--prefix=/usr (was Re: GCC vs GLIBC: why this stance, Drepper ?!?)

[Phil Edwards <pedwards at disaster dot jaj dot com>]

On Mon, Jul 02, 2001 at 09:57:16AM -0700, Daniel Jacobowitz wrote:
> On Mon, Jul 02, 2001 at 08:16:38AM +0100, Joseph S. Myers wrote:
> > On Sun, 1 Jul 2001, Daniel Jacobowitz wrote:
> > 
> > > >    ./configure --prefix=/usr
> > > >    make bootstrap
> > > >    make install
> > > > 
> > > > would replace my current GCC with a new one without too much hassle.
> > > > But I think it already does, despite what HJ says.
> > > 
> > > Does it behave correctly in overwriting libgcc?  My reading of
> > > install-sh suggests that it should be fine.
> > 
> > My reading of install-sh says it doesn't work correctly here (it removes
> > the old file, then tries to move in the new file - which would fail if mv
> > were dynamically linked against libgcc).  I made some attempts to
> > reproduce this in a chroot some time ago but they failed to show the
> > problem because (a) the shared libc didn't get linked against the shared
> > libgcc and (b) through fragile means, the C program "install" was used
> > instead of install-sh
> > <URL:http://gcc.gnu.org/ml/gcc/2001-03/msg01327.html>.
> 
> I take it back; you're quite right.
> 
> As a purely technical issue, should we correct this?

I have a sneaking suspicion that we can't get it right in all cases.

For starters, I /would/ like to see a detailed list of the problems
involved when --prefix=/usr is given, sort of a "list of known things
that go nuts".  Then we'd have something to point to, both when discussing
possible solutions, and when users submit bug reports.


Phil

-- 
Would I had phrases that are not known, utterances that are strange, in
new language that has not been used, free from repetition, not an utterance
which has grown stale, which men of old have spoken.
                                     - anonymous Egyptian scribe, c.1700 BC