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

H . J . Lu hjl@lucon.org
Sun Jul 1 20:24:00 GMT 2001


On Sun, Jul 01, 2001 at 08:06:20PM -0700, Richard Henderson wrote:
> On Sat, Jun 30, 2001 at 10:26:20PM -0700, H . J . Lu wrote:
> > Do you seriously believe
> > 
> > # ./configure
> > # make
> > # su
> > # make install
> > 
> > should override a system DSO which may be used by every single binary
> > on the system? That is the central issue for any system which gcc
> > is used as a system compiler.
> > 
> > If your answer is
> > 
> > 1. Don't know. I have nothing to say.
> > 2. Yes. It is too bad. I have nothing to say either.
> > 3. No, it shouldn't override a system DSO. Then we can start to talk.
> 
> My answer is 3, in that I disbelieve that "make install" could
> ever be made robust enough for that to work.

What do you mean by that? Did you mean sometimes "make install"
wouldn't work right?

> 
> However, I'll also state the Red Hat srpm (and Debian whatsit, etc)
> for gcc should build a libgcc rpm that _does_ install libgcc_s.so
> in /lib.

I have no doubt about that. My questions are

1. Will every Linux distribution come up with the similar scheme to
deal with libgcc_s.so so that they are as compatible as before gcc
3.0?
2. Will

# ./configure
# make bootstrap
# make install

made by a normal gcc user ever make his/her machine unreliable where
there is a system libgcc_s.so under /lib?


H.J.



More information about the Libc-alpha mailing list