(r1) CoreSoftwareMeetingMinutes20070822 < LHCb

LHCb Web>LHCbComputing>CoreSoftwareMeetingMinutes>CoreSoftwareMeetingMinutes20070822 (revision 1) (raw view)~~EditAttachPDF~~

<noautolink>
---+!! LHCb Core Software Meeting
---++!! Date and Location
22 August 2007 %BR%
10:30 - 12:30 %BR%
CERN (2-R-030) %BR%
---++!! Attendees
Gloria,
Hubert,
Marco Cl. (minutes),
Markus,
Patrick (EVO),
Philippe,
Thomas
<!--
---++!! Apologies
-->
---++!! Subjects
%TOC%
---+ Software Releases
---++ Gaudi (Hubert, Marco Cl.)
   * Python modules/scripts installation almost completed
      * By default the <package>/python and <package>/scripts will installed respectively in InstallArea/python and InstallArea/scripts
      * The recommended policy for the python modules is to add a subdirectory called <package> under python. The import of the modules will have to be done using "from <package> import <module>" or something equivalent (e.g. "from GaudiPython import gaudimodule").
      * The migration will be sort of adiabatic. E.g.: together with "GaudiPython.gaudimodule" a module called "gaudimodule" will be provided to print an obsolescence warning, but allow old python code to run (the behavior can be turned to failure in any moment).
      * Still to test on Win32 because AFS Service not always working on CERNTSDEV01.%BR%
      Should contact NICE support to have a newer version installed.

---++ Gauss (Gloria)
   * Problem in the production in a new generator
      * The generator (FORTRAN) needs to open a file, but the string for the file name was long 100 chars (a bit too short for the GRID). The problem has been solved setting the size to 512.
   * Preparing a new DC06 version:
      * Fix for the problem just described.
      * Fixed discussed last week (empty events).
      * A new generator (for Vanya).
      * New features in the particle gun.
   * A post-DC06 version will follow.
   * The summer student is looking at the timing of Gauss with a special eye to the effect of the cuts at generation level.
      * It is not clear how to interpret the time spent because the scaling factor returned by the ChronoStatSvc and the one provided by LSF do not agree.%BR%
      It is probably better to use always the same machine and not LSF.

---++ OnLine (Markus)
   * Tried to prepare a new release.
      * Rich people prefer to have their monitoring tools in Online instead of Orwell, but they will not be part of the release until they build successfully on Win32.

---++ Release and Deployment (Hubert)
   * Prepared the php scripts to handle archival of projects versions. The python script to do the archival are being prepared.
   * Tomasz asked for a new release of Vetra.
---+ Round Table
   $ *Gloria*: 
      * Few changes should be done in Velo geometry (DDDB). Will need a new DDDB tag and possibly a production.
      * We need a way to be able to do simulation with a special detector description. The new informations will have to go in the SIMCOND database, which Marco Cl. will start to prepare as soon as back from CHEP07+LHCbWeek.
   $ *Thomas*: Recently notices that in some code we use the function pow(int,int) which returns double on Linux and int on Windows. The specific problem has been avoided by removing the call to pow (not actually needed). For the future, we should avoid that function or use it converting the arguments to double before passing them to the function.
   $ *Patrick*: Problem with ParticleID and vertex fitting. There is a constructor of Particle taht acceps and int, which is used by mistake when trying to create a particle from the particle id, while it is needed by the ParticleMaker to give to the Particle (a KeyedObject) the same key of the MCParticle from which it was generated.%BR%
   We agreed to remove that constructor and add a setter for the object id.
   $ *Markus*: Beat proposed to dedicate a Core-Software meeting to a discussion about adding a couple of states to Gaudi (something more that "initialize", "run", "finalize") and to the possibility of having more processes on one machine sharing the memory for services like IMagneticFieldSvc or the DetectorDataSvc.
   $ *Philippe*: The reconstruction has been restarted without the minimum bias events and it is running smoothly. The problem with the bug in Boole v12r1 and L0 decision, does not affect the reconstruction, so the affected sample is being reconstructed too (with a different production number for safety). If the fixed L0 decision is needed, we can fix it during the stripping.

----
</noautolink>

-- Main.MarcoClemencic - 22 Aug 2007

Topic revision: r1 - 2007-08-22 - MarcoClemencic

LHCb

Public webs

Welcome Guest

- Cern Search
- TWiki Search
- Google Search
LHCb All webs

Copyright &© 2008-2021 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
or Ideas, requests, problems regarding TWiki? use Discourse or Send feedback