On 08/01/2010 05:29 PM, Timothy Rice wrote:
> No, I'm not currently building a base system. You're correct that we
> should focus on updating the hint first. I can probably still contribute
> to Part I of the hint, and to the scripts.
>
> There are three more ideas I want to put out there before we decide what
> changes to make to the hint. So, I will now take my turn to do a bit of
> rambling - that way you won't feel all alone:-D
>
Hi yourself, Tim--
I snipped your three ideas, but I want to tell you that they intrigued
me. With that, I "rambled" yesterday because I wanted anyone who's
interested to know my mind set when I "throw something out there." Your
three ideas allowed me to articulate another thought that I hadn't quite
put together yesterday when I rambled.
For the sake of this discussion let's assume that basic LFS and a "no
frills" package users system comprise a "first build." If we update the
hint properly, then this build should go without a hitch. [BTW. I think
there may be some "stuff" to ferret out in Ch. 6 up to and including
GCC-4.5.0. For the first time in my LFS experience, using package users,
I got some depressing test results. I don't know if it has anything to
do with my typing or whether it's package users. I do know that the test
results were "lovely" when I used jhalfs.]
So now, in our scenario, a builder is ready to move on to BLFS, which
takes the discussion out of the realm of lfs-dev. IMO, the frustration,
which lead to the learning, I've had with package users came during
"BLFS time."
OK, that's scenario #1. LFS and a package users hint that will get
someone ready for BLFS.
Now an observation on our discussion and my package user efforts. My
scripts, your scripts and both sets of our ideas lead to reducing the
frustration and increasing the ease in which package users operates
during BLFS. Another way to say it is, our comments, and the ones
William made about your ideas, come from experienced builders--those who
have moved beyond developing the basic skills the combined {,B}LFS
experience provides.
So I'm thinking that it may be reasonable to bring the hint "up to
date," and put the information for our ideas on the wiki and make
example scripts, or whatever, available for use. These could be in the
form of "to automate install directories add these lines to your build
script."
Please don't misunderstand. I'm not discounting or denigrating your
ideas. I'm just wondering if it's a good idea to incorporate the
"improvements" after LFS. In addition to firmly believing in the
learning that LFS provides by manual builds, I am a great believer in
Occam's Razor. But maybe I tend to *over*-simplify. For example, you
mentioned a system that can be used for "versioning control." Isn't that
inherent in package users as it exists now?
ls /usr/src | grep man-pages
gives you every version of man-pages you have ever installed if you
don't delete the package user's home directory.
Right now I'm trying to develop a philosophy of what needs to go in the
"bare bones" hint. I'd like to start separate threads for each of your
ideas, or, process if that process involves more than one of your ideas.
Thoughts? Gut reaction?
Great discussion.
Dan
--
http://linuxfromscratch.org/mailman/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page
