Re: embedded octave: how to define new functions?

[Przemek Klosowski]

Andrea,

You said that you want to embed octave into my code, and in your latest
posting you added: 

    As I didn't want to write any command line interface or a
    main's-argv parsing routine (the octave's ones are ok and I
    actually need only to extend the octave's instruction set with my
    own functions and data structures, not more)

Please explain why don't you just want to add your functions and
structures to Octave, using the 'mkoctfile' facility. The main reason
for 'embedding' octave is to avoid the octave interpreter loop
(command line handling and evaluation), presumably because you can't
replace your own pre-existing main eval loop for some reason.

Octave has real, mature language facilities, so it makes a lot of
sense to refactor existing code to use it: chances are that you'll
need some commandline facilities to read/write data, set up internal
variables, start your calculation, test for convergence, etc. No sense
in maintaining another inferior scripting language.

There is the issue of licensing, too. Obviously, code written as
extensions to Octave inherits its GPL license, so if it is at all
distributed, it must be Open Sourced. This does not change if one uses
embedded Octave: since liboctave is under the GPL, it still applies
even if you only link to liboctave---so licensing can't be a reason to
embed Octave.