[Pkg-octave-devel] Presentation of Debian Science/Octave/Scilab/Scicomp, roundtable on free software

Thu Oct 2 08:27:55 UTC 2008

Hi,

I was wondering if there are some presentation of Debian Science and the 
Octave team  available, and in particular information like the following
  - objectives, 
  - policies, 
  - how it functions, 
  - common issues faced during software packaging( licensing, technical 
issues, build system, upstream...), eg. Salomé issues
  - some numbers: 
    o approximate number of working on the project, 
    o approximate number of people in mailings lists, 
    o origin of the people: continents represented, number of countries 
represented 
    o number of packages maintained

I already have some of these information for pkg-scicomp and one slide ready.
If the information is not available, that's ok too :), perhaps you can just 
send me some of these data.

Silvestre, 
do you have anything like that for Scilab and its integration in Debian ?

Why would I need that
---------------------

Next week I organize with some colleagues a workshop in Paris (EDF/Clamart 
[1]) October 8 in the context of a project (OPUS) financed by the French 
national research agency  (ANR). OPUS stands for Open Platform for Uncertainty 
quantification in numerical Simulation. The partners are some 
international/french  companies EDF, EADS, Dassault Aviation, Softia, some 
institutions CEA and INRIA and some academics Supelec, Université Joseph 
Fourier and école Centrale Paris. At the end of the project, we should deliver 
a LGPL platform fit for industrial problems.

 1. EDF is the main electricity supplier in France, EDF has a big R&D
   department

OPUS will use OpenTURNS [2] (a EDF, CEA and Phimeca software) which I packaged 
recently. The announcement of OpenTURNS being in Debian generated very good 
feedback from the OpenTURNS community and the companies involved: they are 
very excited/proud (and stressed out) of having their software in Debian. I 
learned also that Debian was one of the development platform and daily used by 
the researchers and engineers in these companies which is a good sign for us.

 2. http://trac.openturns.org

With this recent success, I would like to use this first OPUS workshop to 
advertise the work being done by the Debian Science community and the other 
related projects: med, scicomp,...  Indeed in the afternoon of the workshop we 
organize a roundtable on free software  in the context of OPUS and hence 
engineering, scientific computing and industrial numerical simulations.

During this round table we will have the pleasure to have, among others, 
 - R Gentleman lead developer of R (by teleconference)
 - D Bateman one of the main developer of Octave 
 - some people from Scilab (Sylvestre and some others)

and some people from Debian (Sylvestre, me and perhaps others I don't know)

The round table will be split into 2 parts:
   1- developing an open platform like OPUS : issues, process, 
architecture,...
   2- continuation of the project (financing, licensing, services, 
maintenance,...)

The following questions will certainly be raised but not only

 - what should we be careful about ?  what are the pitfalls to avoid ?
    o intellectual property, know-how, what/how to protect it ?
    o copyright
    o licenses
    o proliferation of licenses: many software integrated in the platform
      means potentially many new licenses which might be incompatible and
      non-free, hence forbidding distributing the software
    o technical issues, architecture, design...
 - what are the good processes for free software with industrial and academic
  partners ?
    o steering committee ?
    o platform architecture
    o communication  (mailing lists, web, ...)
    o building tools
    o releases, milestones, release often ?, ...
    o dissemination, acceptance by scientific/industrial community,...

With this round table, I would like to position Debian Science/Scicomp  as a 
partner and a community that has over the years provided with Debian a leading 
platform for science in general and scientific computing/numerical simulations 
in particular. I would like to voice the issues that we face with packaging 
distributing scientific software and make sure that our work is somehow 
simplified by having a better collaboration and feedback with upstream. We 
certainly developed a good sense of what should be done and what should not. 

I believe this is a very good opportunity, there will be not only people from 
OPUS/OpenTURNS but also people from or connected to Salomé, Code Aster 
possibly Open Cascade and certainly quite a few other big software. This is an 
opportunity to place our community under the radar of these projects and 
possibly create some interesting momentum for the development of Debian 
Science/Scicomp.

By the way, if you happen to be in Paris on October 8 feel free to send an 
email to Guennadi Andrianov <guennadi.andrianov at edf.fr> if you would like to 
attend the workshop [3] with some information about your professional 
affiliation. The web site is not open yet, we just display some information 
about the agenda, location and content of the workshop. Unfortunately the 
number of places is limited, so be quick. I shall send a report on the 
discussions at the round table.

 3. http://opus.softia-systems.net

Ok that's it. 
Don't hesitate to comment on this, remarks and advices are most welcome.

Adam,

could you send me an update on Salomé: hours of work, amount changes that were 
done, number of people involved in the packaging, communication with 
upstream...) ? Salomé might provide some good examples of what should not be 
done in such platforms.  I will be cautious about criticism though: I want the 
discussion to be constructive.

Best regards
C.
-- 
Debian Developer
Annecy - Grenoble
Scientific computing related software

-- 
Christophe Prud'homme
Université Joseph Fourier      christophe.prudhomme at ujf-grenoble.fr
LJK - Room 55                  Tel: +33476635497
51, rue des Mathématiques      Fax: +33476631263
BP53X 38041 Grenoble Cedex 9   
      <http://ljk.imag.fr/membres/Christophe.Prudhomme/>      