Re: Status of MacOS X Porting?

[Mumit Khan]

On Sun, 13 May 2001, Matt Pfenninger wrote:

> Hey,
> 
> Just wondering if anyone has addressed the MacOS X linking problem described
> in http://www.octave.org/mailing-lists/help-octave/2001/384.  I've
> duplicated the problem, but don't really have any ideas for a solution.
> Anyone else had any luck?

The bug report leaves out a very important bit of info -- the actual
link step. Are the needed libraries, liboctave and liboctinterp, indeed
being linked in and in the right order? It's always hard to troubleshoot
incomplete bug reports.

To track down the problem, you'll need to look at the demangled names
and see why these are not being resolved. I took at a quick look at
a few samples, and looks like the demangled names have an extra underscore
that may be causing the problem. I don't know the ABI for OS X, so don't
know if that may be the problem. 

To demangle, use c++filt (grab the message 384 from the archive and
run it through c++filt -- `cat 384.txt | c++filt > 384-demangled.txt').
Let's take two symbols: (1) Array2<double>::_get_size(int, int) const,
and (2) Array2<octave_value>::_get_size(int, int) const. The first one
should be in liboctave/liboctave.a (or whatever the archive/shlib
extension is), and second one in src/liboctinterp.a. Now take a look
at those two archive/shlibs, and see what is different. The linker may
have options to increase verbosity as well, and that is another way to
check why certain symbols are being ignored.

Regards,
Mumit




-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------