Re: CVS update: MODIFIED: ac-helpers ...

On Fri, Jun 08, 2001 at 01:09:17AM -0000, brane@tigris.org wrote:
> User: brane
> Date: 01/06/07 18:09:17
>
> Modified: ac-helpers berkeley-db.m4
> Log:
> Libsvn_ra_dav was missing the an explicit dependency on $(SVN_DB_LIBS),

No it wasn't missing :-). ra_dav does not depend on the DB. It depends on
Neon, which talks over to the wire, to code which *does* depend on the DB.

Looking at your checkin, it appears you actually meant ra_local. And yes:
that does have a dependency.

> which didn't help building the client with --disable-shared (although
> I don't see why libtool doesn't figure out a transitive closure of
> dependent libraries when doing a static link).

Because we don't feed it .la files, so it can't do the closure (even if it
does). I don't recall why we're giving it .a files instead of .la files.
Need to test that.

> And, interestingly
> enough, $(SVN_DB_LIBS) wasn't being set anywhere. Weird.

Long-standing issue. IMO, we should have always had this. When Jim set up
berkeley-db.m4, he modified LIBS and CPPFLAGS directly. That was a poor
choice because it alters how future autoconf tests operate.

Mo's patch was working in the same area, and I went "woah! that should be
fixed."

>...
> --- configure.in 2001/06/05 09:58:30 1.90
> +++ configure.in 2001/06/08 01:09:16 1.91
> @@ -125,6 +125,8 @@
>
> if test "$svn_lib_berkeley_db" = "yes"; then
> FS_RULES="fs-libs fs-programs"
> + AC_SUBST(SVN_DB_INCLUDES)
> + AC_SUBST(SVN_DB_LIBS)
> fi

This should *always* do the AC_SUBST(). Otherwise, you'll end up with
@SVN_FS_INCLUDES@ and @SVN_DB_LIB@ in the Makefile. If you do the SUBST,
then those forms will be replaced with "" in the Makefile.

Cheers,
-g

-- 
Greg Stein, http://www.lyra.org/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org