Mailing-List: contact new-httpd-help@apache.org; run by ezmlm
Precedence: bulk
Reply-To: new-httpd@apache.org
Date: Mon, 29 May 2000 15:48:33 -0700 (PDT)
From: Greg Stein <gstein@lyra.org>
To: new-httpd@apache.org
Subject: Re: cvs commit: apache-2.0/src/lib/apr/misc/unix errorcodes.c
In-Reply-To: <20000529154417.B1911@innovation.ch>
Message-ID: <Pine.LNX.4.10.10005291545200.14857-100000@nebula.lyra.org>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII

On Mon, 29 May 2000, Life is hard, and then you die wrote:
> On Mon, May 29, 2000 at 03:22:21PM -0700, rbb@covalent.net wrote:
> > On Mon, 29 May 2000, David Reid wrote:
> > 
> > > I agree - no hard coding of -ldl.
> > 
> > +1 here.  -ldl should only be added if APR is configured with DSO's and we
> > are on the correct platform.
> 
> FYI, this is currently broken on Linux (Mandrake 7.0). I did a
> 
> ./buildconf
> ./configure --enable-auth-digest --disable-cgi --disable-echo
> make
> 
> and it won't link because no -ldl is being given to the linker.

Sure, it needs to be added. My original point is that it should not be
hard-coded. DL is /not/ the only way to load DSOs.

The list of mechanisms (that I presented earlier) can easily be seen from
Python's dynamic loading varieties: AIX, BeOS, DL, HP/UX, Mac, NeXT, OS/2,
SHLIB, Win32.

Now that David is back, we'll at least have the BeOS variety in
consideration.

btw, Python uses SHLIB on Linux. Dunno why we are choosing DL. Read: who
is wrong? :-)

Cheers,
-g

-- 
Greg Stein, http://www.lyra.org/