problems with iproute in lfs 6.0

Schafer Frank Frank.Schafer at t-systems.cz
Fri Jan 7 00:11:42 PST 2005


Agree,

... and _lots_ means !!!!! LOTS !!!!!. If the purpose of teaching people
what makes Linux works (as stated once in at the beginning of LFS) is
still true, there is much too much script fu-matic in the initscripts.

Well, the init scripts (too) are work in progress and every programmer
knows, that changes of software over time lead to a state at which even
the programmer itself can't tell out of the box what every line is for.

The initscripts are modular. This is a very good programming style but
where is the border between good and overhowl. If a script sources a
script to set a variale used in a script to source to set the final
value of this variable which leads to the fact that someone trying to
read the script has to open up to 5 virtual consoles to ??be able?? to
figure things out ...

... hmmm.

This needs at least strong documentation.

Furtheron (I made complete own initscripts and checked if I forgot
something yesterday) there are some tricky things used. Every programmer
likes to realize such things in his programs bu this needs DOUBLE STRONG
documentation.

Let's take the cleanfs script for instance:

Removing everything from except utmp from /var/run if it is older than
/proc is a good idea but I don't know if everybody understands this.
I couldn't figure out why this script creates files, directories, pipes
and even device nodes. Well it doesn't but it is prepared to do so.
The line:
	exec 9<&0 < /etc/sysconfig/createfile
... I couldn't even explain it from the bash man page. Maybe Gerard
could explain this. Again, in the script itself ;o))

Or in rc itself:
	i=$(ls -v $rc_base/rc$runlevel.d/K*)

Does exactly the same, what
	i=`ls $rc_base/rc$runlevel.d/K*`

Does, what is more readable but this is maybe a matter of taste.
$rc_base is one of these variables set shelling 2 files ... And a
constant for the whole system. :(

If I'm at this. Finding out what services are started at this moment by
looking for the S* scripts in $rc_base/rc$prevlevel.d is IMHO a bad
idea. If someone add or remove sevices to the init system and then
changes the runlevel a not running subsystem will be tried to be stopped
or remains running. This is what /var/lock/subsys is for.

Just my 2 pence
Frank

PS: Uses LFS / BLFS still these plain text HINT files I knew some time
ago? If yes - where can I place a WIRELESS-HINT?

-----Original Message-----
From: lfs-support-bounces at linuxfromscratch.org
[mailto:lfs-support-bounces at linuxfromscratch.org] On Behalf Of Kevin P.
Fleming
Sent: Thursday, January 06, 2005 4:26 PM
To: LFS Support List
Subject: Re: problems with iproute in lfs 6.0

Nathan Coulson wrote:

> hmm, I'll see if I can simplify it some...  [not sure when the 
> networking part got congested]

In the spirit of education, rather than dumbing-down the scripts, maybe
it would better to just adds _lots_ of comments to them so users can
learn what they are doing and why? Seems to me that would solve both
problems.
--
http://linuxfromscratch.org/mailman/listinfo/lfs-support
FAQ: http://www.linuxfromscratch.org/lfs/faq.html
Unsubscribe: See the above information page





More information about the lfs-support mailing list