[bindings] const-correctness

[Uldarico Muico-2]

Hi.  I noticed that matrix arguments for the blas functions like gemm  
are no longer const-correct as they were in numeric_bindings-v1.  Is  
this intentional, or am I missing something?
_______________________________________________
ublas mailing list
ublas@...
http://lists.boost.org/mailman/listinfo.cgi/ublas
Sent to: lists@...

[Rutger ter Borg-2]

Uldarico Muico wrote:

> Hi.  I noticed that matrix arguments for the blas functions like gemm
> are no longer const-correct as they were in numeric_bindings-v1.  Is
> this intentional, or am I missing something?

You're right. I've modified the generator a bit, and committed the result to
svn. All BLAS bindings should be const-correct now. Please let me know if it
works for you.

Cheers,

Rutger



_______________________________________________
ublas mailing list
ublas@...
http://lists.boost.org/mailman/listinfo.cgi/ublas
Sent to: lists@...

[Uldarico Muico-2]

Perfect.  Thank you!

On Sep 17, 2009, at 12:51 AM, Rutger ter Borg wrote:

> You're right. I've modified the generator a bit, and committed the  
> result to
> svn. All BLAS bindings should be const-correct now. Please let me  
> know if it
> works for you.

_______________________________________________
ublas mailing list
ublas@...
http://lists.boost.org/mailman/listinfo.cgi/ublas
Sent to: lists@...

[Uldarico Muico-2]

I forgot to mention there seems to be an extraneous argument

const integer_t k

that appears in several functions like gemm, herk, her2k, syrk, syr2k.

On Sep 17, 2009, at 12:51 AM, Rutger ter Borg wrote:

> You're right. I've modified the generator a bit, and committed the  
> result to
> svn. All BLAS bindings should be const-correct now. Please let me  
> know if it
> works for you.

_______________________________________________
ublas mailing list
ublas@...
http://lists.boost.org/mailman/listinfo.cgi/ublas
Sent to: lists@...

[Rutger ter Borg-2]

Uldarico Muico wrote:

> I forgot to mention there seems to be an extraneous argument
>
> const integer_t k
>
> that appears in several functions like gemm, herk, her2k, syrk, syr2k.
>

I've hacked in support for detection of number of columns of op(A); it's
committed, too. You will need to pass uppercase chars for the transa, transb
argument(s). I've not been able to test it yet, please let me know your
findings.

Cheers,

Rutger



_______________________________________________
ublas mailing list
ublas@...
http://lists.boost.org/mailman/listinfo.cgi/ublas
Sent to: lists@...

[Uldarico Muico-2]

There is a typo

(transa=='N'?traits::matrix_num_columns(a),traits::matrix_num_rows(a))

It should be

(transa=='N' ? traits::matrix_num_columns(a) : traits::matrix_num_rows
(a))

On Sep 17, 2009, at 4:39 AM, Rutger ter Borg wrote:

> I've hacked in support for detection of number of columns of op(A);  
> it's
> committed, too. You will need to pass uppercase chars for the  
> transa, transb
> argument(s). I've not been able to test it yet, please let me know  
> your
> findings.

_______________________________________________
ublas mailing list
ublas@...
http://lists.boost.org/mailman/listinfo.cgi/ublas
Sent to: lists@...

[Rutger ter Borg-2]

Uldarico Muico wrote:

> There is a typo
>
> (transa=='N'?traits::matrix_num_columns(a),traits::matrix_num_rows(a))
>
> It should be
>
> (transa=='N' ? traits::matrix_num_columns(a) : traits::matrix_num_rows
> (a))
>

Fixed. I've also removed the extra level-namespace, added the right files in
level1.hpp through level3.hpp, and removed the high-level stuff.

They should be close to useable now :-)

Cheers,

Rutger




_______________________________________________
ublas mailing list
ublas@...
http://lists.boost.org/mailman/listinfo.cgi/ublas
Sent to: lists@...