LFS-SVN: 6.9. Glibc-2.7

[Dan Nicholson]

On Tue, Mar 4, 2008 at 1:59 PM, Jonathan Oksman
<jonathan.oksman at gmail.com> wrote:
> On 3/4/08, Dan Nicholson <dbn.lists at gmail.com> wrote:
>  >  This has been fixed upstream to use a better search for the bash
>  >  interpreter that won't use the BASH variable, which is set by the
>  >  shell. I don't recall when this went in, but the sed might not be
>  >  needed anymore with glibc-2.7.
>
>  That's sounds to me like what has happened.  The ldd produced by the
>  build process in my tests pointed to bash specifically and not sh.
>  Which I also noted was the same as the toolchain's ldd.
>
>  If what you say is true (and I imagine you are) then chapter 5 is
>  fine.  The sed in chapter 6 is harmless, but no longer needed.

I cut off part of your original message that I wished I hadn't. If, at
the time of building glibc, /bin/sh points to dash, then configure
will search for a bash program and set the variable BASH to that
location. So, in the dash building glibc case, you'll always end up
with /bin/bash in ldd.

In the case where /bin/sh is /bin/bash, configure thinks you're
overriding the location of bash since bash itself sets BASH=/bin/sh.
This has been fixed upstream to use a different variable (BASH_PROG
maybe?) where you'd really have to set the variable to override the
configure search.

--
Dan