The latest LFS book (nightly version)

Björn Lindberg d95-bli at
Tue Jul 15 13:37:05 PDT 2003

James Robertson <jameswrobertson at> writes:

> Björn Lindberg wrote:
> > I was browsing the latest LFS version, and I realised that a lot has
> > changed since I last had a look at it. I suppose the new division into
> > stage 1, pass 1 & pass 2 is a consequence of the "pure LFS" stuff. The
> > thing I'm wondering about is the program called Dejagnu. I suppose
> > that the reason for installing TCL and Expect is that Dejagnu requires
> > them, but I don't see why Dejagnus is installed. I assume it is used
> > somewhere in the build process, but my casual browsing of the book
> > didn't reveal where exactly it is used, and for what purpose.
> > Although I don't want to criticize anything about the great LFS
> > project, I must say that it seems to me that building an LFS system is
> > becoming more and more complicated for every book release -- and it is
> > not necessarily always a good thing. Just my personal opinion of
> > course.
> > Björn
> If you want more info, I would read the pure_lfs.txt hint.  The book
> is making a MAJOR leap forward in its implementation of the
> toolchain/toolbox for ch5 before moving on to ch6.  Greg and Ryan did
> a bunch of excellent work in this regard.  The book's text has not
> begun to be flushed out as of yet, so for now, refer to the hint for
> explanations to the commands and order of them.

Ok, I will look up the hint.

> As Linux gets more complicated, so will the book.  This is a fact of
> life.  The way the toolchain is being implemented now will provide
> LFS'ers a better solution going forward and will provide the project
> some pretty good press in the long run.  All good if you ask me.

But these changes are not reflected by any major changes in
'Linux'. It is just a more elaborate way of building a GNU/Linux
system from scratch. Personally, I liked the move to use a separate
static/ directory, as was first intriduced in a hint (by Matthias
Benkamann I believe). I'm not sure about this new pure LFS bit yet,
but I haven't understood it enough to form an opinion. It does seem
more complicated though, but perhaps for good reasons.

One thing that strikes me though is that if I'm already using a
not-too-old LFS system as host system when building, I probably don't
need to take the extra precautions of the extra pass, and the old
static/ + dynamic might be sufficient. Comments?

Unsubscribe: send email to listar at
and put 'unsubscribe lfs-chat' in the subject header of the message

More information about the lfs-chat mailing list