Re: who uses Octave?

[George Kousiouris]

On Fri, March 26, 2010 18:09, Søren Hauberg wrote:
> fre, 26 03 2010 kl. 11:32 +0200, skrev George Kousiouris:
>>
>
> I have to admit that I don't really understand the above as I don't
> really know the first thing about software-as-a-service and similar
> concepts. But it sounds interesting. Could you explain briefly to an
> ignorant such as myself what this means in practice?

Well, when one is offering software as a service, it could mean two
things: offering its use through some kind of web interface, like the one
that circulated some days ago with the web server where you are able to
upload scripts and execute them, but it can also mean enable one software
to be used in a service sequence diagram. Our contribution is the latter
case. In many distributed environments, there are for example toolkits,
most of them written in Java, that perform a variety of functions, such as
service registry, service management, execution etc. One form of these
management actions may be performance estimation. For example when you
want to have an estimation regarding how much time a job will take to
execute, you take into account a set of historical data and perform some
kind of processing. This is a service that exposes an interface and can be
used by other services automatically in an integrated fashion. Each
service performs an action that gives added value to the sequence. For
example in this case we would have a service that takes as input a data
set, does some internal processing and produces an output. This can be
done by being invoked by another service that knows this interface and has
incorporated the "client" code inside.

Up to now, in the service oriented performance estimation field, this
meant for example to write a Java program that implements one specific
estimation method. This is of course time consuming and pretty complex,
not to mention Java drawbacks in implementing mathematical algorithms. So
we thought that an implementation that has this aforementioned service
interface but uses in its core Octave as the "processing" program would
decouple the method from the framework. Furthermore, it would minimize
development time, since Octave provides a wide range of tools for
implementing complex methods.




> As I said, I don't quite understand what you have achieved, but it
> sounds interesting to me. So, if you are willing to share, then I'm sure we
> can help you. How you share this stuff depends on what kinds of changes
> you've made. If you've changed how the interpreter works or something like
> that, then you should start a thread about your work on
> address@hidden If you provide a set of scripts and functions then
> I guess it would be more suitable to share them on Octave-Forge as
> a package. If this is the case, then you can start a thread on
> address@hidden about your work.
>

We didn't change the core octave, we built a framework around it, that
enables it to  be executed in an automated service oriented way. This
consists mainly of suitable Java classes, interfaces and according Octave
scripts, so I guess the second list is what we must pursue.

If you need any more info don't hesitate to contact me.

Hope I have clarified things and not complicate them :)


Best regards,
George


>
> Thanks
> Søren
>
>
>


----

George Kousiouris
Research Associate
Division of Communications,
Electronics and Information Engineering
School of Electrical and Computer Engineering

E-mail: address@hidden

National Technical University of Athens
9 Heroon Polytechniou str., 157 73 Zografou, Athens, Greece