Readline does not belong and why (long) - was My Plans

Larry Lawrence larry at linuxfromscratch.org
Fri May 14 19:46:34 PDT 2004


On Thu, 13 May 2004 23:03:13 -0700, Jeremy Utley wrote:


> OK, I'll explain this ONE MORE TIME, and this will be the end of it.
> 

4) Readline
> provides functionality to Bash and E2fsprogs (Bash, if readline is not
> present, will statically link it's own internal copy of the readline lib,
> but will not provide this library for other programs such as e2fsprogs). 
> So in effect, by NOT including readline we are limiting the functionality
> of these programs.
> 
> With that, I'm going to take some time away from these lists, cause noone
> is listening anyways.
> 
> -J-

I'd like to address this from using some BLFS experience.  Install Bash
and efsprogs with out readline.  Is there any instruction command in the
book that will not work?  If I use the standard of who should be building
LFS, that person should be 1) able to tell that readline is missing or 2)
unsure why the command line is working differently than in the host, but
it is not interfering with progress.  Once the system is running, there
are a couple of things that can happen. After installing readline from
BLFS, a LFS package is upgraded, functionality returns, Person 1 from
above knows exactly why and may have even reinstalled the same version to
get that effect, Person 2 may find the up cursor producing history
commands instead of control characters and they still may not know why.

The same thing happens to libxml2 in BLFS, there is no reason to install
Python prior to installing libxml2, but by the time you upgrade to the
next version, Python is probably going the be installed and the
installation is "complete".  The trick is to make sure the missing feature
is not impeding progress.

That's why ALSA is in BLFS, You don't need sound functionality to complete
LFS, but you'll probably figure out it's missing.

I have yet to understand why the first build has to be "complete" (and
will the community standard for "complete" ever live up the the
expectations  of the user).  First, I think you are overestimating the
expectations of the user, and if your not, then your user was expecting a
complete mini distribution after he finished the book.

Each step needs to be a necessity for future steps, with every step going
to the goal.  Goal moving has been tried before, it doesn't ever work
(mainly because everyone's goal is the way they want the system to look
after completing LFS!).  So we end up back to where Gerard thought the
book should end and a whole lot of side projects were formed.  Hints for
the real aggressive, ALFS for the automaters, BLFS for those of us crazy
enough to use LFS for everyday functions.

I doubt that being useful was ever a goal of LFS in the beginning.  It
came later when it was necessary to "eat your own dog food" and prove that
the process worked well enough to be usable instead of a distribution. 
Many of us have probably taken this a bit far, but it has worked out.  A
real satisfaction for us, but few in the real world really give a damn.

So, yes, I am a minimalist.  LFS trained me that way. 

I'll say one more thing -- Stop making the book a set of instructions
"you" (as in this is not your first time) need to follow to create a
working system. This book has always been about the next guy, lost but
willing, to learn what you learned the first time.

Larry



More information about the lfs-dev mailing list