On 12/13/2009 [email protected] wrote:
> Date: Sun, 13 Dec 2009 13:41:32 -0500
> From: Eugeniy Mikhailov <[email protected]>
> Subject: Re: [OctDev] optim/leasqr.m style changes, diff
> To: [email protected]
> Message-ID: <20091213184132.ga26...@beam>
> Content-Type: text/plain; charset=us-ascii
> 
> Hi Olaf,
> 
> Thanks for clarification of your diff.
> 
> >> > > I see that your changes are not Matlab compatible. Personally, 
> I have no
> >> > > interest in maintaining Matlab compatibility, but perhaps 
> others do?
> > > 
> > > Yes, maybe some people who need leasqr must write code that runs 
> both
> > > in Octave and Matlab. I could revert all incompatible changes, 
> making
> > > life a bit harder for me ... Note, however, that there is no way 
> (?)
> > > to get the helptext (without copyright/license notice) displayed 
> in
> > > Octave _and_ Matlab without changes to the file (if a
> > > copyright/license notice is present). One could also leave it to 
> the
> > > respective users to make some changes for compatibility, as they 
> would
> > > have to do also with other Octave functions if they have need for
> > > Matlab compatibility. leasqr was originally Matlab compatible 
> because it
> > > came from a Matlab-related repository.
> I would also strongly argue that we should not brake Matlab
> compatibility. While I am personally try to stay away from the Matlab
> beast, I sometimes have to use it since it supported by our cluster
> maintainers and Octave is not (octave still missing easy parallelism
> despite the effort of the people in the parallel thread).
> 
> But I do debug my code in octave, and I don't want to run octave to 
> Matlab
> converters just because Emacs cannot highlight staff properly.
> 
> I also would be happy to review your leasqr patches , since I use this
> function quite often.

My understanding is that Matlab compatibility is encouraged for the case 
of _functionality_ and not necessarily for coding syntax and style.

Regarding leasqr, there is not a "leasqr" function in Matlab but instead 
a "lsqnonlin".  This function does not yet exist in Octave (or 
octave-forge), although recently lsqnonneg was created.  If there is 
interest in working on leasqr, I would suggest instead (or in addition) 
creating a lsqnonlin function in a way that is compatible with Matlab's. 
  As always, this should be done without looking at the contents of 
Matlab's code for legal reasons.

Jonathan

------------------------------------------------------------------------------
Return on Information:
Google Enterprise Search pays you back
Get the facts.
http://p.sf.net/sfu/google-dev2dev
_______________________________________________
Octave-dev mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/octave-dev

Reply via email to