LHCb Core Software Meeting

Date and Location

20-Jul-2011

Attendees

Subjects

News

• The Core Software Meetings are now in a dedicated category in indico (thus exported in a dedicated iCal file).
  • Already created the meetings until the end of the year, with correct details for the EVO meetings.
• Starting from today, the LCG nightlies feature gcc 4.5.3 builds (x86_64-slc5-gcc45-dbg)
  • to be added to our nightlies when stable
  • Hubert instrumented LbScripts to understand the new platform (from next version)
• Password leaked in a Subversion commit
  • the password for the affected account has been changed
  • there is an "official" procedure to clean the repository, but we never used it so far because we need to test it first
    • no hurry for the massaging of the repository (anyway the old password cannot be used anymore)
    • test the procedure with a small repository (e.g. the unused lbdirac, but we have to wait for Joel)
    • try the first part of the procedure (dump + filter) to estimate how long we need to keep the repository offline (Hubert)
• Atlas started discussions on a possible "CMT2"
  • Marco Cl. and Hubert will follow the presentation during an Atlas meeting tomorrow
  • Marco Ca.: it should be made clear that LHCb requires the continuation of the support of CMT1 and is considering the adoption of CMake.

Round Table

Marco Cl.

• Working to apply Gaudi patches (clean up a bit the savannah backlog)
  • the release of LHCb (and the rest of the stack) is starting early next week, and will not use/need a new Gaudi
  • aiming for the next release around the middle of August (using the latest stable version of root, 5.30/00)
• Preparing a shared Eclipse installation based on the new major version Indigo (3.7)
  • It will include the latest versions of Greg's plugins (now working on the migration of the Gaudi basic tutorials)
• Working to write a tool for the DQ Filtering to be paired with the work by Rob, Jaap and Marco Adinolfi
  • given an IoV, it must return a boolean (good or bad)

Martin

• Writing the tests for the environment configuration parser.
• Next step is to define how the environment configuration has to be used (tools).

Hubert

• LbScripts v6r3 in production (no problems reported so far).
• Preparing the next version (minor changes).
• Preparing a new version of LHCbGrid to distribute the latest version of tcmalloc (to be used on the grid).
• Changing the way of getting the up-to-date list on known projects and packages
  • from a python module (requiring a new release of LbScripts for each new entry) to a couple of http-distributed XML files with local cache automatically handled by install_project.py.

Marco Ca.

• Next week will bee a "releasing week": the full stack to be ready for the Stripping.
• A new release shifter is going to join the team.

Illya

• Moving on with the implementation of the tools for the new way of deploying the SQLite (CondDB) databases.

Marco Ca.: there has been a question from the Tier-1s about our plans for Frontier and oracle Streams (see thread).

Rob

• Prepared a script to parse CMT requirements directly on SVN
  • integrating it in LbScripts.
• Trying to detect MDF files without opening them with Gaudi (i.e. from some signature)
  • will ask Markus for hints.
• We want to use tcmalloc for the stripping
  • the working version will be distributed with the LHCbGrid that Hubert is preparing and Rob will test it as soon as available on AFS.
• Working on the optimization of the code, starting from the protection of the debug print-outs, using the UNLIKELY macro too.
  • Marco Ca.: there is a test in Brunel which uses the new detection of un-protected debug messages (initialization only); it should be a good starting point.
  • Marco Cl.: to simplify the protection and the usage of the UNLIKELY macro, we need a macro like ON_DEBUG (to be introduced before changing too much code in LHCb), e.g.:

ON_DEBUG {
  log << MSG::DEBUG << "Hello, world." << endmsg;
}

-- MarcoClemencic - 20-Jul-2011