some thoughts about the future of Octave

[courtois]

Hi All,

I work both as a researcher and a teacher and I am an everyday user...
but I am not a power user.

First, thanx to jwe for bringing up this marvelous tool that is called
"Octave". We are currently using it in replacement of Matlab for
teaching in our college of Engineering. It has 2 advantages:
        -we will be able to get rid of our problematic NT server that
         crashes so often for unknown reasons
        -it costs much less (linux+octave / NT+Matlab in a client
         /server environment).
        -it is cross-platform : UX + Win
        -it allows us to give a copy of octave for free to the students
         for their future work (Matlab would be so expensive)

This can work mainly because Octave is close to be compatible with
Matlab... the "de-facto" standard.

We, here, enjoy the possibility to easily port the m and cmex files from
Matlab to Octave. The close compatibility is a major advantage that led
us to choose Octave. The fact we can compile it, taking advantage of 2
processors for example is a nice thing.

I must precise that I have been working with another similar software
called Scilab (freeware available on Win, UX and Mac)... I even made the
port from UX to MacOS. I spent many hours to accomplish that for finally
little interest. Scilab has many powerful toolboxes that can compete
with Matlab but it isn't very compatible with it. Major problem is that
you have to do
getf("myfile.sci")
comp(myroutine)
myroutine

instead of
myroutine

If I wouldn't be so interested in Matlab compatibility I would have kept
scilab, sincerely.

I have heard of other very interesting Matlab-like environnements like
RLab, R, Yorick...

As far as I know, Scilab is the only one that propose a matlab2scilab
converter (written in Scilab language). Why Octave wouldn't offer the
same ?

Clearly, I understand jwe but I confirm our interest (both for research
and education) in Octave being as close as possible to Matlab. Octave is
already quite compatible with Matlab 4.2 except that there are still
missing features that are quite annoying :
        -spline (pkienzle's package doesn't solve this)
        -fmins (there are downsimplex and levenberg etc... but nothing as 
generic)
        -plot (they don't look very nice, more like a raw output)
        -multiplot (same + problems to print them)
        -documentation is still very poor
        -no easy doc to help understand how to make octfiles

I have take a look at vtk and it looks so great ! Actually I am using
epstk which is very nice even if it can't do everything we can ask.

the command line interface is quite enough for us and we don't care for
any other interface. We would rather prefer to see more work on the
plotting library and the documentation.

I personnally think that a short documention explaning very precisely
(with examples, like a canvas) how to create standardized re-usable
mfiles to contribute to the community would be very important. For
instance, I did a library allowing to use rs-232 communications and I am
just wondering how to make it standard (help, parameters, manual) to
offer it as a contribution. I am afraid that if I release it as it, it
won't be understandable by others.

I feel that there is a gap between some of us who are very into Octave
(Kienzle, Grossman...) and people like me who are not mastering Octave
so well but are able to give contributions as well, provided that we are
given more help (documentation mainly).

If I would give priorities in the TODO List it would be:
        1- write a canvas and a manual for adding new mfiles
        2- add a fmins function
        3- improve the website and documentation
        4- start a project for an automatic convertir Matlab2Octave
        5- change gnuplot for something else

I think these "users comments" can contribute to the debate and assure
again all of the OCtave developpers our sincere congratulations for
their wonderful work.

-- 

Francis

 =========================================================================
 Dr. Francis COURTOIS               I  ENSIA - INRA
                                    I  1, avenue des Olympiades
 Food Process Control Lab.          I  91744 MASSY Cedex
 Food Engineering Department        I  FRANCE
 -----------------------------------I  Tel : (+33) (0) 1 69 93 51 29
 mailto:address@hidden      I  Fax : (+33) (0) 1 69 93 51 85
 http://ensia.inra.fr/Šþcourtois     I  Cell: (+33) (0) 6 84 13 73 92
==========================================================================



-------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.

Octave's home on the web:  http://www.octave.org
How to fund new projects:  http://www.octave.org/funding.html
Subscription information:  http://www.octave.org/archive.html
-------------------------------------------------------------