Re: Metis

[David Bateman]

John W. Eaton wrote:
> On 17-Oct-2007, David Bateman wrote:
>
> | The check for cholmod at the moment looks for METIS and defines
> | HAVE_METIS if it exists.. We in fact don't use the HAVE_METIS flag.
>
> But we do, in liboctave/sparse-base-chol.cc, liboctave/CSparse.cc,
> liboctave/dSparse.cc, liboctave/oc-sparse.h.  Also, the file
> src/DLD-FUNCTIONS/spchol.cc uses METIS_VERSION.
>
> jwe
>
>   
Opps yeah I forgot about that bit of code.. In fact its the same bit of
code everywhere and is just block of code that forces versions of metis
later than 4.0.2 to use a particular implementation of malloc and free
and to test for out-of-memory conditions before calling metis if that
can't be done. The free/malloc bits can be dropped and in later versions
of CHOLMOD the memory guard code is the default.. So if we don't want to
set the free/malloc functions for metis, this is just cruft and can be
completely removed.

D.

-- 
David Bateman                                address@hidden
Motorola Labs - Paris                        +33 1 69 35 48 04 (Ph) 
Parc Les Algorithmes, Commune de St Aubin    +33 6 72 01 06 33 (Mob) 
91193 Gif-Sur-Yvette FRANCE                  +33 1 69 35 77 01 (Fax) 

The information contained in this communication has been classified as: 

[x] General Business Information 
[ ] Motorola Internal Use Only 
[ ] Motorola Confidential Proprietary