LHCb Core Software Meeting

Date and Location

01-Jun-2011
10:30 - 11:50
CERN (2-R-030)

Attendees

Gloria Corti, Hubert Degaudenzi, Illya Shapoval, Joel Closier, Marco Cattaneo, Marco Clemencic (minutes), Markus Frank, Rob Lambert, Thomas Hartmann

Subjects

News

  • The frequency of the Architects' Forum changed from every two weeks to roughly once a month (the second Thursday of the month).
  • SLC6 public login service available for testing (lxplus6)

There is a technical problem for the eventual migration to SLC6: the version of the C++ compiler we use for SLC5 (gcc 4.3) cannot work on SLC6 (because the native compiler is gcc 4.4). This implies that on SLC6 we can run SLC5 applications, but we cannot them (as we used to do for SLC4 on SLC5). The long term solution is to change the compiler we use on SLC5 to gcc >= 4.4: gcc 4.4 will be easier to handle on SLC6, but the currently tested 4.5 has better support for automatic vectorization (faster code).

Round Table

Marco Cl.

  • Working on the cmake configuration for Gaudi.
    • Next steps require the tool to set up the environment (summer student project)
  • Starting to apply the pending patches in Gaudi (including Sacha's patch for the new implementation of the option parser).
  • Prepared a new version of the shared Eclipse installation:
    • usual updates of plugins
    • added Gregoire's plugin for getpack integration (requires LbScripts v6r2)
    • 'prod' link to be changes when LbScripts v6r2 will be in production

Joel

  • Please report ASAP problems on SLC6 (lxplus6.cern.ch).

Hubert

  • Development version of LbScripts available (v6r2)
    • improvements to getpack (for the Eclipse plugin)
    • fix to SetupProject (change in the usage of packages to avoid picking up the wrong AppConfig)
    • automatic doxygen documentation missing (problems?), BTW, the Python scripts are not documented
  • Working to get an error if the generation of the tarball fails.

Rob

  • Working with Marco Ca. for the adoption of the new persistency format
    • some changes are required in Gaudi (to simplify the usage)
  • Prepared scripts to
    • check if the application supports the new persistency format (for Dirac)
    • convert old style data cards in the new style

Question about the possibility to write POOL files with the new versions of the software. The answer is no: if we need to process a new file with an old version of the software, we produce a patched version that uses the new persistency.

Illya

  • Prepared the changes to use tag-based slices of the CondDB partitions instead of the full DB files
    • the problems is that the duplications increase (the size of 2 sliced of DDDB are as big as the full DDDB, etc.), so it is still better to distribute full DBs
  • To solve the problem of duplicated full copied of DDDB and to have a quicker distribution cycle, we can exploit CVMFS: instead of packaging new versions of SQLDDDB, just update the SQLite files in CVMFS, so that they are automatically picked up by the clients when they change.
    • the technical details have to be defined

Markus

  • Problem with the checkpointing after the upgrade of the kernel version on lxbuild

-- MarcoClemencic - 08-Jun-2011

Edit | Attach | Watch | Print version | History: r3 < r2 < r1 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r3 - 2011-06-08 - 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