Re: GCC vs GLIBC: why this stance, Drepper ?!?

[Paolo Carlini <pcarlini at unitus dot it>]

Hi,

Justin Guyett wrote:

> On Sat, 30 Jun 2001, H . J . Lu wrote:
>
> > I am happy with gcc 2.96 from RedHat. I will recommend it over gcc
> > 2.95.x for glibc 2.2. But it is only my opinion.
>
> That's fine, but someone building a LFS system may not realize that the
> optimal solution is to get the gcc 2.96 srpm from redhat and install that
> as the system gcc before beginning.  It would be nice to have a release of
> gcc that builds all critical linux packages (excepting broken code that
> gcc 2.95 allowed).  Not to mention gcc developers publically denounced the
> rh gcc 2.96 snapshot.  Does the 2.95.4 snapshot used by debian have those
> two patches in it?

For sure the "atexit" patch was applied to the 2.95.4 branch:

    http://gcc.gnu.org/ml/gcc-patches/2001-04/msg00130.html

The other one, was hacked by Jakub Jelinek for me ;-) , never formally tested,
and I'm almost sure that is not present in the official GCC 2.95.4 branch.

> Would gcc 3.0 with static libgcc_s work for recompiling glibc?

This is a nice question!!

I would add: what features *exactly* are broken when building glibc2.2.x with
gcc3.0? (I'm asking this because I tried on my Linux-x86 (using 3.0 pre) and
had no major problems for 3-4 weeks, when I learned from Andreas Jaeger that
my PC was about to explode  :-)

What *exactly* is *still* not ok when a static libgcc_s is installed?

Cheers,
Paolo Carlini.