Boost logo

Ublas :

From: Andreas Klöckner (lists_at_[hidden])
Date: 2008-08-16 09:10:17


Hi Thomas, all,

I've merged the bindings svn changes as of r48172 into my tree. The remaining
diff is at http://tiker.net/tmp/bnb-2008-08-16.diff

 lapack/gbsv.hpp | 136 ------------
 lapack/geev.hpp | 20 -
 lapack/gels.hpp | 213 -------------------
 lapack/gelsd.hpp | 420 --------------------------------------
 lapack/gelss.hpp | 358 --------------------------------
 lapack/geqrf.hpp | 8
 lapack/gesvd.hpp | 10
 lapack/heev.hpp | 8
 lapack/heevd.hpp | 22 -
 lapack/heevx.hpp | 12 -
 lapack/lapack.h | 74 ------
 lapack/lapack_names.h | 27 --
 lapack/syev.hpp | 11
 lapack/sygv.hpp | 211 -------------------
 umfpack/umfpack_inc.hpp | 2
 15 files changed, 54 insertions(+), 1478 deletions(-)

Here's what it consists of:

- Vardan Akopian's gbsv
- Georg Baum's complex type flexibility
- Jesse Manning's gels*
- Some compiler fixes where adding an <int> template parameter made something
better.
- A few #ifdefs that prevent unused variable warnings.
- A syev() without a workspace argument.
- Change <umfpack/umfpack.h> to <umfpack.h>. This is sensible because umfpack
is mostly distributed as part of suitesparse, so that users have to hack the
header to get rid of the extra umfpack/ path. (or clutter their include
directories with a symlink)

At least the last four of those should be easy merges. Thomas?

Another point: when unifying over complex and real matrix properties
(orthogonal/unitary, symmetric/hermitian), we should pick one, and stick to
it. There's some inconsistency in there already: we use hegv, but orgqr. I'd
suggest we stick to the 'real' terms (orthogonal, symmetric).

Andreas