Can't get past Bzip2

barry at barry at
Wed Dec 27 08:56:22 PST 2000

Jesse Tie Ten Quee wrote:
> Yo,
> On Wed, Dec 27, 2000 at 08:01:06AM -0800, WarmFuzzy wrote:
> > lrwxrwxrwx   1 root    root    21 Dec 25 16:10 /usr/bin/gcc ->
> > /etc/alternatives/gcc.
> Nono... inside chroot, where your having this problem..this is in chroot
> right? (or did i read something wrong..?)
> > Should I tell you that this is on a Mandrake 7.2 distro? I also did gcc
> > -v and it says this is version 2.95.3 19991030 (prerelease).
> *checks running MDK7.2*  yup standard MDK it
> seems..interesting that there using a prelease of 2.95.3 (the unreleased
> bug fix for 2.95.2) allthough that isn't as bad as Redhat 7.0 *shudders*
> -
> Jesse Tie Ten Quee - highos at highos dot com

at least Red Hat gives the kgcc package, though...

and, isn't bzip2 also in the static compilation?  

I think that the important question here is whether this was the static
bzip2 or the dynamic (chroot'ed) bzip2...

if so, that is if it were the static compilation, they would have only
had to compile bash and binutils initially.  Did either of these
packages show any strange behavior during compile time?  That might clue
us in on what may be happening.  Further, was $LFS set?  If not (and
assuming the command syntax was copied exactly from the book) then it's
possible that binutils could have been installed on his mandrake system,
which can cause all kinds of bad monkey style wierdness if they're using
a gcc snapshot...

beyond that..the symlink is odd.  Why are they putting this in /etc? 
Doesn't it seem more logical to place alternative packages in /usr or
/usr/local?  Or even /opt?

I think that this could stand as a direct violation of the FHS... but,
in any case, the fact that the symlink exists in /usr/bin should make it
possible for make to find gcc..I've done something like this before.

Perhaps a fix might be to skip to compiling gcc on the current system
(which may not work if this problem is that severe) and then start from
bzip2 again...

Unsubscribe: send email to lfs-discuss-request at
and put unsubscribe in the subject header of the message

More information about the lfs-dev mailing list