Present : Dieter Schlatter, Giuseppe Bagliesi, Andrea Valassi, Andreas Pfeiffer,
          Dino Ferrero Merlino, Arno Heister, Guenther Dissertori


 

 
 Andreas Pfeiffer and Dino Ferrero Merlino have given a short 
      report about the next plans with the LHC++ project. At the moment they are
      working on the definition of the interfaces for analysis tools, called AIDA.
      The plan is to set up something like a modular C++ version of PAW, with very
      general interfaces and some command line interface. A first prototype is
      expected for the beginning of February. They would welcome comments and
      suggestions by the users, also from us. The IRIS Explorer analysis package
      will not be developed further, but nevertheless there will be some support.

      They are also in contact with the JAS developers, and for the end of February
      a JAS interface to Tags is expected.


 
 Guenther gave a report about his efforts to do a ALPHA++-port
      to Linux. First he prepared a list of MINI banks by writing a small application
      which generates a list of all banks found on the MINIs, since his goal is to
      write now all banks to the database. He run his application over
      50k LEP1 data events, 50k LEP1 MC events, 50k LEP2 98-data events and 50k
      events of a KORLW02 production (note that the different  sets can have small
      differences in banks encountered). In the end he generated a list of 125 banks.
      
      On the software side (first on the population code) the following change is
      applied: cfortran.h is used (a platform-independent definition of calls 
      to FORTRAN routines from C++).

      Next he has generated a new database with the MINI bank list, which means that
      he generated a new federated database and compiled the ddls and the write application
      on BOTH RedHat 5.1 and OSF1 !

      The results are the following:
      The executables work, and the population of the database has been tested, by writing
      from his PC to the database which resides on a shift50 disk.

      10k LEP1 data (clas 16) events :                 size   141 Mb (no ootidy)
      --------------------------------                 time   real= 14 min
                                                              user=148 sec
                                                              sys =  8 sec  (19%)
                           this corresponds to 38 runs, 10k events, 241k banks.
                           so he estimates that for 4M events we need 57 Gb and about 4 days.
							       
       10k LEP1 MC (qqbar) events :                    size   217 Mb (no ootidy)
      -----------------------------                    time   real= 13 min
                                                              user=185 sec
                                                              sys =  7 sec
                           this corresponds to 1 run, 10k events, 327k banks.
                           so he estimates that for 8M events we need 176 Gb and about 8 days.     

      He points out that the executables compiled on RH5.1 also run on RH6.0! This is
      important for the production in Munich which runs RH6.0, so eventually we simply
      copy the executables and shared libs from his PC (5.1) to Munich.
			   
      The test of the read application showed: Earlier problems (such as missing banks,
      LEP energy and MC banks) are gone!!

      Problems encountered:
      ---------------------
      First there were some problems with the shared libs on Linux (an ASIS problem).
      On OSF1 a problem with Ospace STL and HepODBMS was found, and Dirk had to
      install a patch. Note that it is still possible to run on the old database, on OSF1.

      For some banks it was found that there is no description or a wrong discription
      around (X1TT:rows and columns inverted, ARUN: 15 instead of 10 columns, ...)

      One problem is still there: when reading a MC database with the full analysis code,
      an objy-problem occurs after 3500 events. Dirk has to help here.

      Plans:
      ======
      First he wants to implement cfortran also on the analysis package for Linux, which
      needs some coding. Then some further tests on the database should be performed, and
      then he proposed to go to Munich: 17.-21.1.2000. Project:
          -- install Objy according to latest certifications
	  -- install ALPHA++
	  -- perform tests
	  -- start the big production

      In the meanwhile he invites the people to study on upgrades to the application
      code, Tags and further LEP2 analysis requirements.