Re: embedded octave: how to define new functions?

[Andrea Latina]

Hi,

> | I've considered this possibility but, as our simulation code is
> | already productive and working fine with its -rather complicate- data
> | structures, I thought that the embedding is the best solution. (But
> | maybe I'm wrong.)
>
> Do you need to expose all of these things in a scriptable interface,
> or can you get away with just having a few functions to control your
> code from Octave along with some functions to get at the data that is
> interesting for later processing?

The first option you said: all these things must be exposed in a
scriptable interface. Our code (I'm not the main author, I'm taking
over and improving it) counts more than 200 functions and commands. It
is a code for simulating particle accelerators.

A typical simulation script consists mainly in internal function
calls, with some loop and a few control statements. The amount of data
produced is moderately small. Another typical study is performance
maximization: where the results of some simulation are optimized with
respect to some parameters. In this case, a language such as octave
would be perfect.

> BTW, are you distributing your code?  If so, is it freely available
> under terms that are compatible with the GPL?  If not, then you would
> certainly not be able to embed Octave in it and distribute it unless
> it has a license that is compatible with the GPL.

Yes, our code is freely available and already distributed under the
GPL license. You can have a look at this page:
https://savannah.cern.ch/projects/placet/ . Look at the manual -rather
outdated, unfortunately- if you want to have a flavour of a typical
simulation script.

In the meanwhile I'm proceeding in the embedding and the results seem
amazingly promising!

Cheers,
Andrea

P.S. Sorry, John, for the double reply. I clicked "reply" instead of
"reply all".