[MMTK] Re: Problems with FC4, gcc4.0.1 build

khinsen at cea.fr khinsen at cea.fr
Mon Nov 7 10:08:13 CET 2005

On 07.11.2005, at 09:45, seth.olsen at gmail.com wrote:

> I am having problems building MMTK on a Pentium M machine running  
> Fedora 4 with gcc 4.0.1.  I have previously installed Scientific  
> Python v. 2.4.9 without significant problems.  When I perform  
> 'python setup.py build' in the MMTK source directory the build ends  
> abruptly during the lapack_mmtk extension build.  The output after  
> this command is listed below.  (The cut and

GCC 4 is stricter about the interpretation of the C language rules  
than its predecessors (and most other C compilers). It doesn't accept  
static function declarations inside functions. The C-LAPACK code that  
is included in the MMTK distribution (it is generated from the  
Fortran version using f2c) uses such declarations, but it is rather  
straighforward, though a bit of a chore, to take them out of the  
functions. This has been done in MMTK 2.4.4 and 2.5.10, so please  
update to one of these two releases.

> Any ideas what the problem may be?  I have installed the newest  
> updated lapack libraries from the fedora yum rpm repository.   
> Oddly, I am getting similar 'Invalid storage class'  errors for a  
> SciPy build that also has failed...

All f2c-generated code suffers from that problem. I suppose that  
SciPy also has some of it.

Konrad Hinsen
Laboratoire Leon Brillouin (CEA-CNRS), CEA Saclay,
91191 Gif-sur-Yvette Cedex, France
Tel.: +33-1 69 08 79 25
Fax: +33-1 69 08 82 61
E-Mail: khinsen at cea.fr

More information about the mmtk mailing list