LHCb Core Software Meeting

Date and Location

18. October 2006
10:00 - 10:45
CERN (2-R-030)

Attendees

Florence, Gloria, Hubert, Marco Ca., Marco Cl. (minutes), Markus, Matt, Stefan, Thomas

Apologies

Juan

Notes

I have to apologize to the attendees trying to call the phone conference. For a reason that I do not know yet, the system was not letting me in.

Subjects

News

• LCG 47b delayed because of a bug in ROOT. The build of a fixed version should be ready during the morning. COOL and POOL will re-run the tests.

Software Releases

Gaudi (Hubert)

• v19 still on hold
• v18r8 ready, just waiting for LCG 47b
  • Changes in the properties of DataOnDemandSvc (Vanya)
• Few problems were encountered on AMD64
  • lxplus machines do not have enough memory to compile the Parser
  • problems with random seds
  • problems with ROOT

LHCb, Boole, Brunel (Marco Ca.)

• Stuck after the discovery of the problem with Reflex
• Next release (Brunel) will be early next week (with new Gaudi and new LCG if they are ready in time).
• The following release will be at the beginning of November.

Gauss (Gloria)

• v25r4 released last week and used in production.
  • 4-5 failure for time-out from IN2P3 (Andrei is aware of the problem)
• A build of v25r5 (~ identical to v25r4) with the new version of GENSER is available in DEV for testing.

DaVinci (Marco Cl. for Juan)

• Test jobs with pre-selections exposed variuos problems (fixed by Juan and Patrick)
• DaVinciAssociators revers association have problems. Olivier found a possible cause, but the solution requires big changes (not ready yet).
• Problems with calorimeter association: require latest versions of Calo/CaloReco and Calo/CaloUtils.
• No Brunel v30r10 in the book-keeping. He will ask for a small production (~10K events) to test the pre-selection.
• Matt committed changes in DetDesc to speed up the transport service (few functions made inline).

Panoramix (Thomas for Guy)

• Pending issues on dictionaries and ROOT.

Round Table

Gloria

It happened twice in one week that files ready to be migrated wait for days.

Thomas

We should probably do some clean up in CASTOR. Marco Ca. reminded that we do not have anybody for that task.

Markus

There are efforts in ROOT to be able store vectors of pointers as branches of a tree (as it happens with TClonesArray). If we move to it, we will not be able to read back old data. Anyway it is not going to happen soon.

Marco Cl. and Florence

After the iscussion of last meeting concerning the possibility of overriding the version of LCG externals like CASTOR, the are implementing a solution based on SetupProject and CMTUSERCONTEXT, with changes in install_project.py to be able to disptribute tarballs with the new versions of the needed libraries. The system is half ready: if you use SetupProject at CERN for an application based on LCG 46, you will use CASTOR 2.1.1 (LCG 47) instead of 2.1.0.

---

-- MarcoClemencic - 19 Oct 2006