Re: looking for comments on porting code

[Jonathan King]

On Sun, 11 Feb 2001, Mostyn Bramley-Moore wrote:

> I'm investigating the possibility of porting the SPM99 Matlab code
> (http://www.fil.ion.ucl.ac.uk/spm/spm99.html) to Octave.  SPM99 is a
> research tool "used to test hypotheses about [neuro]imaging data".  

Wow, something I know a little bit about, for a change... :-)

I know that the porting of SPM has been something that a *lot* of people
have thought about for quite some time, but then nobody ever quite pulls
the trigger.  My presumption is that this is because when you're
shelling out anywhere between $200 and $700 per subject to acquire data,
the cost of a Matlab license (as opposed to a porting effort) gets lost
in the noise.

[snip]

> I imagine that the major work would be in implementing 3D arrays (last
> I was aware, Octave didn't support greater than 2D arrays), and the
> GUI features of Matlab 5.2 which are used.

Multidimensional arrays are a big deal, and I think there may be a couple
of other "gotchas" in there, too, although it's been awhile since I've
looked at the SPM code.
 
> The actual work would be done by a small team of software engineering
> students, over the course of a year.  
> 
> Do people here think that this would be a worthwhile project?  Is anybody
> working on some of this already?  

Worthwhile, certainly.  Also often contemplated, but I don't think
anybody's really done it; I can ask around a bit to make sure, if you
like.

> Would this be too hard for a group of 5-6 third year software
> engineering students?  Is Octave a good environment to port to/is
> there anything else that would be a better choice?

Five or Six people can get a lot done, especially if there's an Important
Deadline involved, that's for sure.  I know there are other Matlab-like
open or free software efforts out there, but I don't think any are nearly
as well-developed or close-to-Matlab as Octave.  An adaptation of SPM to
R, a free implementation of S/S+, could be extremely interesting given
the greater number of stats nerd using R/S, but that would be a much,
much larger project.

jking



-------------------------------------------------------------
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
-------------------------------------------------------------