Minutes of the offline meeting held on 23 February 1999

Present: P.Mato, P.Maley, J.Harvey, G.Corti, D.Steele, I.Korolko,
            E.van Herwijnen,   G.Gracia, M.Cattaneo, M.Franck, F.Ranjard

Minutes : Florence Ranjard

      1. short resume of the 1st Geant4 Technical Steering board (F.Ranjard)

        The first Geant4 TSB took place at CERN on Tuesday 4 February 1999 . The minutes are not openly avalaible.

• people attending the meeting were developers (but D.Jacobs/CERN, myself/LHCb) : after few official statements on behalf of Universities or Laboratories, it was a Geant4 workshop.
• next release end of May: the target is consolidation, bug fixes and time performance improvement.
  • update PhysicalConstants and SystemOfUnits to the corresponding CLHEP ones.
  • use the CLHEP RandGeneral.
  • migrate to STL.
• documention is avalaible on the web.
• introduction of granular libraries: a solution for the next TSB meeting end of March.
• hadronic packages
  • the current implementation is an implementation "a la GHEISHA" entirely rewritten by Vancouver.
  • another implementation is being worked out starting from theoritical models. This new implementation should be avalaible in the end of July release.
• cpu time consumption
  • the problem occurs when tracking real events ==> when you mix physics process and tracking; tracking of geantino is going as fast as in Geant3: the Tracking and Electomagnetic working groups should study the problem.
• huge data files
  • the huge data files are mandatory when you want to track low energy neutrons.
• SPIDER: the TSB agreed that Geant4 will watch the developments of the SPIDER-SRT group.
• Both CVS check in and out are restricted to Collaboration members.

 

2. Is it possible to split SICB in small packages?

Florence presented a way to split SICB in small packages: one sub-directory will be a package.(ppt)

==> remarks:

• cyclic dependencies must be avoided ==> some include files must be moved from one package to a lower one.
• working with 35 packages will be very difficult with the existing Makefile avalaible in the Task package.
• So the use of a Configuration Management Tool is recommended.
• SICB version 115 has been used to check the feasability of the split and to check CMT the tool written by C.Arnault.
• The same tool could be used in GAUDI and SICB.

==> decisions:

• Florence will install SICB v116 under CMT and provide the necessary documentation on the Web.

3. A.O.B

• SICB should run on OSF1
• installation of the new version on NT should be done as soon as possible after the release. Is it possible to automate it.
• next SW weeks (J.Harvey)
  • reduce the SW week to 3 days: June2-4, Sept1-3, Nov24-26.
• Software planning (J.Harvey)
  • meet with subdetector convenors and SW contact people to discuss programme of work in1999.
  • establish the goals and timescales for producing them.
  • explain what features will be supported by the framework and by when.
  • to understand plans of the subdetector groups  and the resources,
  • to see what additional help can be provided in terms of training, support, etc..

Next Offline meeting Tuesday 2 March 1999