will parts of sandbox move to boost?

Hi.
  Is there any work to submit sandbox files
( especially "boost/numeric/bindings/traits" ) to
BOOST ?

Thanks,
  Dima.
_______________________________________________
ublas mailing list
ublaslists.boost.org
htt
p://lists.boost.org/mailman/listinfo.cgi/ublas

Am Freitag, 23. März 2007 19:35 schrieb Sorkin Dima:
> Hi.
>   Is there any work to submit sandbox files
> ( especially "boost/numeric/bindings/traits"
) to BOOST ?

The question is: Do we really need the complicated
bindings-traits logic 
if we add support for (vendor) BLAS to uBLAS? 

mfg
Gunter
_______________________________________________
ublas mailing list
ublaslists.boost.org
htt
p://lists.boost.org/mailman/listinfo.cgi/ublas

On Fri, 23 Mar 2007, Gunter Winkler wrote:
> Am Freitag, 23. März 2007 19:35 schrieb Sorkin Dima:
>> Hi.
>>   Is there any work to submit sandbox files
>> ( especially
"boost/numeric/bindings/traits" ) to BOOST ?
>
> The question is: Do we really need the complicated
bindings-traits logic
> if we add support for (vendor) BLAS to uBLAS?

But BLAS is not the only math-lib to which one wants to pass
vectors.
Traits are important building block for creating interface
for any
legacy lib. It could be helpful to "standartize"
it.

See for example the following example
http://ganymed.iwr.uni-heidelberg.de/piper
mail/dealii/2007/001837.html
(thread starts with
  http://ganymed.iwr.uni-heidelberg.de/piper
mail/dealii/2007/001822.html
)

Regards,
  Dima.
_______________________________________________
ublas mailing list
ublaslists.boost.org
htt
p://lists.boost.org/mailman/listinfo.cgi/ublas

Am Freitag, 23. März 2007 20:59 schrieb Sorkin Dima:
> But BLAS is not the only math-lib to which one wants to
pass vectors.
> Traits are important building block for creating
interface for any
> legacy lib. It could be helpful to
"standartize" it.

IMO this standardization was the goal of the bindings.
However it is 
impossible to include bindings to all existing libraries
that are used 
together with ublas. We have only a chance to add support
for already 
standardized libs like BLAS, LAPACK and maybe UMFPACK.

mfg
Gunter
_______________________________________________
ublas mailing list
ublaslists.boost.org
htt
p://lists.boost.org/mailman/listinfo.cgi/ublas

On Fri, 23 Mar 2007, Gunter Winkler wrote:
> IMO this standardization was the goal of the bindings.
However it is
> impossible to include bindings to all existing
libraries that are used
> together with ublas. We have only a chance to add
support for already
> standardized libs like BLAS, LAPACK and maybe UMFPACK.

I afraid we misunderstood each other.
I meant just the traits classes (vector_traits,
matrix_traits etc).
I didn't meant the libs themselves, although having BLAS and
LAPACK is a
good bonus too.

Dima.

_______________________________________________
ublas mailing list
ublaslists.boost.org
htt
p://lists.boost.org/mailman/listinfo.cgi/ublas