Re: Wikibook on octave

[David Bateman]

Nicolas Pettiaux wrote:
> 2007/1/29, David Bateman <address@hidden>:
>> Nicolas Pettiaux wrote:
>
>> Hi Nicolas,
>>
>> I'm not sure I see the interest in creating a separate WikiBook for
>> Octave, though of course we value the effort. My reasoning is that as
>> incompatibilities between Matlab and Octave are in most cases considered
>> as bugs, then the Matlab WikiBook should in an ideal world equally apply
>> to Octave. Therefore you are creating additional work for yourself in
>> writing a separate contribution and you'll have to continually keep
>> chasing the original document and porting changes between version.
>
> I fully agree.
>
> Please read
> http://en.wikibooks.org/wiki/Talk:MATLAB_Programming#Octave_.2Aand.2A_matlab_programming
>
> where you'll find my suggestions to first rename the book about
> Matlab to "Octave and matlab programming" . There and in and
> http://en.wikibooks.org/wiki/User_talk:Npettiaux you'll find some
> replies of the heavy contributors of the above book, who are not
> fravorable
>
Expense is not the argument, as discussed in the comment but rather the
number of differences between octave and matlab and whether the syntax
is close enough to justify that a single book can discuss both..... That
being the case I'd argue that the differences aren't that many and they
are being worked on the bring similar levels of compatibility. The only
areas of difference in the core matlab/octave functionality between
2.9.9+ and matlab r2006b are

1) OO programming in octave is not supported in the same manner as in
matlab. Its not impossible as the OO functionality of C++ itself can be
used in an oct-file to get this type of functionality, but it certainly
isn't compatible. John has mentioned in the past that he has funding to
develop matlab compatible classes. cf.
http://www.cae.wisc.edu/pipermail/help-octave/2006-November/002337.html

2) The graphics handles are handled in a matlab compatible way.. There
are numerous external toolkits to bring this functionality (octplot, 
yapso, etc etc), and there are moves to include similar functionality in
octave 2.9.10 (cf
https://www.cae.wisc.edu/pipermail/octave-maintainers/2006-December/001435.html)

3) There is no profiler in octave. There are recent patches to do
something about this, so maybe soon there'll be something..

4) There are no compatible GUI functions... There is some code for this
using tcl/tk and zenity, but nothing compatible. I also know of no work
to make it compatible

5)  Simullink. Ok simullink is a toolbox and not a core functionality of
matlab, but lots of people criticize octave as not have this
functionality.. Scicraft (http://www.scicraft.org/) supplies similar
functionality and can call R and octave as the compute engine

6) Missing toolbox. Point them to octave-forge which has a lot of the
missing functions already available..

D.



-- 
David Bateman                                address@hidden
Motorola Labs - Paris                        +33 1 69 35 48 04 (Ph) 
Parc Les Algorithmes, Commune de St Aubin    +33 6 72 01 06 33 (Mob) 
91193 Gif-Sur-Yvette FRANCE                  +33 1 69 35 77 01 (Fax) 

The information contained in this communication has been classified as: 

[x] General Business Information 
[ ] Motorola Internal Use Only 
[ ] Motorola Confidential Proprietary