Ben Collins-Sussman <sussman@collab.net> writes:
> Philip Martin <philip@codematters.co.uk> writes:
>
> > Karl Fogel <kfogel@newton.ch.collab.net> writes:
> >
> > > * support Berkeley DB 4.0 *or* 4.1
> >
> > Unfortunately, I think rev 3550 broke BDB 4.1, I get
> >
> > $ ./config.nice
> > [snip]
> > checking for Berkeley DB in /usr/local/db4 (as db4)... no
> > configure: error: Could not find Berkeley DB 4.0.14. with libname "db4"
> > $ ls /usr/local/db4/lib
> > libdb-4.1.a libdb-4.1.la libdb-4.1.so libdb-4.so libdb.a libdb.so
>
> I'm not sure what the problem is. I added this snippet to
> configure.in:
>
> # Look for libdb4.so first:
> SVN_LIB_BERKELEY_DB($SVN_FS_WANT_DB_MAJOR, $SVN_FS_WANT_DB_MINOR,
> $SVN_FS_WANT_DB_PATCH, "db4")
> if test "$svn_lib_berkeley_db" = "no"; then
> # ...try libdb.so otherwise.
> SVN_LIB_BERKELEY_DB($SVN_FS_WANT_DB_MAJOR, $SVN_FS_WANT_DB_MINOR,
> $SVN_FS_WANT_DB_PATCH, "db")
> fi
>
>
> I tested this patch on Karl's machine before committing; it simply
> tried to find -ldb4 in 4 locations, and then attempted to find -ldb in
> the same 4 locations, one of which matched.
>
> Why is your ./configure script simply bombing out after the first
> failure?
The top of ac-helpers/berkeley-db.m4 sets status to "required", then
when the test fails the following code in ac-helpers/berkeley-db is
run
case "$found" in
"not" )
if test "$status" = "required"; then
AC_MSG_ERROR([Could not find Berkeley DB $1.$2.$3. with libname $4])
fi
svn_lib_berkeley_db=no
;;
I think AC_MSG_ERROR must be fatal, I don't know much about autoconf,
but if I put an echo immediately after the AC_MSG_ERROR its output
doesn't appear.
So I think it affects BDB 4.0 as well as 4.1, this will affect lots of
people. Perhaps we should pull the 0.14.4 tarball?
--
Philip Martin
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Wed Oct 30 00:13:39 2002