strange problems when building Maxima with Ecl

On Tue, Aug 26, 2008 at 03:25:37PM -0600, Robert Dodier wrote:
> On Tue, Aug 26, 2008 at 2:42 PM, Oliver Kullmann
> <O.Kullmann at swansea.ac.uk> wrote:
> 
> > So I need a local installation of ecl, and the only possibility which I see
> > to work around the broken *local* ecl-installation seems to wrap it like:
> >
> > #!/bin/bash
> > export LD_LIBRARY_PATH=/Path/Ecl/lib; /Path/Ecl/bin/ecl "$@"
> >
> > (so "ecl" is this script).
> 
> No, remove the shell script & just execute
> 
> export LD_LIBRARY_PATH=/Path/Ecl/lib
> 
> at your shell prompt, and then ecl can run successfully.
>

You see, this is not an option:

This is not just for my private use, but it's an open source library,
and I absolutely should not mess with the LD_LIBRARY_PATH of my users
(many of them are programmers, and wouldn't use my system if I mess
with their system).

(By the way, it's available at http://www.ok-sat-library.org; yet it's
very early, "pre-alpha", but you might have a look.)

There is a lot of "LD_LIBRARY_PATH is hell" etc. on the Internet,
and everybody (e.g., http://blogs.sun.com/ali/entry/avoiding_ld_library_path_the)
says that the only reasonable way to avoid trouble is to wrap it like
I did.

So the real question seems to me: what things is the Maxima installation
process doing so that the standard wrapper script is not working?

> There is some better way to solve this problem via a program named
> libtool, but I don't know how to use it, and LD_LIBRARY_PATH should
> solve the problem (although it is clumsy).
> 

But libtool seems also only to be usable when compiling: That's the whole
problem, if ecl would compile their binary correctly, there wouldn't
be any problem.

Oliver