Re: Octave project involvement

[John W. Eaton]

On  7-Sep-2007, David Bateman wrote:

| A.B. Kinsman wrote:
| > Dear maintainers of Octave:
| > 
| > First let me say this is a great project, and I am excited about taking
| > a crack at contributing something. I have begun to work on the
| > suggested improvement of extending the meaning of .* to include by row
| > and by column scaling, and wanted to be sure this will not be redundant
| > effort. Please let me know if you have any suggestions.
| > 
| > Thanks,
| > 
| > Adam Kinsman
| > McMaster University, 
| > Ontario, Canada
| > 
| 
| 
| In fact there was a patch for that one against a much older version of
| octave (version 2.1.31), though its probably not that hard to port to a
| current version of octave. See
| 
| 
http://octave.cvs.sourceforge.net/octave/octave-forge/extra/patches/M-v-ops-2.1.31.patch?hideattic=0&revision=1.1&view=markup
| 
| However, this patch wasn't accepted at the time (ask John why) and
| although the idea is nice, any patch might not be accepted now. So some
| feedback from John is probably a good idea. before embarking on this
| particular task.
| 
| Also note that the same behavior might be had with the use of repmat, or
| the bsxfun function I just introduced.. Maybe these are sufficient,
| though some benchmarking is probably in order.

Using repmat for this job is OK but can use up a lot of memory
unnecessarily.

Although it would probably be good to have some kind of function to do
row/column scaling, I no longer think that it is desirable to overload
the .* operator to do it because of the potential for confusion about
when row or column scaling should occur that could easily lead to
undetected errors.  Probably it would be better to add a function for
this.  I don't know what the best name or interface would be.

jwe