This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: "./contrib/gcc_update --touch" doesn't work with make
On Tue, Jul 16, 2002 at 05:18:54PM -0400, Hans-Peter Nilsson wrote:
> On Tue, 16 Jul 2002, H. J. Lu wrote:
> > On Tue, Jul 16, 2002 at 04:53:25PM -0400, Hans-Peter Nilsson wrote:
> > > (Reply-draft cleanup.)
> > >
> > > On Tue, 2 Jul 2002, H. J. Lu wrote:
> > > > On Tue, Jul 02, 2002 at 09:58:05PM +0200, Gerald Pfeifer wrote:
> > > > > On Sat, 29 Jun 2002, H. J. Lu wrote:
> > >
> > > (About the need to silence make in gcc_update when invoked from
> > > other make.)
> > >
> > > > > > while ${MAKE-make} -s -f Makefile.$$ all | grep . > /dev/null; do
> > > > >
> > > > > If you just want to add -s to make, this is portable, as far as I can
> > > > > see, so please go ahead and commit your change with a proper ChangeLog.
> > > >
> > > > It turns out that -s still doesn't work in all cases. I didn't bother
> > > > to track it down.
> > >
> > > Example case where it doesn't work? It works for me in "all
> > > cases". I've had the same patch in my tree for too long.
> >
> > I have a build tree for my own Linux based on RedHat. gcc 3.1 won't
> > build inside my tree because gcc31.spec does
> >
> > %prep
> > %setup -q -n gcc-%{version}-%{DATE}
> > ...
> > ./contrib/gcc_update --touch
> >
> > and ./contrib/gcc_update goes into an infinite loop when I do
> >
> > # make
>
> I'm confused: Isn't this one case where you *need* the -s option
> for make in gcc_update? I asked why you say -s isn't
> *sufficient*.
I have several targets in ../common.mk. "prepare", which does "rpm -bp",
works with -s. But "all", which does "rpm -ba", doesn't.
H.J.