Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 29 Nov 1998 11:44:00 +0100 (CET)
From:      Lars Gerhard Kuehl <kuehl@lgk.de>
To:        "John S. Dyson" <dyson@iquest.net>
Cc:        advocacy@FreeBSD.ORG, tlambert@primenet.com, (Wes Peters) <wes@softweyr.com>
Subject:   Re: Linux to be deployed in Mexican schools; Where was FreeBSD?
Message-ID:  <XFMail.981129114400.kuehl@lgk.de>
In-Reply-To: <199811290732.CAA35872@y.dyson.net>

next in thread | previous in thread | raw e-mail | index | archive | help

On 29-Nov-98 John S. Dyson wrote:
~
~
>> > >       Fix: Change the FreeBSD "init" process.  This is political
>> > >       suicide, but technological necessity.
>> > >
>> > No question about that.  SYSV init or something close to that
>> > is necessary.
>> 
>> It shouldn't be all that difficult; have the arguments in that past 
>> been "it's just not BSD-ish?"
>> 
> I believe that you are right.  Anyone (IMO) who has really used a
> SYSV style init (and understands it), will find that it is a valuable
> tool.  Expecting it to solve *all* problems is probably a little too
> demanding.  However, IMO, it is a good tool that could help manage
> system startup and package startup/shutdown, etc.  As any valuable
> tool, SYSV init can cause problems -- but then if it hurts (SYSV init
> does something evil), then don't make it do the evil thing!!! :-).
> 
> One can hack a solution with BSD init, but it ends up implementing
> subsets of SYSV init.  Why not go all the way and just do it?  If
> SYSV init has serious problems (which it is indeed NOT perfect), then
> implement either a better version, or start moving forward with an
> existant version, and then move forward from there.

The strong side of BSD init is its simplicity. A good redesign of init
adding SYSV style capabilities could get the proc 1 image even simpler.
Why not putting all the getty/ttys stuff into a different process? (Kind of
special getty parent process, not necessarily a ttymon.) Further I'd add a run
level manager process  which wouldn't need to have a dedicated proc id,
as well as a seperate program initctl. If run level information was stored
persistently, the 'init system' could recover from an rlm crash without
getting the entire system into an unhealthy state.

Good old V7 init would not retire - hence no reason for wistful lamentation. ;-)

        Lars

/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\/\
Lars Gerhard Kuehl      Phone: +49 40 54768010  Mobile: +49 171 9307085
                        Fax  : +49 40 54768012  Email : kuehl@lgk.de
#ifdef is your friend, and everyone's else enemy
#endif /* is your friend */

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-advocacy" in the body of the message



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?XFMail.981129114400.kuehl>