1 |
On 03/06/2014 23:01, Marc Stürmer wrote: |
2 |
> Am 03.06.2014 22:14, schrieb Alan McKinnon: |
3 |
>> This whole systemd thing looks awfully like the switch from a hosts file |
4 |
>> to DNS so many years ago. |
5 |
> |
6 |
> Not really. What many people bothers about systemd is that it is getting |
7 |
> more and more |
8 |
> |
9 |
> a) a hard dependancy for software projects, e.g. like GNOME, although |
10 |
> there's no such thing like systemd e.g. on FreeBSD, (MATE instead tries |
11 |
> to be init system agnostic), making it harder to port |
12 |
> |
13 |
> and |
14 |
> |
15 |
> b) that systemd seems to be on a track to reinvent the wheel or so more |
16 |
> and more. |
17 |
> |
18 |
> They are really working on their own DHCP server and client at the |
19 |
> moment, also their own NTP client. Some people coined the term |
20 |
> Lennartware for it, because it's from Lennart Poettering, like also |
21 |
> pulseaudio and avahi. |
22 |
> |
23 |
> Some people are already joking that it wants to become the next Emacs. |
24 |
> |
25 |
> Even Linus Torvalds himself ranted about the attitude of systemd's |
26 |
> developers at the beginning of May this year. |
27 |
|
28 |
|
29 |
|
30 |
I am very familiar with all of that, along with every other regular here |
31 |
- the topic has been discussed to death here repeatedly[1] |
32 |
|
33 |
Also, please don't assign all the evils of the free software world to |
34 |
Lennart. systemd is a large team comprising many more persons than just |
35 |
Lennart. |
36 |
|
37 |
Now, ranting about Lennart ain't gonna solve nuthin'. |
38 |
Writing code will. |
39 |
|
40 |
The systemd devs have an itch to scratch and they are scratching it by |
41 |
writing code. I don't see too many other folks writing anything like the |
42 |
same amount of code and yet for those that want to counter systemd, that |
43 |
is the only thing that will. |
44 |
|
45 |
So where's the alternative code? It's not in SysVinit. |
46 |
|
47 |
For the record, I'm not a systemd fan, I actually run it nowhere. |
48 |
FWIW, I agree with Lennart's ideas as they have sound technical merit in |
49 |
principle. It's his implementation I don't like. |
50 |
|
51 |
Incidentally, what exactly is wrong with systemd writing a dhcp server & |
52 |
client, and an ntp client? Is that project prohibited from writing such |
53 |
software? Are they not allowed to do it? Does it break legal laws? Is |
54 |
there an NDA or non-compete clause in the mix that I'm not aware of? |
55 |
Because they are the only things that could stop systemd from writing |
56 |
such code; without such prohibitions they are free to spend their time |
57 |
doing whatever they damn well please and if that means yet another dhcp |
58 |
implementation, so be it. |
59 |
|
60 |
Let's take that argument a little further: |
61 |
Paul Vixie wrote a dhcp package. Per your logic, it would then not beOK |
62 |
for Roy Marples to write dhcpcd. but he did, and no-one is complaining. |
63 |
|
64 |
I can predict the likely response - systemd will bundle their dhcp and |
65 |
ntp code. So what? It's their time and effort, they are free to choose |
66 |
to spend it how they wish. |
67 |
|
68 |
And you are equally free to write something better if you so choose. |
69 |
|
70 |
|
71 |
|
72 |
|
73 |
[1] For the purpose of this exercise, lets assign a value of "more than |
74 |
10" to "repeatedly" |
75 |
|
76 |
|
77 |
|
78 |
-- |
79 |
Alan McKinnon |
80 |
alan.mckinnon@×××××.com |