[linux-elitists] How about a fork of Jessie without systemd?

Don Marti dmarti at zgp.org
Sun May 15 10:18:13 PDT 2016


begin Greg KH quotation of Sun, May 15, 2016 at 02:14:22AM +0200:
> On Sat, May 14, 2016 at 04:38:49PM -0700, Ed Carp wrote:
> > I'm particularly annoyed by systemd, period. Nothing wrong with the
> > way things were
> 
> You are kidding, right?  Have you ever worked at that level?  Tried to
> manage processes and services in a sane manner?  I've been dealing with
> that problem since my first paid Linux job in 2000, and it didn't get
> finally resolved until systemd.  There were lots of things wrong with
> how things "were" before, perhaps you never worked down there?

I have written both init scripts and systemd unit
files, for server-side packages of comparable
complexity.  From the point of view of the package
maintainer, systemd does have an edge.  You're writing
fewer total lines and telling a well-documented
program what to do, not invoking shell functions that
can run arbitrary stuff.

I also run a "mainstream" desktop distribution (with
systemd) and the thing starts up faster than my Linux
laptops used to.  Hard to break out how much of this
is systemd socket activation, though. (this is where
someone links to the obvious LWN article I should
have cited.)

> > - systemd goes completely against the Unix idea of one
> > small utility doing one thing well, and it just comes across as change
> > for the sake of change as a monument to someone's ego.
> 
> Really, you want to drag this old disputed argument out again?  Tell me
> how your old init system was a "do one thing well" program?

I don't get the "Unix Philosophy" thing.

Posts about "this functionality is broken on systemd
but not OpenRC" are great, but "Unix Philosophy"?
What did that get us?  A bunch of failing vendors in
the 1990s, and the inevitability of Windows NT.

  http://gabriellacoleman.org/blog/?p=1729

The "philosophers" of Unix let themselves be rounded
up and made irrelevant.

If you want to build a better init system, build a
better init system.  But philosophy-based OS advocacy
is a failure.  Designing working software based on
philsophy is like writing real network software based
on the OSI 7-layer burrito model.

(Nothing against Rick's article.
  http://linuxmafia.com/kb/Debian/openrc-conversion.html
I enjoyed reading it.  No philosophy, which I
appreciate.)

> Remember back when Linux users were the ones pushing the boundries of
> things, solving real problems and being happy to handle major changes in
> the quest to making something better?  I sure do.  The real question is
> why have the others on this list, who have been complaining about a
> system library being replaced that they didn't have anything to do with,
> have forgotten about why they started using Linux in the first place.

One cool systemd feature that I want to try is
multiseat:

  https://bluehatrecord.wordpress.com/2015/01/15/multiseat-configuration-in-fedora-21/

Debian is one of the distributions I regularly
use, so I wouldn't mind setting up a system with
an alternate init system if there is something I
want to do that works with OpenRC but not systemd.
Right now I don't see a compelling reason not to use
the "out of the box" init system.

> I mean, if you like openrc so much, why not just move back to solaris,
> there is an "open" version floating around somewhere that desperately
> needs users and developers...
> 
> Or just use Gentoo or slackware if you don't like change and wish to
> remain in the 1990's.

Or if you want to go the other direction, Plan 9 or GNU Hurd.

-- 
Don Marti <dmarti at zgp.org>                   
http://zgp.org/~dmarti/
Are you safe from 3rd-party web tracking?  http://www.aloodo.org/test/


More information about the linux-elitists mailing list