LHCb Core Software Meeting

Date and Location

8 August 2007
10:40 - 12:30
CERN (2-R-030)


Gloria, Hubert, Juan, Marco Cl. (minutes), Markus, Patrick (EVO), Philippe, Thomas


Software Releases

Gaudi (Hubert, Marco Cl.)

  • Hubert
    • Completed the fragments and patterns to install Python modules and scripts. Now working on installation of configurables.
    • Working to produce relative symbolic links in the install area instead of the current absolute links.
  • Marco Cl.
    • Applied a couple of patches by Vanya to allow the usage of AIDA:IProfile[1D,2D] in GaudiPython (the feature was available in Atlas private code), and to allow access to histograms and n-tuples from outside of algorithms and tools.

Gauss (Gloria)

  • Cambridge people have been active to the port to Win32 of Gauss, they managed to use a particle gun. The required changes have been sent by mail.
  • A version of Gauss that can use alignments is available in DEV.
    • Tested with Geant4 OpenGL visualization tool (DAWN was not built for SLC4).

The organization and requirements files of Geant4 visualization tools need to be reviewed. We have a lot of ad-hoc solution and missing libraries.

Gloria and Thomas decided to create a new package in LHCb to host the XML files for the configuration of the visualization parameters, which are still in XmlDDDB.

DaVinci (Juan)

Patrick: Patrick Spradlin is reimplementing MCEFBuilder and DaVinciEff that reverse-associates proto-particles to truth. He was asking which is the correct package for it. We agreed to have it in Analysis, and move it somewhere else only if it is ever needed in Brunel.

HLT (Patrick)

  • Development going on in DEV. Waiting for Marco Ca. for a proper release.

OnLine (Markus)

  • Some success on the serialization of DataObjects, to be able to send them over the network for monitorin.

Panoramix (Thomas)

  • New release available.
  • Problems with the dictionaries:
    • few duplicated dictionaries in SoDet, but they cannot be removed otherwise other dictionaries are missing
    • missing dictionaries for classes correlated to event model classes
      Marco Cl. observed that the problem was about a class returned by a method of a G.O.D. generated class. One possibility is to extend G.O.D. XML description to allow addition to the Reflex selection file, but is is felt as unmaintainable. The best solution seems to add the missing classes at the project level.

Release and Deployment (Hubert)

  • Released the full stack.
    • Problems with Panoramix on AMD64, genmap complains about unresolved symbols.
  • Prepared the md5sums of all the tarballs in our distribution area, install_project.py will soon be able to check if a download was successful or not. If not, the installation directories should be left clean (so that a second call to install_project.py will be able to try again).

Round Table

Marco Cl.
  • Completed the fix to be able again to clear the Detector Transient Store and reload the XML files. It will need a new Gaudi for a proper implementation in Panoramix. Currently can be used only with a recipe (and few patches) to compile some packages of Panoramix and LHCb and it can only reload from files (the CondDB cache is not cleaned).
  • Fixed a bug in SetupProject that was preventing the usage of the user release area if called after setenvProject.csh (setenvDaVinci, etc.)
  • Currently it is possible to generate the Doxygen documentation only with the release scripts, not by hand. The proposal is to extract from "Project"Sys the documentation parts, and put them in a new package called "Project"Doc. That will avoid unneeded dependencies and clean up the procedure to build the documentation, with the side effect of having to maintain another package which version should be kept in sync with the project version.%BR$ We agreed that it should be done.
  • After a long discussion with Marco Cl., it seems useful to split the LCG external tarball in one tarball per external package.
    (The discussion has been long...) It is not completely clear which are the advantages and disadvantages of such a solution. both in terms of download time and code maintenance. We agreed to have an implementation of install_project.py in which is easy to chose any of the two approaches (one big tarball or many small ones), so we can test and decide.
Progress in the understanding of DaVinci crashes?
(Markus) The situation is that it is still impossible to reproduce the problem and in this case we can only try code inspection.
Marco Cl. proposed to try to re-run DaVinci in the same job if it failed at the first attempt. Markus proposed to run it twice always and produce two ETC files. Philippe will discuss with Joel to produce the needed work-flow.
Recently reported a bug to Gaudi savannah. The problem is that when an option file is included twice, the second time it is ignored without a warning. It seems reasonable that in the context of option files the inclusions are always processed, since to override a setting you have to append the new setting to the file. Unfortunately, this behavior will probably create a lot of problems in DaVinci because all the pre-selection options are including DaVinciCommon.opts. The proposed solution is to keep the current behaviour but issue a warning.

-- MarcoClemencic - 08 Aug 2007

Edit | Attach | Watch | Print version | History: r2 < r1 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r2 - 2007-08-08 - unknown
    • 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 & 2008-2021 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