Reference Software Meeting at CERN
   Monday 18 January 1999, 14:00 in 40-4-C01
 
 Present:  J. Baines, R. Bock, A. Bogaerts, M. Dobson, F. Giacomini, S. Gonzalez, 
           R. Hauser, S. Qian, M.Sessler, T. Shears, S. Tapprogge, P. Werner, T. Wildish
           
 
 phone connection with: ANL (J. Schlereth)
                        RAL (A. Belias, D. Botterill)
                        Saclay (M. Huet)
             

1. Linux kernel repository

The proposal by Marc and Tony that has been circulated to the Ref Sw group
was discussed. Outcome:
- the proposal was well received
- Tony and Marc volunteered to be responsible for the kernel repository
- Marc will also be the Linux contact person of the Ref sw group
- an updated version of the proposal will be circulated
- implementation of the scheme will start

2. SRT revisited

SRT has been considered in the past but was too immature to be used.
The tool has been improved and is currently used by the ATRIG
community and the DAQ/BE. Simon and Reiner are investigating the possibility
of introducing the improved version of SRT in the reference Sw.

3. Progress in Single Farm multi-CPU system

RAL has been working on the application framework and Robin produced some
documentation; CERN has taken this as a starting point and Reiner produced
a note with a few refinements. Action points:
- RAL and CERN need to collaborate to unify the approach
- proposals should be circulated before they are frozen
- this is becoming urgent as Robin will come to CERN next week
  for re-integration

4. Run Control

Postponed

5. Progress in HPCN

Per prepared an action plan for running the communications benchmarks
on a 16 node DG system next weekend. The note will be circulated to 
the Ref sw group.

6. Algorithms

Saul & Tara reported on porting a simplified version of the EM cal
algorithms to the OO reference sw environment. They expect to have a first
version ready by the end of Jan.

SCT: John plans to extend the SCT software to low Pt physics.
     Qian is improving the efficiency of the Kalmann filter based SCT
     algorithms. Many Fortran routines need to be converted into C++
     versions for the reference sw.

data files: physics performance measurements need large data files
            (~ 6 Gbytes) which at the moment are kept at RAL.
            We should look into the possibility of making a local
            copy at CERN using e.g. an NFS exported disk.
            Rudy will contact Tony.

7. A.O.B.

"Data in the Rob for the Pilot project": Michel produced a draft note;
to be discussed at the next meeting.

Rose/StP: there is no decision/recommendation on a unique design tool.
StP is the standard tool for ATLAS off-line. ROSE is gaining more
popularity and is available at CERN; a subset of ROSE is part of the 
WNT C++ Visual Studio. Robin pointed out the problems related to
licence cost for laboratories with a small user community.
 

date of the next meeting: the next meeting on Monday 25 Jan will be
postponed to Friday 29 Jan (to be confirmed by e-mail)
because of a clash with the meeting in Amsterdam. 


 ______________________________________________________________
 |           Next meeting:                                     |
 | Friday 29 Jan 1999 2pm (Geneva time), to be confirmed       |
 | ANL, RAL, UCL and Saclay  please dial +41 22 767 7000       |
 |_____________________________________________________________|