On Sat, Jul 7, 2012 at 6:43 PM, Nico Kadel-Garcia <[log in to unmask]> wrote:

> There were multiple bug reports, all closed, some marked "not a bug".
> Our favorite upstream vendor did not include the "glibc-static.i686"
> package in their main "channels", probably as part of the separation
> and clean-out of non-x86_64 packages from i686 packages. This was also
> not a problem on their "enterprise release" version 5 which
> auto-installed i686 packages along with x86_64 packages by default.
> Whoever closed the bug reports considered "manually install the
> package to compile grub" to be enough of an answer.

I filed a bug report for this issue with respect to RHEL-6 here:

https://bugzilla.redhat.com/show_bug.cgi?id=641739

I did my best to explain what the real issue was but it was also
closed as NOTABUG. One outcome was that glibc-static.i686 was
eventually made available for x86_64.

> I'm a bit confused about why it didn't show up in the Scientific Linux
> build setups, and would like to get a better handle on how those are
> done.

I believe that SL developers try not to modify the source if the
'problem' can be dealt with by manipulating the build
environment/setup. I'm sure grub is not the only one that needs such
endeavour.

> Unfortunately, my patch had a bug. Using "BuildRequires:
> glibc-static.i686" does not, in fact, find the .i686 version of
> glibc-static, it expects a package named "glibc-static.i686" of an
> arbitrary architecture.

You can add "BuildRequires: /usr/lib/libc.a" as noted in comment #6 of
the above bugzilla.

Akemi