From octave-sources-request at bevo dot che dot wisc dot edu  Fri May 31 09:58:33 2002
Subject: Re: Replacing LINPACK with LAPACK
From: Laurent Mazet<mazet at crm dot mot dot com>
To: "Steven G. Johnson"<stevenj at alum dot mit dot edu>
Cc: <octave-sources at bevo dot che dot wisc dot edu>,    David Bateman<David.Bateman@motorola.com>
Date: Fri, 31 May 2002 10:16:53 +0200 (MEST)


On 30-May-2002 Steven G. Johnson wrote:
> On Thu, 30 May 2002, David Bateman wrote:
>> The change to configure.in adds the flag "-without-linpack" that
>> forces the use of the code in my patch. The code in my patch is
>> entirely inactive without this compile flag, though I imagine that
>> this behaviour should eventually be changed.
> 
> If you're going to change it to use LAPACK instead of LINPACK (which I
> heartily approve...I didn't realize that Octave still used any LINPACK,
> since it already links to LAPACK), I think you should just delete the old
> code.  What's the point of leaving it as an option?
> 
> Steven
> 

(I worked with David on this patch)

We wanted to test that the inversion function in LAPACK and LINPACK gave us
same results. So we let both codes in liboctave (switched by PREFER_LAPACK
define). If you want to delete the old code, just cancel the patch on
configure.in and delete old code in liboctave/dMatrix.cc and
liboctave/CMatrix.cc

If John wants to integrate it into CVS, we can rewrite it without the old code.


    Laurent
-- 
Dr. Laurent Mazet: Research Engineer /V\ Centre de Recherche de MOTOROLA
Tel: +33 (0)1 69 35 48 30 =-=-=-=-=-=-=-=-=-=-= Email: mazet at crm dot mot dot com