LFS 5.0 is a serious problem

Paul Campbell paul at multimedia-campbell.com
Tue Nov 18 08:49:42 PST 2003


^(o,o)^ Turk2000 wrote:

> Whi? This procedure took about 40% less time than LFS 5.0 traditional.

Exactly, why?  The answers are in the book and the system it builds.

If just you want a linux system, built from source code, then just unpack
the kernel, glibc, core utils and bash + whatever apps you want and compile
em.  Thats not the point of LFS is I have found anything out.  It has to be
"from scratch" without any distro blood in the works.

In the above method a lot of already compiled code in package development
RPMs from your distro will be used in creation of the new binaries.  They
will be made by one version, to make another.  Make version X makes Make
version Y and so on. While these new binaries will probably run outside
their mother enviroment, they are effectively mungrels, worse they are
bastard binaries (LOL).  

LFS repeats packages in the various stages to filter out (not sure on
this) ?all? the code assilimated from the initial host build system,
effectively breeding a pedigree LFS with no distro blood left in it.

Just my 2pennies (cents) worth.  

My LFS 5.0 went to plan, though. (achmmm) I did remove the make test/check
on a few passes....  the system was a slow one.  I also accidentally
(ooops) built the kernel I am running with Gcc-3.x not the prepared
gcc-2.x.  I dont know nor see any major problems yet....

Paul



More information about the lfs-support mailing list