[patch 1/10] Don't depend on Kerberos headers when checking librpcsecgss in configure

[Neil Brown]

On Tuesday July 4, kwc at citi.umich.edu wrote:
> On 7/4/06, Greg Banks <gnb at melbourne.sgi.com> wrote:
> > On Tue, 2006-07-04 at 10:46, Neil Brown wrote:
> > > On Monday July 3, kwc at citi.umich.edu wrote:
> > > > -  AC_CHECK_LIB(rpcsecgss, authgss_create_default, [librpcsecgss=1], [AC_MSG_ERROR(librpcsecgss needed for nfsv4 support)], $KRBLIBS)
> > >
> > > In my tree, this line looks like
> > >      AC_CHECK_LIB(rpcsecgss, authgss_create_default, [librpcsecgss=1], AC_MSG_ERROR([librpcsecgss needed for nfsv4 support]), $KRBLIBS)
> > >
> > > Note the different bracketing near AC_MSG_ERROR.  Obviously I could
> > > just make the relevant change anyway, but I'm curious where that
> > > discrepancy comes from, so I thought I would ask first.
> > > So consider this a question (without any question-mark!!).
> >
> > Sorry, that was me:
> >
> > http://linux-nfs.org/cgi-bin/gitweb.cgi?p=nfs-utils;a=commit;h=24d303ffae686192bda0a5996e8590219dcc82e7
> >
> > Greg.
> 
> Phew!  Thanks, I didn't have an answer.  However, I did rebase the
> patches on 1.0.9-pre1, so I'm curious why I didn't notice.

Greg's patches went in after -pre1 (but are dated earlier).

Now if I had just done
  git blame configure.in
I would have found.

24d303ff (Greg Banks 2006-06-22 17:16:19 +1000 193)   AC_CHECK_LIB(rpcsecgss, authgss_create_default, [librpcsecgss=1], AC_MSG_ERROR([librpcsecgss

and I might have understood (though given my recent track record -
don't bet on it).

All pushed out now.

Now to figure out why I sometimes get

Making all in support
make[1]: Entering directory `/home/git/nfs-utils/support'
Making all in export
make[2]: Entering directory `/home/git/nfs-utils/support/export'
test -f mount_clnt.c && rm -rf mount_clnt.c || true
../../tools/rpcgen/rpcgen -l -o mount_clnt.c mount.x
test -f mount_xdr.c && rm -rf mount_xdr.c || true
../../tools/rpcgen/rpcgen -c -o mount_xdr.c mount.x
test -f mount.h && rm -rf mount.h || true
../../tools/rpcgen/rpcgen -h -o mount.h mount.x
ln -s ../export/mount.h ../../support/include/mount.h
ln: creating symbolic link `../../support/include/mount.h' to `../export/mount.h': File exists
make[2]: *** [mount.h] Error 1
make[2]: Leaving directory `/home/git/nfs-utils/support/export'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/home/git/nfs-utils/support'
make: *** [all-recursive] Error 1

and maybe remove

./support/export/keys.c
./support/include/rpcdispatch.h
./support/include/rpcsec.h
./support/include/version.h
./support/include/ypupdate.h
./support/nfs/clients.c
./support/nfs/keytab.c
./support/nfs/ypupdate_xdr.c
./support/rpc
./support/rpc/include
./support/rpc/include/Makefile.am
./tools/rpcdebug/neat_idea.c
./utils/mountd/mount_xdr.c
./utils/rquotad/pathnames.h

none of which seem to be used

and maybe get rquota_???.c to be generated from
rquota.x

and maybe wonder why we have our own rpcgen (but now that we have
fixed it, I'm loathe to remove it) when glibc seems to provide one.

NeilBrown