seg-fault of octave on MacOSX

[Jean-François Mertens]

On 05 Dec 2007, at 06:34, John W. Eaton wrote:

> On  5-Dec-2007, Jean-François Mertens wrote:
> > switch from g95 to gfortran 4.2.2
> Was that the only change, or did the C and C++ compilers also change?
Yes, it was ! C and C++ were left at Apple's !
But I sent a copy of this problem to a friend, jhowarth at cinci.rr.com,
who is maintaining the gcc and g95 etc pkgs for fink, and he reported
just now that if using gcc4.2.2 for all 3 compilers (c, c++, and  
gfortran),
this error disappeared.
So this seems really a compiler problem (and specific to the Mac- 
Intel back-end?),
that we have to solve for ourselves _ if we can just hope for your  
guidance
in reducing this to an example not involving octave ..
>
> | 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..
>
> How is HDF5 involved?  I seem to remember that there was a report that
> the crash occurred when saving structures in text mode.  In that case,
> I don't think the HDF5 library could have any effect.

Just saw that the second part of the test (lines 1218-1240) seemed to be
in an enormous #if  ... #endif  :
   687  #if defined (HAVE_HDF5)
     ....
  1243  #endif
and started with :
1219  %! if (!isempty(findstr(octave_config_info ("DEFS"),"HAVE_HDF5")))

so that, if the type of fortran was involved, this might be a thing  
to look at.

Jean-Francois