Re: alphabetical list of Octave functions

[David Bateman]

Brian Kirklin wrote:
> Jordi Gutiérrez Hermoso wrote:
>   
>> On 10/06/2008, gOS <address@hidden> wrote:
>>   
>>     
>>>  simply because no one programming for my
>>>  company should be using something that won't be present in Matlab.
>>>     
>>>       
>> If only it were that nobody in your company should be using Matlab. :-/
>>
>> All this wasted effort on proprietary software for code that could
>> just as easily be made free, when we as a numeric community obviously
>> have the geekpower for it, and are already getting paid by employers
>> and academic institutions.
>>
>> But I digress.
>>
>> - Jordi G. H.
>>
>>   
>>     
> As long as there are advantages to using Matlab over Octave there will 
> always be people who use Matlab. In particular, you rarely have to worry 
> about bugs when working with Matlab, there are significant speed 
> advantages due to JIT compiling, the graphics end doesn't crash, and 
> everything is well documented, etc.
>   
Ok the JIT and a large number of toolbox functions is a given for
matlab, but I'd disagree on the issue of bugs.. I'd had a number of bugs
in Matlab in the past and the response from their support was always
along the lines of  "known bugs aren't bugs but rather are features".
The classic one was "speye(n).^0" which I believe they finally fixed in
2007b even though I reported it 5 years ago. The only bug I've had to
worry about in Octave recently have been ones that push the limits of
the compatibility between Octave/Matlab with things like function
handles defined in sub-functions.. Also having the Octave source means
that any bugs I find I can and do fix myself rather than waiting on the
whim of mathworks support staff to fix.

I've not had crashes with the graphics with gnuplot 4.2.3, though there
were issues with gnuplot 4.2.0 due to a bug in the handling of piped
data in gnuplot. The graphics in Octave is evolving quickly however and
so is incomplete relative to the equivalent matlab functionality, though
that doesn't mean it crashes. The Java 1.4 or earlier issue for
jhandles, can't really be considered the fault of Octave.. Matlab
delivers their own copy of java to avoid having to address such problems
which is a bit of bloat.. This is a matter of philosophy, If you have
upgrades available then use them and avoid incompatibilities and/or
bloat. Modern linux distributions all have auto update features for the
entire system (not just the bits supplied by Microsoft as is the case
for windows) to address this.

D.

PS. FreeMat is implementing a JIT based on LLVM and Octave will most
likely copy their code once it stabilizes and get the equivalent JIT
functional of matlab.

-- 
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