Bas Wijnen wrote:
> I'll speak for myself here: I don't really care about the init system.
> I am unhappy with the emotions that this debate is causing, but I'm not
> very interested in the technical parts. From what I see on the mailing
> lists, it seems that a few users are very unhappy and they keep bringing
> this up.
Since there continues to be interest on -user about why no DDs are
proposing a GR to overrule this decision, I want to expand on that.
First, you have to understand that every single argument that has been
posted to debian-user about systemd was already hashed out on
debian-devel over a year ago. The discussions about systemd on -devel
went on for at least a full year. It was a major topic at DebConf13,
which included presentations by both upstart and systemd upstream
developers. Then we had the -ctte process which dragged on for quite a
while longer and rehashed most everything all over again. So at this
point, most of us are pretty tired of the subject.
Secondly, Russ Allbrey did an amazing job during the -ctte decision of
weighing systemd vs the alternatives. He was unbiased; he dug deep. It
really cut through the fog. When you see such good work being done,
there is less tendancy to second-guess it, even if you might disagree
with his conclusions. We really appreciate Russ[1].
Thirdly, DDs feel empowered to fix problems. Not because they can upload
packages to Debian, but because they can file bug reports and work with
others to get them fixed. It's what we do. An example: Yesterday, DD
John Goerzen had a really, really bad experience with systemd on his
laptop, which uses an unusual zfs+encryption setup. He ranted, like
anyone would in such a situation:
http://changelog.complete.org/archives/9237-first-impressions-of-systemd
... But then he got on #debian-systemd on IRC and filed several bugs, and
got help to get his system working, and followed up on the bugs with
the details that will let them be reproduced and get fixed.
Just now, he wrote there:
<kini> CosmicRay: glad to see you got some problems resolved :)
<CosmicRay[John]> kini: yes, me too ;-)
<CosmicRay[John]> I plan to post an update. I must say, this is one of
the most helpful communities I've seen in Debian.
<CosmicRay[John]> that is something *huge*.
I'd hope that anyone who has the time and expertise to participate in
1000+ message theads about systemd that dig into the source code and
discuss rather rarified theories of software engineering also feels
empowered to file bug reports and work to get actual problems fixed.
If you do, you will probably feel less need to engage in such threads.
And, if you appreciate this process of how software is improved, you'll
start to, perhaps, become a little bit suspicious that some voting-based
GR process can have as good results overall.
Fourthly, I think that many DDs feel that releasing jessie with systemd
as the default won't make it appreciably harder to revert to
non-systemd-as-default later than it would have been if we stuck with
sysvinit for this release.
Not that it would be easy to ditch systemd. But there's a lot of FUD
going around here about sysvinit support rotting because systemd is the
default, while the fact is that Debian fFreeBSD doesn't have systemd at
all, and all the init scripts will be kept working for that reason if
nothing else. Also, the tech committe decision was that Debian continues
to support multiple inits to the best of our ability[2]. And, the init
scripts are a relatively miniscule portion of the code in Debian, and
don't tend to bit rot much anyway[4].
So most of our concern about being locked into systemd is that desktop
environments are coming to require it, and that systemd-shim may be hard
to keep working in the long term. But desktop environments like Gnome
were already requiring systemd before Debian switched to it; Debian
cannot hold back the tide.
I'd say that the chances of a GR at this point in the release process
are about 1 in 1000. It'd take 5 DDs simulantaneously having a bad day
like John did, or massive evidence of unhappy users. And I mean, hard
statistical evidence of that on eg [3], not a few users posting
arguments against systemd that are often highly slanted and innaccurate
and have in any case been seen over and over again before.
--
see shy jo
[1] Russ was awarded a handcrafted plaque for this at DebConf14.
We have never awarded anyone such a thing before. We really
appreciate Russ!
http://vincentsanders.blogspot.com/2014/08/without-craftsmanship-inspiration-is.html
[2] https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=746715#278
[3] https://qa.debian.org/popcon-graph.php?packages=sysvinit-core+systemd-sysv&show_vote=on&want_legend=on&want_ticks=on&from_date=&to_date=&hlght_date=&date_fmt=%25Y-%25m&beenhere=1
[4] If I still maintained a daemon and was concerned about its init
script bit rotting, I'd write a simple autopkgtest check that the
init script worked properly; we've gotten increasingly good infrastructure
to test such things.
http://www.piware.de/2014/10/running-autopkgtests-in-the-cloud/
Attachment:
signature.asc
Description: Digital signature