LFS-6.0 print process

Nathan Coulson conathan at conet.dyndns.org
Wed Jul 7 15:38:47 PDT 2004

> Well, I've created the testing branch at
> svn://svn.linuxfromscratch.org/LFS/branches/testing.  So far I've
> removed hotplug.  So, still to do are:
> 1) Confirm with BLFS that ed can migrate across
> 2) Confirm with BLFS that net-tools can migrate across
> 3) On a "yes" to 2) apply Jim's iproute2 patches to both books
> If none of the above 3 get done in time, I don't see them as
> show-stoppers, they're relatively minor in comparison to the other
> changes that 6.0 brings with it.  However, they are "nice to haves"!
> I'm trying to get render-lfs-book.sh to produce an online copy for
> everyone to access but for the time being folks will just have to
> download the copy at
> http://www.linuxfromscratch.org/lfs/downloads/testing/LFS-BOOK-SVN-20040705-HTML.tar.bz2
> or do an svn checkout and render it themselves of course.
> As mentioned before, I don't foresee too many technical problems with
> the testing branch.  However, there is a *lot* of documentation to put
> in still.  I'll attempt to get this into the branch by the end of this
> weekend.
> Cheers,
> Matt.

The next iteration of the LFS Bootscripts [which shall be 2.1.3] currently
have the following changes

a)  The network bootscript now uses source within subshells, to obtain the
value of the ONBOOT variable
b) added -depth to the find command in the cleanfs script
c) moved the hotplug script from lfs/init.d to contrib/init.d.
d) ifup and ifdown now have the option of being utilized by hotplug [if
ONHOTPLUG is defined in ifconfig.eth0].  This has no value to the testing
branch at this time.  (I decided to keep it, as nobody said otherwise, and
prevents hotplug from doing whatever it pleases with network interfaces
unless you want it to)
e) Added an additional check to udev.  If it is unable to mount /dev as a
ramfs, print a pretty message, and halt the computer.  (Currently
untested, but committed).

Further changes I wish to commit
Configuring the network interfaces via the interface name, instead of an
arbritary name.  [aka, all protocols on eth0 should be in ifconfig.eth0]. 
There are 2 systems for consideration,

a) SERVICE="static,ipx,pppoe" #[with or without ,'s].
    [all variables that'll be utilized by the static, ipx, or pppoe
scripts included]
  Patches are on lfs-dev, and would need some testing to verify.
  Organized simular to grub's menu'lst, and the different services can be
better organized.  A bit of extra scripting to implement, there was a
basic example on how this could be done on lfs-dev, but it was
psedocode.  (Anything above the first SERVICE= line would be set for all

I would prefer b, if we even want this for the 2.1.3 release.  (Depends on
the testing timeline).

Someone on blfs-dev wanted to migrate the createfiles configuration file
from a file to a directory, I have not seen any further discussion on
that.  (hint, now's the time to speak up if you like)

More information about the lfs-dev mailing list