Minutes of the offline meeting held on 30 March 1999

Present: P.Mato,  T.Cass, P.Colrain, M.Cattaneo, V.Breton, F.Harris, A.Jacholkowska, 
              P.Maley, G.Corti, J.Harvey,  D.Steele, G.Gracia, F.Ranjard

Minutes : Florence Ranjard

      1.  Conclusions of the LHC++ workshop (P.Mato)

• LHC++ workshops are organized twice a year: users and developers can meet.
• between these workshops there are regular meetings at CERN: everybidy can attend.
• the referee is J.Knobloch.
• the LHC++ team needs manpower and would like more interactions with end users.
• the LHC++ team is interested in Java:
  • a browser of objectivity database objects developed by BaBar,
  • JAS (positive interest from various experiments).
  • nagC does not intend to write a C++ interface but a Java one.
  • ==> how to move from C++ to Java ?  using CORBA or rewriting C++ classes in Java
• All packages developed in LHC++ assume an Objectivity model
  • ==> how does that fit in the Gaudi framework?
• a Calibration Database is being developed following the model of the BaBar Conditions Database.
• ATLAS would like a low level persitency package for histograms : something lighter than Objectivity:
  • ROOT?
• packages are installed on various platforms using the default compiler.
• It was agreed that packages will be also installed using the g++ compiler : but porting should be made by us.
• We must remember that we cannot link together packages compiled with different compilers:
  • ==> it will be difficult to use Objectivity ( no g++ version) in the Gaudi framework (only g++ version).
• Today if we want the best Unix compiler we should probably use PC's and Linux.

2. Geant4 Technical Steering Board meeting (F.Ranjard)

     The 2nd Geant4 TSB meeting took place at CERN on 25 March 1999: Video Conference with Japan, Vancouver and SLAC.

• the next Geant4 release is forseen for end of May: 3 possibilities would be avalaible
  • global (as now): one huge library
  • full granularity (78 libraries) : should be the default
  • medium granularity
• Geant4 binary release policy:
  • binary releases for major releases only,
  • patch releases every 2 weeks
  • study the use of shareable libraries.
• Rogue-Wave will be removed in the next release and replaced with STL.
• the CPU timing problem mentionned in the 1st TSB meeting is not real:
  • the electomagnetic physics processes take more time in Geant4 than in Geant3, but they are better.
  • the navigation inside volumes and the tracking is much more efficient in Geant4 compared to Geant3
  • ===> in a real application usinf a complicated geometry Geant4 and Geant3 CPU time are similar.
• The G3toG4 package does not work well, it is not part of Geant4, it is using Rogue-Wave and cannot be
  move to STL quickly.
  • ===> users should be discouraged to use it.
• StdHep the standard interface to event generator : nothing to report.
• bug reporting system: Buzilla is free software with a very nice Web interface.
• Documentation : LIGHT is a very nice system but IPT does not support LIGHT1 and LIGHT2 does not exist yet.
• The next CLHEP version will provide a way to store the random number generator seeds on a file and to initialize
  the random number generator with them to reprocess an event.

          3. Sicb under CMT (F.Ranjard)

• The March 16th presentation has been updated (ppt)
• SICB v116 is avalaible now in a granular form on $CVSROOT and $SOFTWARE/LHCbCMT
• A presentation of the new way of working could be made at a next physics meeting.
• developers are invited to use it as soon as possible and to report any difficulty to Florence.
• Florence is avalaible at any time to give help.

   ================

    next meetings     no meeting on Tuesday 6 April
                             Tuesday 13 April    ( at 9.00  <===) - CORBA presentation by N.Neufeld