Re: sysvinit: rc vs. r2d2 bahavior
Miquel van Smoorenburg <miquels@cistron.nl> wrote:
>>> Instead of this here's what r2d2 does:
>>>
>>> 1. Run all K??-scripts in new runlevel, which are not K??-scripts in
>>> the prevlevel. Do this in _descending_ order of the scripts
>>> (stop first, what is started last). Use "stop" as parameter.
>>> 2. Run all S??-scripts in new runlevel, which are not S??-scripts in
>>> the prevlevel. Do this in _ascending_ order of the scripts. Use
>>> "start" as parameter.
> Oh this will break existing setups _horribly_.
It is somewhat different to the actual behavior, but is it really
worse? BTW: What setup is broken by this? Most people don't use the
possibilities, sysvinit gives them but use only three runlevels (0 for
halt, 2 for normal use and 6 for reboot) using update-rc.d defaults.
Such a setup shouldn't be broken by replacing rc with r2d2 or miss I
something?
On the other hand r2d2 is much more intuitive (for me), so someone who
wants to use a more complex sysvinit setup, can do this with less
problems, because it makes it possible to understand what happens.
BTW: neither rc from sysvinit nor from file-rc behave exactly in the
way Debian policy 3.3.1 describes (r2d2 also doesn't ;-).
>>> So my question is the following: Why do we keep this non intuitive
>>> behavior of sysvinit and file-rc instead of intuitive behavior of
>>> r2d2?
> Because it will break your system?
Can you give me an example for this?
I don't see why we should stick on a behavior of rc, which isn't
intuitive or logical instead of changing it.
I don't say, that this change is trivial, but I don't like to stay on
a worse solution when a much better one is possible.
> Really I've explained the backgrounds of the init system a few times
> here and a few times in private email. I'm not going to do it again
> unless someone puts it in a FAQ somewhere.
Maybe you can forward some of these mails to me? I asked for the
second time here now, but didn't get any reaction. Nobody was able to
explain me the ideas behind this strange behavior of rc. Seems that
someone implemented this in a strange way years ago and nobody
understood it but simply copied the bahavior...
Tschoeeee
Roland
--
* roland@spinnaker.rhein.de * http://www.rhein.de/~roland/ *
PGP: 1024/DD08DD6D 2D E7 CC DE D5 8D 78 BE 3C A0 A4 F1 4B 09 CE AF
Reply to: