Minutes of the Computing meeting held on 26 May 1999

Present: P.Mato,  M.Cattaneo, M.Frank, P.Maley, P.Colrain, G.Corti, E.Van Herwijnen, I.Videau, E.Tcherniaev, 
              N.Smirnov, P.Binko, M.Needham, G.Gracia, J.Closier, T.Cass, F.Ranjard

Apologies:  F.Harris, J.Harvey

Minutes : Florence Ranjard

      1. Pending questions (everybody)

• AFS project space: J.Closier sent a mail to various physics group on 6.1.1999 asking those group to cleanup their
  AFS project space area, in order to reorganize this pace in a better way.
  Only G.Corti for the muon group has done its home work.
  ===> Joel will re-send a mail to responsible people.
  ===> if no answer it will be assumed that the project space is no more needed and could be assigned to another group:
            the read access will be removed.
• gmake on rsplus still does not work.
  ===> Tony will follow the problem.
• still problem with rshell and xterm.
  ===> Tony and Joel will follow the point.
• CERNLIB 99 is not accessible in the normal way on NT, F.Hemmer is aware of this fact.
  ===> Frederic, Joel and Tony will follow the problem.
• RFIO is now working on NT: thanks to Ian McLaren .
• CLHEP : the new version is avalaible on all platforms compiled with egcs-1.1.1.

2. Review code meeting (M.Cattaneo)

Following Paul Colrain's request , Marco will organize a meeting next Friday to review the muon code before
his presentation during the software week.

3 HTL (P.Binko)

Pavel presented the current status of the Histogram Template Library HTL (ppt)

• compact, extensible, modular and performant.
• same signature as HBOOK calls.
• Transient and Persistent histograms are decoupled.
• Interface to PAW exists.

          It was written by a cooperant who has now left CERN with some advices from V.Inoccenti.

• some points are missing, but more than missing points: integration in Gaudi would be difficult  because the
  strategy is different, neither the HTL code nor the HTL library are compatible with Gaudi strategy.
  • changes were made by D.Ferrero Merlino and Pavel in HTL
  • HistogramSvc and IHistogram interface have to be changed.
  • needs to recompile the library.as LHCb is currently using the single-threaded libraries

  The presentation was followed by an intensive discussion:

• Pavel suggested to mask HTL to the end-user in Gaudi.
  • to add missing features,
  • to maintain our strategy

• but it implies that using HTL in the Gaudi Framework and outside would be different:
  • i.e. the same code could not run in test-beam (no Gaudi) and offline (in Gaudi),
    unless test_beam uses Gaudi (as suggested by Pavel)

• if HTL is replaced by another package, the Gaudi interface has to be re-written.

• Tony suggested that potential users (LHCb, ATLAS, CMS, ALICE,...) and IT should agree
  on an interface which fulfills  physicists requirements and this interface should not change when
  the histogram package is replaced or modified.

• It was agreed that LHCb should work with IT to get an HTL package which fulfills our requirements.

======

next meeting:  Wenesday 16 June 1999 9:00-11:00  blg 2/1-034