Bug#727708: personal views of Debian users
Hey.
Well not sure whether this is actually welcomed or not,... but since
some people have already started to share their personal feelings about
the debate, I want to do so as well.
I've been using sysvinit for countless years (as most of us)... and I've
tried both, systemd and upstart when the recent discussion began (which
was, I guess, actually sparked indirectly by a post of mine, when I
"asked" whether systemd was now mandatory due to GNOME depending on
it))...
I haven't really looked in depth at OpenRC or other solutions, since
from the descriptions made by other people, I think it's not comparable
to systemd/upstart.
I'm maintaining a large Tier-2 for the LHC Computing Grid... so I guess
I do have "some" ;) experience in what is useful in real life (like most
other people here have of course as well).
Now I guess it doesn't make much sense to repeat all technical arguments
people have already brought up over and over again in this bug, so to
make it short:
>From a technical POV, I'd clearly go for systemd.
Not only are (IMHO) it's core concepts and design superior... it also
seems to provide much more and better features.
Speaking only about Debian GNU/Linux... I'd even go as far, that I'd say
we should in the long term, think about integrating the "other" features
of systemd, like the Journal replacing rsyslog, or perhaps even having
it in the initramfs (well, that is of course something one would really
need to investigate closely)...
In any case we should try to get something like the un-initramfs at
shutdown, which systemd seems to support quite well.
I think however, that a main part (50%) of the question systemd vs.
upstart vs. something-else is not a technical one.
Code, design and features can be improved or added.
I think there is a strong political part in this decision.
- At most upstream projects (the kernel, wayland, X, etc. pp.) people
seem to at least think first about systemd... if they support upstart at
all.
Just look at recent developments like kdbus, which are clearly strongly
"influenced/triggered" by systemd.
So I fear that when going for upstart, Debian might sooner or later sit
on a lone island (next to *buntu's island), having to spend a lot time
to keep things working and adapted to upstart.
- Most other major distros (except *buntu) have decided for systemd,...
so again here,... with upstart we'd sit on a lone island, which
ultimately would lead to many problems for sure.
- In my opinion (and I'm sure some people will agree and others will
contrary): RedHat has proven to be more "neutral" to projects it
"governs" than Canonical.
Actually, many people seem to think,... that Canonical has recently gone
some strange paths, which somehow seem to lead them away from the
community and classical open source ecosystem (just think about the
whole Mir-story).
- With upstart there is the contributor license agreement issue... which
I think is a major political problem.
- Last but not least... there are people (including myself, I guess),...
which are concerned about the Debian/*buntu relationship in several
ways... so having upstart the default init system, would give Canonical
for sure some bit more of influence on Debian (and if it was just by
technical decisions they make upstream).
Of course one can argue, that this kind of influence might now be taken
by RedHat.
As another side note (which is not really a reason against upstart), but
has also some political "impact", I guess...
I really wonder what the decision "systemd vs. upstart in Debian" means
to upstart?
systemd for sure wouldn't bother much, if Debian decides for upstart.
But it seems to me, that if Debian decides for systemd, this could be
the end of upstart itself.
Why?
- *buntu would then permanently be completely alone on the
upstart-island.
- And since Debian packages would then focus on systemd, *buntu would
get proper support for that for free - so why continuing to spend much
efforts just for having an own init-system, which even provides no real
technical benefits?
Actually Canonical *is* known for dropping support or at least active
development for their praised products,... think about bzr.
Some last things:
- While I think there should be a default init-system which all packages
MUST support (which I'd want to be systemd)... others should be allowed
as well.
- I do have a big problem with projects (especially like GNOME) which
sometimes seem to have an agenda of enforcing people to use the
techniques they want. IMHO, open source IS about choice. But reality is
probably, that one cannot do much about it.
- I strongly like the idea of having k/freebsd and other non-Linux
Debians,... and if it is just for diversity.
Whatever decision is taken in the end,...care should be taken, that
these ports can continue to exist.
Best wishes,
Chris.
Reply to: