LHCb Core Software Meeting

Date and Location

11. October 2006
10:00 - 12:00
CERN (2-R-030)

Attendees

Florence, Gloria, Hubert, Joel, Juan, Marco Ca., Marco Cl. (minutes), Markus, Matt, Nick (phone), Olivier, Patrick, Philippe, Thomas

Apologies

Pere

Subjects

News

• LCG 47 officially announced
  • ROOT 5.13.02
  • CASTOR 2.1.1-1
  • etc...
• (Joel) Reconstruction of DC06 restarted (good DSTs at CERN and RAL)

Software Releases

Gaudi (Hubert)

• Markus noticed a performance problem in the StatusCode improvements by Atlas: v19r0 on hold.
• Branch for v18 to release v18r8
  • based on the new version of CMT (cannot be built with the old one)
    • Florence will change the login script to use the new version of CMT and send an e-mail to LHCb
  • Marco Cl. added the function System::setEnv and a new option to ApplicationMgr to set environment variables from job options (needed to work around limitations of DIRAC/Ganga)
  • problems with Windows build because of externals

Marco Cl. proposed to change the behavior of System::getEnv to return an empty string if the environment variable does not exist. Now it returns "UNKNOWN". The problem is that on Windows it is not possible to set a variable to "", because it is removed from the environment, so, currently, we can set a variable to "" and get back "UNKNOWN".

LHCb, Boole, Brunel (Marco Ca.)

• Brunel v30r10 released
• Waiting for Gaudi

Gauss (Gloria)

• New release for DC06 (cuts at the generator level)
• Fixed a problem with an uninitialized variable

DaVinci (Juan)

• DaVinci v17r2 released: more peopole is using it (more feedback)

Panoramix (Thomas for Guy)

• Released a version with bug-fixes by Thomas. Guy proposes a different way of fixing the bugs.

Round Table

Florence

Fixed a problem with the pattern to set PATH and LD_LIBRARY_PATH in LCGCMT.

Olivier

Added 2 new properties to algorithms (like the context one) to handle Time Alignment Events.

Patrick

Adding a new package DVMCKernel to group Montecarlo related classes.

Philippe

The new version of the CASTOR client/server introduced a problem with the way we deal with the versions of libraries needed by plug-ins. We should be able to use a version of CASTOR (gfal, dcache, etc.) different from the one that was used for the release. It should be useful to do it in a generic way (should work on the GRID and at CERN).

Marco Ca.

We soon will move to SLC4. He asked for which linux platform build the debug versions.
It has been agreed to stop building debug versions for SLC3 and start to build them for SLC4.

---

-- MarcoClemencic - 16 Oct 2006