1 |
On Mon, Feb 17, 2014 at 11:52 AM, Andrew Savchenko <bircoph@×××××.com> wrote: |
2 |
> On Sun, 16 Feb 2014 15:16:36 -0600 Canek Peláez Valdés wrote: |
3 |
>> On Sun, Feb 16, 2014 at 2:58 PM, Volker Armin Hemmann |
4 |
>> <volkerarmin@××××××××××.com> wrote: |
5 |
>> > Am 16.02.2014 21:08, schrieb Canek Peláez Valdés: |
6 |
>> >> On Sun, Feb 16, 2014 at 12:59 PM, Volker Armin Hemmann |
7 |
>> >> <volkerarmin@××××××××××.com> wrote: |
8 |
>> >> [ snip ] |
9 |
>> >>> or it is an idiotic decision. Because features means complexity. |
10 |
>> >> Yeah, like the kernel. |
11 |
>> >> |
12 |
>> >>> Complexity means bugs. |
13 |
>> >> Bugs get reported, bugs get fixes. Life goes on. |
14 |
>> |
15 |
>> You didn't answered this, did you? |
16 |
> |
17 |
> Bugs are different. |
18 |
|
19 |
Bugs are bugs, period. And they get reported and fixed. |
20 |
|
21 |
> Bugs in the critical system components are |
22 |
> critical to the whole system. |
23 |
|
24 |
Yeah, that's why we have unit testing and QA teams and stable and |
25 |
unstable releases, etc. |
26 |
|
27 |
> If Libreoffice or browser |
28 |
> segfaults, some data may be lost and inconvenience created, but the |
29 |
> system will continue to run. If PID 1 segfaults — everything is |
30 |
> lost, you have a kernel panic. |
31 |
|
32 |
And the world will end? The same happens if the kernel has an error. |
33 |
|
34 |
> That's why critical components should |
35 |
> be as simple and clean as possible. |
36 |
|
37 |
Like the kernel? You call that "simple"? |
38 |
|
39 |
I'm sorry, but you are (IMO) wrong: critical components should be |
40 |
thoroughly tested and debugged, and have integrated unit testing, and |
41 |
a large enough group of volunteers to test new releases before they go |
42 |
into the general public. |
43 |
|
44 |
> SysVinit code size is about 10 000 lines of code, OpenRC contains |
45 |
> about 13 000 lines, systemd — about 200 000 lines. |
46 |
|
47 |
If you take into account the thousands of shell code that SysV and |
48 |
OpenRC need to fill the functionality of systemd, they use even more. |
49 |
|
50 |
Also, again, systemd have a lot of little binaries, many of them |
51 |
optional. The LOC of PID 1 is actually closer to SysV (although still |
52 |
bigger). |
53 |
|
54 |
> Even assuming |
55 |
> systemd code is as mature as sysvinit or openrc (though I doubt this) |
56 |
> you can calculate probabilities of segfaults yourself easily. |
57 |
|
58 |
I don't care about probabilities; I care about facts: FACT, I've been |
59 |
using systemd since 2010, in several machines, and I haven't had a |
60 |
single segfault. FACT: almost no bug report in systemd involves a |
61 |
segfault in PID 1. |
62 |
|
63 |
>> >> All of them are different tools providing one capability to systemd as |
64 |
>> >> a whole. So systemd is a collection of tools, where each one does one |
65 |
>> >> thing, and it does it well. |
66 |
>> >> |
67 |
>> >> By your definition, systemd perfectly follows "the unix way". |
68 |
>> >> |
69 |
>> > |
70 |
>> > no, it isn't. |
71 |
>> > |
72 |
>> > How are those binaries talk to each other? |
73 |
>> |
74 |
>> dbus, which is about to be integrated into the kernel with kdbus. |
75 |
> |
76 |
> And this is a very, very bad idea. Looks like you don't know matter at |
77 |
> all: to begin with kdbus protocol is NOT compatible dbus and special |
78 |
> converter daemon will be needed to enable dbus to talk to kdbus. |
79 |
|
80 |
kdbus uses a different wire protocol than dbus; but for clients that |
81 |
doesn't matter; libsystemd-dbus will offer a compatibility layer (talk |
82 |
about "standard" and "replaceable"), so if your application uses dbus |
83 |
today, it will work with kdbus. |
84 |
|
85 |
Of course, new applications will take advantage of the new features of kdbus. |
86 |
|
87 |
> The |
88 |
> whole kdbus technology is very questionable itself (and was |
89 |
> forcefully pushed by RH devs), |
90 |
|
91 |
Sorry, but it's you who doesn't know the matter at hand: kdbus was |
92 |
(and is) written by Greg Kroah-Hartman, Linus' right hand, and who |
93 |
works for the Linux Foundation. |
94 |
|
95 |
> anyway it is possible to disable this |
96 |
> stuff in kernel and guess what will be done on my systems. |
97 |
|
98 |
Good for you. Guess what will be done in mine? |
99 |
|
100 |
>> > Looks broken. Broken by design. The worst form of broken. |
101 |
>> |
102 |
>> By your opinion, not others. |
103 |
> |
104 |
> That is not just an opinion. There is a science and experience behind |
105 |
> system's design. |
106 |
|
107 |
Yeah, what do you think about Greg Kroah-Hartman, Linus' right hand, |
108 |
or Keith Packard of X.org fame? None of them works for Red Hat; both |
109 |
of them know more about Unix and Linux than you and me together, and |
110 |
both of them promote systemd. |
111 |
|
112 |
I mean, I myself know a thing or two about computing and Linux, and I |
113 |
promote systemd (and nobody pays me, BTW), but obviously you don't |
114 |
need to believe in my credentials. |
115 |
|
116 |
And, no offense, but I will always give more weight to the words of |
117 |
Greg Kroah-Hartman or Keith Packard (to name only two), instead of a |
118 |
random user in gentoo-user. |
119 |
|
120 |
There are knowledgeable people who are against systemd. But usually |
121 |
they don't give *technical* sound reasons. |
122 |
|
123 |
> And all that science was ignored during systemd |
124 |
> architecture process if there was any at all. |
125 |
|
126 |
You should read systemd-devel and Lennart's blog posts before saying |
127 |
something like that. I did. |
128 |
|
129 |
Regards |
130 |
-- |
131 |
Canek Peláez Valdés |
132 |
Posgrado en Ciencia e Ingeniería de la Computación |
133 |
Universidad Nacional Autónoma de México |