On Sat, Apr 27, 2002 at 02:15:11PM -0500, Frederick Bruckman wrote:
> On Sat, 27 Apr 2002, Martin Weber wrote:
> 
> > On Fri, Apr 26, 2002 at 08:18:19PM +0300, Jarkko Hietaniemi wrote:
> > > - download
> > > - unpack (unpacks to perl/)
> > > - cd perl && sh Configure -des -Dusedevel
> > > - make all test
> >
> > Success with minor twiddling on 1.5 & -current on i386, I mailed jhi
> > all stuff which went wrong / the patches/tweaks I need (basically a
> > lacking -Wl,-R /usr/pkg/lib when building, and a not properly propagated
> > -lgdbm in ext/GDBM_File, had to tweak Makefile by hand).
> 
> Should "gdbm" be required to build and run perl? Maybe it would be

It's not *required* to build and run perl.  What now happens is that
the configuration process of Perl *detects* gdbm if it is installed,
but then the build process of Perl fails to properly build the
(shared, dynamic) glue library between Perl and gdbm.

> better, on NetBSD, to not require "gdbm" by default, and so not
> hard-code any paths in "/usr/pkg"? The way it is now, "perl"
> out-of-the-box (with "./Configure -de") breaks on most NetBSD boxes,
> and the typical user isn't going to know how to fix it.
> 
> Frederick

-- 
$jhi++; # http://www.iki.fi/jhi/
        # There is this special biologist word we use for 'stable'.
        # It is 'dead'. -- Jack Cohen