Re: Configuring inittab
On Fri, Feb 24, 2006 at 02:50:34PM +0100, Michelle Konzack wrote:
> Hello Tarvin,
>
> Am 2006-02-19 18:02:20, schrieb Digby Tarvin:
> > Debian by default does not make good use (IMHO) of the runlevel mechanism.
>
> Oh yes, it does.
Actually I was expressing my opinion (that is what IMHO means), so
you can't really correct me and say that my opinion is something else...
Of course you are free to say what your opinion is...
> Debian give you the freedom to configure your rcX.d HOW YOU WANT!
Of course - that is what I indicated in the posting.
The earlier statement referred to the default configuration (which
is a characteristic of Depian), not the flexibility of the mechanism
which - which is common go most Linux distros and Unix derivatives
in general.
I believed the default configuration (all multi user levels th e same)
doesn't make good use of a facility that can define different service
profiles.
Nothing would be lost in defaulting to the traditional runlevel
choices, if the default runlevel was the one which is equivelent
to all of 2-5 currently. And it would provide extra functionality
for those that don't want to define their own initially.
> > It bundles all multi-user stuff into runlevel 2 and then leaves 3-5
> undefined
>
> It is defined like the rc2.d.
Which definition are you referring to?
Martin Krafft's "The Debian System" lists runlevel's 3-5 as 'unused'.
Other documents list them as the same as 2. I know that Sarge installs
with them all the same as 2 - but I havn't seen a 'definition' that
would lead me to assume that this can safely be relied on, even if it
was in any way useful..
> > The traditional usage I was familiar with was
> > 2 - multi-user, no network
>
> No Network is stupid by default.
That is rather a sweeping statement to make without any supporting
justification. (and nobody said it had to be the default runlevel)
I wasn't actually saying that it was what I wanted, merely the 'standard'
used by other more traditional flavours of Unix. In any case, having it as
a choice seemed better than having no choice at all as the default
configuration.
For what it is worth, I think it makes a lot of sense on a laptop which
is sometimes booted in an environment where there is no network. Why sit
around while your battery is used to start up NFS servers and SSH daemons
that will never be used, and waiting for DNS lookups timeout, because you
want to edit a file on a train?
What I personally would prefer is a runlevel that includes X but no
other networking for mobile non-networked use, a runlevel that is
appropriate for use behind a firewall (XDMCP support, telnet etc)
and one that is suitable for direct Internet connections (just SSHD,
firewall and client services).
> > Anyway, the infrastructure is there, and you can fix it if you
> > want. I am sure there was a good reason for the change, but I
> > sure as hell can't think what it would have been...
>
> Debian does not force users. - With Debian you have the
> choice or use another Distribution if you do not agree...
>
> This is 100% freedom.
I am 100% aware of this, but it has nothing to do with
appropriatness or otherwise of the default configuration
provided.
The number of options on any Unix derived systems are vast, and
appropriate defaults should aim to provide as much functionality
as possible to the widest ranger of users as far as can be done
without unreasonably compromising security - so that individual
requirements can be satisfied with as few customizations required
as possible.
As far as I can see, the change to the default runlevel configuration
in Debian removes functionality from the default configuration with
no identifiable gain.
> > (maybe to simplify package management for packages that involve
> > additions to system startup - so they wouldn't need to ask
> > about with runlevel things get added to??)
>
> Provide a patch! - The BTS is open for it.
That last statement was speculation about a possible reason for
the change to default runlevels in Debian - not a request or
suggestion for a change.
Or were you referring to something else?
Regards,
DigbyT
--
Digby R. S. Tarvin digbyt(at)digbyt.com
http://www.digbyt.com
Reply to: