On 15.11.11 22:37, Gerald Pfeifer wrote:
GCC fails to build on FreeBSD 10, and debugging this I ran into the
following snippet that's spread all over the place in various ./configure
files:
freebsd* | dragonfly*)
# DragonFly does not have aout. When/if they implement a new
# versioning mechanism, adjust this.
if test -x /usr/bin/objformat; then
objformat=`/usr/bin/objformat`
else
case $host_os in
freebsd[123]*) objformat=aout ;;
*) objformat=elf ;;
esac
fi
FreeBSD 10.0 matches this pattern, and I guess it's from there that
things go downhill.
Sadly this does not come from any of our configure.ac or m4 bits,
but upstream Autoconf 2.64. What can we do here, for trunk and
really also GCC 4.6?
Are you sure that they come from autoconf? I'd say they come from
toplevel libtool.m4.
When I tested this, I only had to adjust libtool.m4 and then
autoconf'ing the needed files.
libtool.m4 change:
------------------
case $host_os in
- freebsd[[123]]*) objformat=aout ;;
+ freebsd[[23]].*) objformat=aout ;;
resulting configure change:
---------------------------
case $host_os in
- freebsd[123]*) objformat=aout ;;
+ freebsd[23].*) objformat=aout ;;
On trunk, wouldn't it be time to switch to a current version of
Autoconf?
Gruss,
Andreas
