Gentoo Archives: gentoo-amd64

From: Marc Joliet <marcec@×××.de>
To: gentoo-amd64@l.g.o
Subject: Re: [gentoo-amd64] Systemd migration: opinion and questions
Date: Wed, 25 Feb 2015 20:28:49
Message-Id: 20150225212831.03290c80@marcec.fritz.box
In Reply to: Re: [gentoo-amd64] Systemd migration: opinion and questions by Frank Peters
1 Am Tue, 24 Feb 2015 16:51:11 -0500
2 schrieb Frank Peters <frank.peters@×××××××.net>:
3
4 > On Tue, 24 Feb 2015 21:15:45 +0100
5 > Marc Joliet <marcec@×××.de> wrote:
6 >
7 > >
8 > > I liked systemd enough to want to switch my desktop to it before I
9 > > managed to finish writing this. And holy crap, Duncan was right: systemd is
10 > > *so* *fast* on an SSD, it's just not funny. It takes *3 seconds* after the
11 > > kernel boots for me to get a login screen
12 > >
13 >
14 > It requires about just as long (3 secs) on my desktop machine *without*
15 > systemd by just using my custom bash init script (no sysvinit or openrc either).
16 > Furthermore, I use no SSD.
17
18 FWIW, I was implicitly comparing systemd to OpenRC, which I was running
19 previously (hell, I started using it in April 2009 when it was at version
20 0.4.3-r1, a bit more than than two years before it was first marked stable).
21 What I was trying to say was that the speed improvement after installing an SSD
22 was noticeable, but not overly so, whereas systemd exploits it to a much higher
23 degree (as Duncan mentioned a while back, which is why I referred to him).
24
25 And just to make it clear: I'm not bashing OpenRC. It was a really nice init
26 system in the years that I used it (I especially grew fond of netifrc, and am
27 glad that netctl is similarly declarative), but systemd serves my purposes
28 better now.
29
30 > But there's also no login required so it's probably even faster -- and
31 > there's also no long list of permanently running daemons as well.
32 > (After a boot, ps ax shows a very sparse process list.)
33
34 A system that barely runs anything will boot quickly no matter whether it's on
35 an SSD or on a hard disk. Though your point still stands: you run a very
36 minimal system, so you see no need for anything to manage it. Put differently:
37 why try to manage complexity where there is none?
38
39 > For me, systemd is totally unnecessary, excessively burdensome, and
40 > highly obfuscating to a complete control and understanding of my
41 > system.
42 >
43 > To each his own.
44
45 Exactly. I fully respect your wish to control your system totally :) .
46 Something like systemd would just get in the way.
47
48 > If you want/need it, fine. But don't expect me,
49 > or everyone else, to slavishly follow.
50
51 Please don't start with this "us vs. them" crap, it's wholly unnecessary. We
52 are not antagonists, and nobody is trying to enslave anybody.
53
54 > As has been more or less officially stated, systemd is an attempt to
55 > provide a uniform and monolithic kernel-user space interface FOR THE
56 > BENEFIT OF POTENTIAL (COMMERCIAL) SOFTWARE DEVELOPERS. RedHat wants
57 > Linux to be just like MS Windows so that *they* (RedHat) can be just
58 > like MS Windows.
59
60 I've come to a different conclusion regarding the motives of the systemd
61 developers (unification, yes, but to the benefit of as many Linux users as
62 possible, and not just for commercial interests).
63
64 But again, we both want different things from our systems, and that's *fine*.
65 So what if I like systemd and you don't? Then I'll use it, and you won't.
66 There's no need to turn this into a conflict.
67
68 [...]
69 > Frank Peters
70
71 Greetings
72 --
73 Marc Joliet
74 --
75 "People who think they know everything really annoy those of us who know we
76 don't" - Bjarne Stroustrup