LHCb Core Software Meeting

Date and Location

21 February 2007
10:30 - 11:50
CERN (2-R-030)

Attendees

Florence, Gloria, Hubert, Marco Ca., Marco Cl. (minutes), Markus, Olivier, Philippe

Apologies

Andres

Subjects

News

Software Releases

Gaudi (Hubert)

  • v19r2 (delayed by Atlas requests)
    • changes in gaudimodule.py
    • IUpdateManagerSvc (by Marco Cl.: to allow access to some U.M.S. functions from Python)
    • lot of problems with configurables [long discussion]
      • since the make fragment is trying to load the component libraries to discover the options, it needs the ApplicationMgr and ServiceMgr, forcing a dependency on GaudiSvc in all the component packages: not acceptable for LHCb. We will investigate further how to make both Atlas and LHCb happy.
  • Port to AMD64
    • Vanya fixed the problem with the seeds:
      • changed vector into vector
        Gloria asks what to do with the vector of integers we have in the event headers to record the seeds. We can keep the vector as it is, but it has to be copied to/from a vector of long when needed.
      • use Ranlux64 (same behavior on 32 and 64 bit machines, but different from the old Ranlux)
        The fact that the sequence of random numbers changes from previous versions of Gaudi will be advertised as soon as this changes are included in the release.
    • Problems observed with the POOL IO Coll. examples: strange results in both writing and reading on AMD64.

Marco Ca.: we have two copies of the RootMapDump script, one in LHCb the other in Gaudi with small differences. We agreed to take to best one and keep it in Gaudi.

LHCb, Boole, Brunel (Marco Ca.)

  • v22r2 available in DEV (both optimized and debug versions)
    • only few changes in LokiGen are missing
    • the main goal of the release is the deployment of SQLDDDB (Marco Cl. will prepare the changes needed to do the step).
  • Dictionaries for interfaces moved from TrackPython to the correct packages.
  • v21r0: moved GaudiMath to GaudiKernel, added headers for backward compatibility. Components moved to PHYS.

Florence: ApplicationMgr.DLLs is not needed anymore.

Gauss (Gloria)

  • Preparing v30, based on LHCb v22r1
    • Changes in XML because of modifications in XmlTools
    • Rich namespace
    • Porting changes from v25 to v30

Round Table

Gloria
somebody asked to have a tool to clone the particle in the transient store to be used to generate MicroDSTs. It is not extremely clear who is going to provide such a tool, but it's clear that it has to be done with care and with input from the physics group (to know where to stop the recursion).
Marco Ca.
Reimplemented Patrick's script to discover used projects.
Olivier
Seems that on AMD64 std::vector::operator[](size_t) is much slower when using numeric constants.

-- MarcoClemencic - 21 Mar 2007

Edit | Attach | Watch | Print version | History: r2 < r1 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r2 - 2007-03-21 - MarcoClemencic
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    LHCb All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback