What is LFS anyway?

Alexander E. Patrakov see at the.sig
Sun May 9 01:24:36 PDT 2004


Jeroen Coumans wrote:
> Matthew Burgess said the following on 08-05-2004 12:10:
> 
>>
>> Firstly, LFS is *not* minimal.  It never has been since I've been using
>> it.  My idea of minimal would be a book that allows one to read the
>> "From-Bash-To-Powerup HOWTO" and see the end result of it (i.e. a shell
>> prompt) via LFS instructions.  Of course, everyone's idea of "minimal"
>> is subject to their own personal ideas, naturally.
> 
> 
> Thanks for raising this rather difficult question. It seems wise indeed 
> to stipulate some rules which we try to adhere to when compiling the set 
> of packages we'll build. The current set can use some re-evaluation 
> (currently, the only justification for the set of packages is that 
> Gerard once decided on it, and that we should be extremely hesistant to 
> add more packages), just as the assumptions which dictate the current set.
> 
> I'd like to offer some guiding principles for evaluating/changing the 
> package set which currently comprise "an LFS system":
> - we build a system which can recreate itself without external tools
> - we configure each package we install
> - we only add packages if they are a (new) build requirement for a base 
> package
> - the base system is able to perform the following functions:
>   * recreate itself
>   * sustain itself (thus: be able to boot, shutdown, login, repair, log 
> etc.)
>   * provide basic file manipulation (the standard UNIX tools)
>   * provide basic network functionality, excluding servers

One more point here is that our user should not need to reboot into his 
base distro once he built LFS. He should be able to read further 
instructions online and download packages.

This does mean that we should have a dhcp client, pppd and lynx in LFS.

-- 
Alexander E. Patrakov
To get my address: echo '0!42!+/6 at 5-3.535.25' | tr [!-:] [a-z] | tr n .



More information about the lfs-dev mailing list