LHCb Core Software Meeting

Date and Location

14-Oct-2009
10:30 - 12:15
CERN (2-R-030)

Attendees

Gloria, Hubert, Illya, Juan, Karol, Marco Ca., Marco Cl. (minutes), Markus, Patrick (EVO), Rob, Silvia, Thomas

Subjects

News

• LCG 57 being released

Software Releases

Gaudi (Marco Cl.)

• Applioed some pending patches
  • Question about patch by Vanya, to use histograms as properties: won't apply
• Should have a look at the pending bugs
  • limited number of output streams
  • Markus: we must get rid of the messages coming from the configurables about changing the type of a configurable (a bug report will be posted on savannah)

LHCb, Boole, Brunel (Marco Ca.)

• Start tagging this week
• Changes by Wouter in track event
  • they are not backward compatible, but needed and not easy to roll-back

Gauss (Gloria)

• Additional generator for MC09
• Compilation problems on SLC5 (mainly FORTRAN)
• Given feedback tyo G4 user forum
• LCG GENSER meeting at the end fo November

HLT (Patrick)

• Debigging a problem with CALO (probably an uninitialized variable)

DaVinci (Juan)

• Will make a release ASAP after LHCb

Online (Markus)

• The online cluster will migrate to SLC5 next week

Round Table

Hubert

• Distrubution kits for Ganga (SLC4 only) and Dirac are ready
• Fixed manpages problem (in the online environment)
• LbScripts v4r3 will be soon in production

Patrick

An offline discussion about Panioramix lead to the question about why we do not support OSX, but we support Windows. The point is that the infrastructure for Windows is in place and we do not have yet another (non-GCC) compiler to validate the code. OSX cannot be supported at the moment because Windows absorbs too much man-power.

Silvia

Switched off the writing in the tests

Marco Ca.

We need to test the workflow in the nightlies. Markus reported a possible memory leak in Brunel.

Karol

• Restarted the broken nightlie
• Testing an algorithm to start/stop profiling

Juan

Would be nice to have links in the project pages that point to the used projects (with therir versions)

Rob

How do we ditribute the XML summary service and the Python modules? LHCb is too big to have Ganga depend on it. We can put the Python (platform-independent) part in DBASE (for example), so that it can be distributed as a stand-alone package, while the C++ part (not needed by Ganga/DIRAC) can stay in LHCb. This gives us some flexibility in the evolution of the Python code too. To be done after the next release.

Marco Cl.

A problem occurred in the propagation of Online conditions during the last TED: they didn't go to the database. Velo position will be fixed in LHCBCOND, magnet current to be checked.

-- MarcoClemencic - 21-Oct-2009