Re: [CHANGESET]: First attempt at a single precision type.

[David Bateman]

John W. Eaton wrote:
> On 29-Apr-2008, David Bateman wrote:
>
> | Agreed that there is the issue of maintenance to address, but if there
> | has to be specializations for the LAPACK/BLAS codes that won't go away.
> | Even functors will just increase the amount of code to maintain in my
> | opinion
>
> Yes, and it tends to be fairly complex, so I agree we need to consider
> whether it is worth it.
>
> | in that case, and the calling conventions of some complex LAPACK
> | functions are not the same as their real counterparts and so the methods
> | with functors in those cases could even be included in the
> | MArray2/MArrayN classes.
>
> Yes, those would cause trouble.  I was thinking about float/double.
> Are those at least the same?
>   
Yes, I can't think of a case where they aren't the same. However, how
could we have two different methods with the same calling convention in
the MArray2/MArrayN classes to handle the real/imaginary cases.. I
suppose something like


#define COMPLEX
#include "MArray2.h"
#undef COMPLEX

in the CMatrix.cc code and then different versions in MArray2.cc might
work..

> | There is also still the issue of the effect
> | such changes would have on the maintainability of user types in
> | octave-forge.
>
> I guess I was hoping to keep the interfaces of the
> Matrix/NDArray/etc. classes the same as they are now.  Would it matter
> if we added functions to the MArray classes?
>   
I'd only matter if the code added had new dependencies on code outside
the Array and MArray classes as far as I can see.

D.



> jwe
>
>   


-- 
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