seg-fault of octave on MacOSX

[Jean-François Mertens]

On 05 Dec 2007, at 03:32, John W. Eaton wrote:

> | On 04 Dec 2007, at 23:34, Jean-François Mertens wrote:
> |
> | | This occurs even when octave is compiled with
> | FFLAGS='-O0 -gfull -Wall'

> I don't see how FFLAGS could be important here.  The ov-fcn-handle
> file is written in C++ and I don't see that the tests in that file
> depend in any way on parts of Octave that are written in Fortran.
>
Was my impression too.
But definitely the first complaints on fink lists about such problems  
started
almost immediately after the switchover from g95 to gfortran 4.2.2
(including for all fortran deps)

And I observe that the second part of the tests in src/ov-fcn-handle.cc
seems hdf5 -related _ so I first re-built hdf5, running make check, and
that passes all its tests..

There might be one or two other very indirect calls... (although if
src/ov-fcn-handle.d is any guide, it contains only a single item in
libcruft _but .f files have no corresponding.h files ..)

> | OhOh !  Now I see some sort of test-sequences (2) within the code of
> | src/ov-fcn-handle.cc itself !
> | Any instruction on how to run those independently, and preferably
> | step by step,
> | in the builddir ?
>
> The tests are simply bits of Octave code.  You can extract them
> (omitting the %! comment marker, of course) and run them at the
> command line.
>
> To run Octave in the build directory, use the run-octave script.

Thanks a lot !
Will try tomorrow evening _ getting really late now ...



JF