:: Re: [DNG] There seems to be some st…
トップ ページ
このメッセージを削除
このメッセージに返信
著者: tito
日付:  
To: dng
題目: Re: [DNG] There seems to be some strong disagreement in Debian regarding usrmerge
On Tue, 2 Jan 2024 01:27:19 +0100
Lorenz <lorenzo.ru.g@???> wrote:

> Il giorno lun 1 gen 2024 alle ore 18:15 tito via Dng
> <dng@???> ha scritto:
>
> > This list cares because the removal of sysv-support in systemd
> > to force all packages that do not have a .service file to create one
> > will be the prelude to the removal by the package maintainers
> > of their sysv init scripts without the need of ctte or vote
> > but just by the law of maximum laziness (common
> > excuses we have already heard are: cannot maintain
> > 2 init systems, sysvinit is not used by debian anyway,
> > systemd is the defacto standard, cannot test etc. etc.) .
>
> Ok, possible, but is there any evidence that the removal of sysvinit
> scripts that we are observing follows this scheme?
>
> two counter examples here:
>
> 1)
> In Debian there were 314 packages shipping a sysv init scripts without
> a .service file (June 2023 [1]).
> On my sistem (Debian unstable) I have:
> # apt-file search /etc/init.d/ | wc -l
> 1251
> So (1251 - 314=) 937 packages in Debian already ship both a systemd
> and a sysv services before this systemd drop-sysv-generator thing; according
> to your reasoning above, why are they still keeping a sysv script?
>
> 2)
> truly lazy maintainers would not bother to drop the init script and just add a
> copy of the service file generated with the systemd-generator.
> Maintaners perfers to not touch things that do not bother them, so unless
> they receive a stream of annyoing bug about an init script (and that's
> rarely the case)
> they are not going to consider the sysv script a burden.
>
>
> Also, you can verify your idea by looking at already resolved bugs
> (~80) in [1]; at the
> end of the bug there is the changelog so you can check if the removal is
> happening there.
> I had a look at a sample and I don't see a pattern of removal.
>
> My impression is that there is a group of maintainers close to systemd
> that are indeed
> removing init scripts from packages that they maintain[2];
> unfortunately they maintain some
> very relevant packages (like rsyslog, mdadm, udev ..) so sysv users
> are going to feel very
> bad about this. But this is not a trend that will continue untill
> there are no sysv script left,
> once this group remove all initscripts in packages they control, the
> removal will stop.
>
> I dont' think there will be a removal of something like 500 or 1200
> scripts, but it's a guess:
> I don't have an estimate of this number (since the task is time
> consuming and not
> relevant for runit) but I think it is relevant to figure out the right
> solution for sysv users.
> You are going to waste a lot of time if you develop a solution that
> scale up to 1200 but
> end up with 50 or 100 scripts removed; the opposite is also true.
>
> [1] https://bugs.debian.org/cgi-bin/pkgreport.cgi?users=bluca@debian.org;tag=missing-systemd-service
> [2] and it's unrelated to systemd dropping support for sysvinit scripts
>
> Lorenzo
>
> > Due to the fact that there are some 1200
> > packages with initscripts in daedalus it could be a huge
> > task that the devuan community probably cannot
> > accomplish to fork them all.
> > So we are thinking about a way to automate
> > or overcome this future problem with devuan's
> > limited manpower and infrastructure before
> > it will become an actual problem.
> >
> > Ciao,
> > Tito


A counter example: bug 1058701

Package: ftp.debian.org
Severity: normal
X-Debbugs-Cc: mario.limonciello@???

The pm-utils package hasn't had any activity in Debian since 2019.
All bugs have been ignored.

Upstream [1] has been dead since 2010.

__Modern userspace uses systemd to perform suspend/resume instead.__

and suddenly package was removed without even contacting the maintainer:

We believe that the bug you reported is now fixed; the following
package(s) have been removed from unstable:

pm-utils | 1.4.1-19 | source, all

------------------- Reason -------------------
RoQA; dead upstream
----------------------------------------------

The package was later re-added after maintainer complained,
but the synthesis of this story is in the last sentence of the bug report:

Modern userspace uses systemd to perform (put whatever you want here) instead.

This meme is widely spread nowadays and many people just don't care
about init system diversity and are ready to accept it.
So as the ancient Romans said:

" Si vis pacem para bellum"

Ciao,
Tito