[Mageia-dev] Proposal: Deprecate draknetcenter+network init scripts after systemd becomes default.

Colin Guthrie mageia at colin.guthr.ie
Tue Aug 23 12:26:25 CEST 2011

'Twas brillig, and Guillaume Rousse at 23/08/11 10:33 did gyre and gimble:
> On 23/08/2011 10:30, Colin Guthrie wrote:
>> 'Twas brillig, and Michael scherer at 22/08/11 13:14 did gyre and gimble:
>>> On Mon, Aug 22, 2011 at 08:44:01AM -0300, Balcaen John wrote:
>>>> Le Monday 22 August 2011 12:57:23 Guillaume Rousse a écrit :
>>>>> On 22/08/2011 11:57, Colin Guthrie wrote:
>>>>>> I also have it on good authority that many of the features lacking
>>>>>> in NetworkManager (such as bridging configuration) will be
>>>>>> available in the not too distant future and many other more
>>>>>> advanced networking features such as fast-start DHCP,
>>>>>> per-interface DNS, 4-8's DNS fallback and several other nice
>>>>>> features will ultimately be possible too.
>>>>> While I don't care about configuration wizards, I do about
>>>>> initscripts. How are you supposed to configure a server in some
>>>>> automated manner without plain-old configuration files ?
>>>> If i'm not wrong you can still drop plain text files in
>>>> /etc/NetworkManager/system-connections/
>>> Provided you want to do nothing fancy like bridge, vlan and
>>> others stuff that are used by sysadmins.
>> As I said in my initial email, but was not clear. All of these things
>> will be supported in a much nicer way in the near future.

> Well, even if supported, I do feel much more confident in shell scripts
> I can read, understand and easily fix if needed to fit my own needs,
> than in a native binary.

This is a endless argument but one which will ultimately not be
sustainable I fear. I do sympathise, but by the same token, I also want
a modern, fast and efficient system too, so I'm kinda torn. I understand
C code pretty well generally and while I can't just "less" the program
on my machine it's not too hard to get the source code and pick through
it when I need to (which is very rarely in reality) so I think both my
need for curiosity is satisfied and while I accept the "on-site
hackability" does get negatively impacted, the larger number of users
using a standard system should result in less problems overall and thus
less need to hack in the first place... this won't always be true, but
I'm happy with that trade off.

> How would removing initscripts support helps enhancing networkmanager
> integration ?

Because the current philosophy of the Unix legacy is lots of individual
utils from various packages cobbled together with some glue shell
scripting code... and it's dying.

The things that these individual tools implement are a few relatively
simply commands to the kernel and it doesn't make sense to do all this
in shell. It makes much more sense to do all these jobs in efficient
code that runs *quickly* without forking hundreds of times. The code is
still perfectly visible and easily hackable, but now things are much
more robust and efficient.

It's also the case that people *talk* about doing stuff lots, but very
rarely actually *do* it. People have talked about tidying up the init
system for years, and they've talked about improving the networking
support for years but these just go in circles and never really result
in real progress. For the first time in a long time, things are actually
being *done* about this and in doing so we can take advantage of a lot
of the modern features of Linux. It's exciting and it means that the
support in the GUI frontends for network management become much more
standardised and less varying over different distros. The fact that
distros all tweak things and do "homebrew" for network management means
that for tools like network manager to support all distros they have to
do a whole bunch of weird shit to work on RH vs Mandrvia. vs Suse, vs
Ubuntu etc. etc. By ripping out the cruft and replacing it with standard
binaries, you get consistency at the expense of the all the variations
but IMO this is a good thing - everyone benefits from a larger community
working on the same thing - more eyes, less bugs, more features.

So, removing the variations in different distros init methods and
network management tools, is very clearly enabling better network
manager integration.

Hope that explains it a bit (tho' I'm typing quickly because I'm "at
work", so sorry if it doesn't read super clearly!)



Colin Guthrie

Day Job:
  Tribalogic Limited [http://www.tribalogic.net/]
Open Source:
  Mageia Contributor [http://www.mageia.org/]
  PulseAudio Hacker [http://www.pulseaudio.org/]
  Trac Hacker [http://trac.edgewall.org/]

More information about the Mageia-dev mailing list