Re: Metis

[David Bateman]

Tatsuro MATSUOKA wrote:
> --- David Bateman <address@hidden> wrote:
>   
>> Not really, as Octave doesn't in fact link to Metis.. It links to
>> Cholmod that then links to Metis. If the user built Cholmod to use
>> metis, then our only choices are to link to metis or to not link to
>> cholmod at all.
>>     
>
> I think it's not true
> The following is the a part of UFconfig.mk of suitesparse 3.0.0.
> #------------------------------------------------------------------------------
> # METIS, optionally used by CHOLMOD
> #------------------------------------------------------------------------------
>
> # If you do not have METIS, or do not wish to use it in CHOLMOD, you must
> # compile CHOLMOD with the -DNPARTITION flag.  You must also use the
> # "METIS =" option, below.
>
> # The path is relative to where it is used, in CHOLMOD/Lib, CHOLMOD/MATLAB, 
> etc.
> # You may wish to use an absolute path.  METIS is optional.  Compile
> # CHOLMOD with -DNPARTITION if you do not wish to use METIS.
>
> Building cholmod with -DNPARTITION option can realise the cholmod library 
> without metis.
> My cygwin and mingw binaries include cholmod without metis.
> Perhaps some features will be lost without metis but the cholmod itself can 
> be build without the
> metis.
>
> The suitesparse libraries with metis is now not allowed to be distributed by 
> the licence problem.
> For the users who would like to use the complete cholmod feature, only they 
> can do is build octave by
> themselves with metis.
>
> So I think the metis selection is not recommended to be removed.
>
> However I do not know how the octave uses suitesparse library.
> My opinion may miss the point.
>
>   

My point was not that its not possible to build a fully GPL compliant
version of Cholmod, if we couldn't then there is no way we could use it
in Octave. Rather my point was that Octave just uses the version of
Cholmod that the user compiles and we can't control whether they built
it with Metis or not.. However, the GPL forbids any distribution of the
resulting binary...

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