Re: [OctDev] linear-algebra: new QR updating functions

[John W. Eaton]

On 19-Feb-2008, David Bateman wrote:

| Well the F77 wrapper makes the DEFUNs almost one liners... If we move
| the lapack/blas wrapper code from F77 to C++ it'd be a bit longer, maybe
| 30 lines or so per function. Encapsulating them in the Matrix and
| ComplexMatrix class would be fairly trivial.
| 
| > 
| > | John, we should include these functions.
| > 
| > Where can I look at them?  I'm not on the octave-dev list.
| 
| You can either check out the octave-forge SVN or look at them on the web
| interface at
| 
| 
http://octave.svn.sourceforge.net/viewvc/octave/trunk/octave-forge/main/linear-algebra/src/

OK.  Which files/functions are you proposing to include in Octave?  I
think it is probably OK to add them as is.  As you said, the Fortran
code should probably go in libcruft, though it seems odd to be putting
much new code there.  In any case, it is the only part of Octave that
is really set up for compiling Fortran code, though maybe it wouldn't
be too much work to add rules for Fortran to the liboctave Makefile.

| > Is there any way to put f77 functions inside a namespace?
| 
| That's not the way the rest of the F77 code in Octave is used..

Yes, I know, I was just pointing out a potential problem with f77
code.  It seems much more likely to have conflicting global symbol
names, especially on systems where the Fortran compiler doesn't append
underscores to external names.

jwe