#
# See also slides in "oscar-data/trunk/2007/jan07-ornl/presentations/" area
#
#==========================================================================
[Thursday, 1/25/07]
 Attendees: 
    o Jean Parpaillon (INRIA)
	o Christopher Heller (Intel)
	o Matthias Hess (NEC)
    o Erich Focht (NEC)
    o DongInn Kim (IU)
    o Stephen Scott (ORNL)
    o Geoffroy Vallee (ORNL)
    o Thomas Naughton (ORNL)

 - Dongarra talk @ 10am
 - Stephen Scott introduction and ORNL overview
 - Voting:
     o OCG Steering Committee 2007 (2yr term)
          - balanced number between industry & academic/research
		  - nominations:

     o OCG Chair (1yr term)
          - nominations:

     o OSCAR Chair (1yr term)
          - nominations:

     o OSCAR Rel. Mgr. (1yr term)
          - nominations:
 - The voting discussion digressed, we may have an OCG but began 
   to look at how to attract folks and adjust the project (OSCAR)
   to support this need for contributions and promote "merit".
   Questions:
     o how to abstract developers?
     o how to keep active research (or can you)?
     o how to get new features -> for being "best" cluster stack, etc.
          * What features do we have to have?
          * How to generate merit w/o having project be led down
            a divergent direction?

 - Discussions let us to come back to a similar idea to the "czar" ideas
   we'd used in the past, where a person can be the lead developer for an
   area of the project.  This provides a good method for getting
   merit/recognition, also assists in communication (who do I talk to
   regarding foo?), and also helps to focus internal design/architecture by
   having clear responsibilities and know if they're getting resolved in
   sufficient time.  Need to identify what the current "working groups" or
   "sections" are for this as well as potential new areas for further
   contributions, e.g., "numerical libraries" person, or "GUI" person (maybe
   just user interface-GUI, etc.

 - <<lunch>>

 - discussion while looking through Erich's slides (see slides):
     o oscar-5 Roadmap, what was done/not-done
     o review oscar-5 features (what ultimately went in v5)

 - discussion of Geoffroy's slides (see slides): 
     o OSCARonDebian status update


#==========================================================================
[Friday, 1/26/07]
 - DongInn presentation on ODA (see slides), discussion points
     o simplification - reduce subroutines, mention of Object-Oriented 
	       approach, e.g., SIS's DB.pm DB:Clients, etc.
     o alternate DBMS engines, possibly look at SQLite
	 o work toward simplifying Database.pm & oda.pm (rename to mysql.pm)
     o DongInn will investigate the simplification of ODA access routines
	   possibly leveraging the generic get_*, set_*, del_* methods for the 
       various tables (doing relations manually).
	   Or, 
	           oda_table_access
			   $a = new oda_table_access("Image");
			   $a->list(name=>"oscarimage");
     o Propose requirements for general group agreement.
	 o NOTE: Not a top-priority as current code works, but is tough for 
       longer term maintenance.

 - Erich discussed the approach taken by SIS for working with their 
   database/tables (see above discussions)

 - Discuss OSCAR CLI work

 - Discuss OPKG Compiler (see slides)

 - Reviewed 2007 release/roadmap items of interests, updated 2006 list 
   see files: "release_plan_06_revised.odt" and "OSCAR07_dev_items.odp"