agent
(MUA) in case of not having control over your MDA. A mailing list
hosting mail server IMHO should be rather conservative in witholding
or even discarding messages automatically.

		Regards, Axel
-- 
 ,''`.  |  Axel Beckert <abe at debian.org>, https://people.debian.org/~abe/
: :' :  |  Debian Developer, ftp.ch.debian.org Admin
`. `'   |  4096R: 2517 B724 C5F6 CA99 5329  6E61 2FF9 CD59 6126 16B5
  `-    |  1024D: F067 EA27 26B9 C3FC 1486  202E C09E 1D89 9593 0EDE