Data Management meeting held the 04/11/2003

Present: C.Cioffi, P.Charpentier, J.Closier, M.Frank, 
         G.Kutznetsov, F.Loverre, S.Ponce, A.Soroko,
         A.Tsaragorotzev

Bookkeeping status:
===================
The new version of the bookkeeping interface was supposed to be ready.
However in the meantime a new requirement came up, that the input
files of a given job may origin from various programs. This makes
the choice of selecting by this program type/version difficult.
It was decided to only select by the program, which produced the
signal events and ignore spillover sources.

The Ganga interface to the bookkeeping will receive this information
in form of a XML file.

Ganga status:
=============
Alexander presented his roadmap to implement the Ganga job options editor.
Here is in brief the outcome of the discussion:
1) Ganga will use a text file to determine the components of each dll.
   Building this description will be part of the dll build procedure.
2) The evaluation of restrictions and dependencies will be postponed.
3) Input/output datasets will require special handling by Ganga.
4) The rpc connection to the job configuration database will be maintained.
   This connection allows to determine defaults for "known" applications.
5) Alexander will contact Ulrik to determine the correct user interface
   look and feel.

Dirac status:
=============
Dirac with the new workload management system is now able to submit simple
jobs to the EDG testbed. The next step is to submit full production jobs.
The software packaging system must be made aware of CMT projects. Also
the LCG software needs such packaging.
Further modifications to the job scripts are planned. Ganga will then 
interface to these job scripts.
Gennady promised a workable form of the production editors by the end
of the week.

POOL integration status:
========================
Markus managed to write Relations tables with POOL. With this the 
basic requirements to write all objects in the LHCb event model
are fulfilled and we have to wait for the updated Gaudi object
description for further tests.
Carmine implemented the POOL based event selector based on the 
proposed interface. He will continue with the main Gaudi event 
selector upgrade.
Pere will have to communicate the changes to the event selector
interface to ATLAS.


Minutes:
     Markus Frank