Re: [ublas] Missing "copy" in BLAS bindings?

From: Gunter Winkler (guwi17_at_[hidden])
Date: 2007-03-28 17:31:48

Am Mittwoch, 28. März 2007 23:23 schrieb Karl Meerbergen:
> Quoting Gunter Winkler <guwi17_at_[hidden]>:
> > Am Mittwoch, 28. März 2007 23:06 schrieb Karl Meerbergen:
> > > Hi,
> > >
> > > The bindings do no longer compile since the functions one1 and
> >
> > one2
> >
> > > do longer exist in the matrix functor_type. Can someone tell me
> > > whether this still exists in another name?
> >
> > Do you remember what one1 and one2 mean?
>
> No idea. I did write the ublas bindings, but I guess you are right:
> it is distance_i or ditance_j. I will try those.

The diff of functional.hpp from 1.37 to 1.38 shows:

row-major:

         static
         BOOST_UBLAS_INLINE
-        size_type one1 (size_type /* size1 */, size_type size2) {
-            return size2;
-        }

-        static
-        BOOST_UBLAS_INLINE
-        size_type one2 (size_type /* size1 */, size_type /* size2 */) {
-            return 1;
-        }

column-major:

         static
         BOOST_UBLAS_INLINE
-        size_type one1 (size_type /* size1 */, size_type /* size2 */) {
-            return 1;
-        }
-        static
-        BOOST_UBLAS_INLINE
-        size_type one2 (size_type size1, size_type /* size2 */) {
-            return size1;
-        }

so one1 and one2 are now 

one1 == address( 1, size1, 0, size2) - address(0, size1, 0, size2);
one2 == address( 0, size1, 1, size2) - address(0, size1, 0, size2);

Do you agree?

mfg
Gunter