[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: 2.9.4 on mac OSX tiger

From: Marius Schamschula
Subject: Re: 2.9.4 on mac OSX tiger
Date: Sun, 22 Jan 2006 13:00:00 -0600


On Jan 22, 2006, at 9:37 AM, Carlo de Falco wrote:

On 21/gen/06, at 18:53, Joe Koski wrote:


Great work! Sorry, but I don't have an answer to your .dylib question. You
might try posting it to the Apple unix-porting list

to see if they have an answer. Most of your libreadline.dylib versions are
in the Fink directory, or are installed into /usr/local/lib, so it appears
that the one with the missing symbol is Apple's /usr/lib version. When you
move the /usr/lib version out of the way, then the linker probably finds the
/usr/local/lib version that works.
As Dmitri pointed out in his post this is a BSD
flavour of redline.
I see in the /usr/include/readline/readline.h header:

/*$NetBSD: readline.h,v 1.11 2004/01/17 17:57:40 christos Exp $   */
* Copyright (c) 1997 The NetBSD Foundation, Inc.
* All rights reserved.
* This code is derived from software contributed to The NetBSD Foundation
* by Jaromir Dolecek.

No news here. The problem still is how to keep it from trying to link, i.e. how to prevent the linker from seeing it.

For what it's worth, on my OS X 10.3.9, I don't have a
/usr/lib/libreadline.dylib; I only have /usr/local/lib/libreadline.a
installed, and I don't recall when or how it got installed, since that
directory is usually empty on a new Mac installation.
Actually I think I got  /usr/local/lib/libreadline.a from installing
the octave 2.1.7 or opendx bynaries from this might also be your case...

You got be be careful when mixing and matching hand-builds with various ports/package systems. You want to stick to one approach if at all possible. I build two versions, into /usr/local and /usr/local/CISM_DX/required_packages (see below), but this happens on different machines...

Which Fortran compiler did you use, perhaps gfortran? Installed via .dmg
from the website?
Actually I configured with --with-f2c so I think fortran to c translation
is used.


You want to use a "real" Fortran compiler. g77 from works for octave 2.1.72 under Mac OS X 10.3.9, however I built the Mac OS X 10.4.x version using my own build of gcc/g++/g77 3.4.5. I have yet to try to build octave 2.9.x since it is the unstable branch.

BTW I'm going to port my FEM semiconductor device simulator
written using from "the other brand" to octave, but I will
need some substitute for the pdetool plotting functions.
It seems very difficult to use gnuplot for that purpose
so I considered two options:

1) Octaviz, but I've not been able to build that on tiger
2) Opendx, but I would need some kind of octave/dx communication
such at that present for gnuplot...

Octave - OpenDX communications are possible in principle, but are somewhat problematic. I'm part of a project (CISM_DX <>) which does this using a dx module, however our current code only works with octave 2.1.50! I'm shipping octave 2.1.72 with the Mac port of CISM_DX, so this support is not enabled. We are using dx 4.3.2 (4.4.0 just came out and currently fails to compile under Mac OS X 10.4.x).

Any suggestions or any other options?




Marius Schamschula,  Alabama A & M University, Department of Physics

    The Center for Hydrology Soil Climatology and Remote Sensing -


reply via email to

[Prev in Thread] Current Thread [Next in Thread]