Re: Devuan

Quoting Ivan Sergio Borgonovo ([email protected]):

[Devuan:]

> While I do really appreciate the effort to disentangle software, I
> think their approach has very few chances to be relevant, and I doubt
> any of their efforts reached upstream.

I see very good signs, starting with the vdev work.

For server use, I actually question the need for even vdev, and am
aiming for either static /dev again or the minimal managed-dev solution,
Busybox's mdev -- but I respect what they're doing.

Also, I got the clear impression that their repos are either already
fixing the xserver-xorg package dependencies on udev (a wholly owned
subsidiary of systemd, at this point) and fixing the other DE dependency
problems, or soon will.

I don't give a rat's ass about DEs (desktop environments), so I've taken
little interest in whether Devuan's repo fixes the GNOME, MATE,
Cinnamon, KDE, and Razor-qt dependency hairballs:  I personally don't
care.  xserver-xorg, I care more about (but obviously not on servers, my
main concern).

If they merely free Debian-compatible xserver-xorg from hard dependency
on udev (and thus systemd), that would make their efforts worthwhile in
my opinion.  (Of course, what they choose to do with their free time is
their business anyway, not yours or mine.)

> systemd is here and it is going to stay and while it is not perfect it
> is better than most of the alternatives in a much wider application
> space (and yeah worse is better etc...).

I politely but firmly reject the raft of (wrong) assumptions you make
when you say 'better'.

o  If I'd wanted socket activation, I'd have used xinetd.  (Why did the
   world not put all daemons under inetd/xinetd to enjoy the benefits of
   socket activation?  Because socket activation is a terrible idea, the
   solution to a non-problem, in almost all cases, and sysadmins aren't
   impressed with reinventing that bad idea, even if desktop-computing
   weenies like Poettering think otherwise.)

o  If I'd wanted a service supervisor, I'd have picked best of breed and
   configured one to manage the services I wanted locally respawned, and
   not the ones I didn't.

o  If I'd wanted a sprawling set of IPC calls to a pandaemonium of
   Freedesktop.org pieces of bugware (PolKit, udisks2, packagekit, and that
   lot), I'd have gotten the GNOME lobotomy and accepted the dependency
   hairball, the code churn, and the inability to understand how my system
   works.

I adopt architecture that meets _my_ needs.  And, as we are in the Age
of Snowden, I care deeply about knowing exactly what everything does,
knowing why it's there, and agreeing to its necessity rather than having
some desktop-software twits tell me that a huge amount of software I
didn't ask for and don't want is for my own good, especially on servers.

Doing it the other way is not 'better', and I'm not going to stand for
it.



> systemd has been very successful...

To quote Mr. Lincoln, it's the kind of thing that will be enjoyed by
those who enjoy that kind of thing.

> and while many software don't have very good reasons to depend on
> it....

Personally, when I see out-of-control dependency problems, and
brittleness that results from that, I see a very good reason to move in
the other direction.  And inability to use network-manager (from GNOME)
and a couple of DEs is about the world's worst inducement to accept the
dependency hairball.

Your Mileage May Differ.{tm}


> Niche distributions are hard to manage unless they really have something 
> very special that solve some of your very special problems.

This is why I specifically mention that Debian works great with OpenRC
(or sysvinit, upstart, runit, or nosh), despite FUD from various people
including a number of anti-systemd flamers.


> Considering the fuss around systemd I would have expected more people 
> putting work in sundering it and most of the work is not getting upstream.
> 
> Judging from
> 
> http://linuxmafia.com/faq/Debian/openrc-conversion.html
> 
> and
> 
> https://devuan.org/os/packages/list/forked-from-debian
> 
> it really doesn't look you needed to fork a whole distribution.

Quite so.  I absolutely concur.

Far as I can tell, Devuan was a operatic overreaction, and by no means
the most efficient way to deal with the problem.  Yet, (1) their repos
are compatible, so their work effectively _does_ supplement and assist
Debian, and (2) again, it's their spare time to use as they please, when
all is said and done.


> Probably and unfortunately if the "debate" hadn't been so hot, 
> contributing to debian packages to make them more friendly to systemd 
> alternatives wouldn't have been so hard.

Probably.  Possibly.  We'll now never know.

(I had no part in that 'debate', and in viewing through archives see
much heat and little light.)

> But if I really had to look around for a systemd agnostic distro I'd be 
> embarassed but I think I'd pick up gentoo for the second time in my life.

I maintain that Debian itself qualifies.

Yes, you get systemd as a distro default if you install using Official
Debian 8 Jessie's installer -- but when did we cease taking charge of
our systems at the end of installation and modifying it to suit our
needs?  I never did.

Thus the point of
http://linuxmafia.com/faq/Debian/openrc-conversion.html .

It says 'Hey, all you people who claimed Debian 8 Jessie is captive to
systemd are wrong.  Here are two simple package commands to install your
choice of other supported init, and four shell commands to ensure
systemd never comes back.'

> I'm not aware of any vital (debian based) distribution that does offer a 
> clean alternative to systemd. 

_Debian_ offers that.

For those of us who don't need the _entire_ Debian metapackage of 
GNOME, MATE, Cinnamon, KDE, or Razor-qt, and who don't give a damn about
GNOME's NetworkManager application.