relocation of the sources

Kev Buckley k.buckley at
Thu Jun 9 06:32:53 PDT 2005

> This brought up a philosophical debate in my mind. If the book mentions
> moving the sources, but then proceeds to move them to a directory where
> only root can write, ISTM that this can be mis-interpreted as "you have
> to download sources as root to be able to save them". If someone has to
> be root to save new sources in the suggested directory then how far is
> that from being root to build?
> Apart from this line of thinking, another thought was why does the book
> suggest this at all? Is this something that should be left as an
> exercise to the reader instead of something that a new reader will
> blindly follow (and they most likely will blindly follow)? That is why I
> added "if so desired", "If you wish", and "wherever you choose".
> Suggestions?
> -- 
> Archaic

Probably a bit late to chip into this now but I have always wondered 
why the LFS doesn't advocate building outside of the untar'ed source
directories where possible.

I appreciate that some packages are unable to do this but many of the
current LFS sources can be compiled in exactly the same way that is
followed for say, glibc and a couple of others where the package
author suggests building outside of the tree.

(assume you are in source dir for package)
patch -Np0 -i ../as_required
mkdir ../build-package_name
cd ../build-package_name
../package-v.r.s.n/configure --prefix blah blah blah

Would this approach not allow one to leave the majority of the sources
"lying around" untouched and ready for use in stage 2/3/... and only 
leaving one having to remove the build directories.

Admittedly this does not gain anything in the newer Cross-LFS contexts
where the second stage is remote from the original build environment
but for LFS constrution on a single system under a chroot, why cant
the sources live in the chroot from the start.

I am also aware that some packages are patched in different ways
depending on the stage of the book you are building at, but again, why
not untar the sources and rename the source directories to relflect
the stage they'll be used in.

So for packages that get patched differently or which cant be built
outside of the source tree, you'ld advocate having two source trees


rather than just assume that there is one single virgin source tree at
the start of every stage.

Hope that makes sense.


*  Kevin M. Buckley              e-mail: K.Buckley at   *
*                                                                    *
*  Systems Administrator                                             *
*  Computer Centre                                                   *
*  Lancaster University          Voice:  +44 (0) 1524 5 93718        *
*  LANCASTER. LA1 4YW            Fax  :  +44 (0) 1524 5 25113        *
*  England.                                                          *
*                                                                    *
*  My PC runs Linux/GNU, you still computing the Bill Gate$' way ?   *

More information about the lfs-dev mailing list