Difference: CoreSoftwareMeetingMinutes20080521 (1 vs. 2)

Revision 22008-05-22 - MarcoClemencic

Line: 1 to 1
 
META TOPICPARENT name="CoreSoftwareMeetingMinutes"

LHCb Core Software Meeting

Line: 23 to 23
 
  • We will have a build of the AA software stack on Win32 with VisualC9, linking against VC71 externals (preliminarily tested by Stefan and seems to work).
  • GCC 4.3 not yet available because of problems with GCCXML 0.9. Under investigation.
  • Decision about HepMC2: as soon as ROOT provides the plug-in to allow the transparent migration (should be in ROOT 5.18 patches), we can try the migration and then move to HepMC2.
Changed:
<
<
  • Atlas don't like the idea to move to the de-SEALed stack before data taking
>
>
  • Atlas don't like the idea to move to the de-SEALed stack before data taking, so the AA will have to maintain 2 versions of the stack (with and without SEAL).
  • OSX 10.4 officially dropped. AA will only support OSX 10.5. (Currently there are problems in the nightlies)

Philippe:

  • What about CASTOR on exotic platforms (OSX, Win32)? Still not available.
  • Gaudi on OSX? It has not been released recently because it requires too much effort (the scripts for the release have problems). Marco Ca. pointed out that the setup of the environment is not automatic (or simple). Hubert's project LbScript should help for that (currently under test).
 

Software Releases

Gaudi (Marco Cl.)

Added:
>
>
  • After the request by Beat, Marco Cl. started to work on the extension of the state machine of Gaudi. After 3 days almost full-time work on it, there is a (private) version of Gaudi that compiles and that can be successfully tested.
    • The main problem for the implementation was that Gaudi didn't have a "global" state machine, but only few components with states (mainly the services).
    • The "restart" and "reinitialize" functions are not implemented (and probably reinitialize is more broken than before).
    • Due to the dynamic nature of Gaudi, it is not enough to remember the state you are in, but the state where you are going to be (if a service is created during the initialize, it has to be created and brought immediately to the INITIALIZED state).
    • The work will have to be redone (almost from scratch) in Gaudi v20 (the current one is a sort of a hack).
    • The changes are not complete, but the test suite is not complete either, so it is difficult to spot mistakes. Marco Ca. will build LHCb/Boole/Brunel and test them to see if there are problems.
  • Marco Cl. is going to review the patch by Eric. It is needed by LHCb, but it seem to be too intrusive for Atlas.
 

LHCb, Boole, Brunel (Marco Ca.)

Added:
>
>
  • New release almost built.
    • It features the long awaited new version of IT and TT structure.
    • The OT final raw decoding is still missing because it breaks Boole. As soon as it is available (Jan Amoraal is working on it), we will have a new release.
  • Boole and Brunel will follow shortly, then we have to produce some events to test the new code and geometry.

Marco Cl.: a new version of SQLDDDB is available, with a consistent 2008 (head) tag. Information about the tags can be found in a preliminary web page containing the release notes of the DB.

Gloria asked about the conditions for the magnetic filed service. Marco Cl. was a bit stuck with other things and it has been delayed. (In principle should not be Marco Cl. to do it, but in this case we make an exception because it is the very first Online condition to be used in production)

Marco Ca.: Preparing an LHCbApplication configurable. We need to add to it shortcuts for the set of tags to be used.

We had a longish discussion about the conditions database, HLT and first pass reconstruction. The outcome is (more or less):

  • the "offline" conditions can be used in HLT only after validation and tag
  • we need the foreseen implementation of the machinery to push conditions to the HLT (for Velo stepping-motors position)
 

Gauss (Gloria)

Added:
>
>
  • Releasing a development version
    • 2 python steering options.
  • Compiling with the latest version of LHCb to test the new geometry.
  • DC06 physics list will not be changed anymore.
  • Preparing v26r1 to use the new magnetic field map (not easy).
  • Gauss meeting on Friday.
 

HLT (Patrick)

Added:
>
>
  • Not yet compiling. No way to get HLT released before middle of next week (if the new LHCb is coming soon enough, we can even skip the current version).
 

DaVinci (Juan)

Added:
>
>
  • There has been a problem with some code that was not in committed, but the compilation worked in the private version because the compiled function was still in the .a file. The temporary solution has been to put the code in DaVinciKernel, but should go immediately in LHCbKernel CVS to be ready for next release.
 
Changed:
<
<

OnLine (Markus)

Panoramix (Thomas)

Release and Deployment (Hubert)

>
>
Philippe: news about the crashes occurring in CCRC?
The problem is due to an event with 4 primary vertexes than makes the combinatoric to diverge. The patch applied currently is a change in options to reject events with more then 2 vertexes.
 

Round Table

Philippe
Added:
>
>
    • Physics people is asking for some minimum bias events reconstructed without trigger, but they didn't say which version of Brunel to use. Thomas will check.
    • Oracle-based CondDB is being successfully used in CCRC. Even the fall-back works (there was a problem with the PIC connection details that made all jobs at PIC go to a different Tier-1 for the conditions).
    • We could dedicate the next software week to physics software (we can still have a parallel core software session).
    • New technical training courses are being prepared. The aim is to give basics about Object Oriented Programming, C++ and Python.
 

Revision 12008-05-22 - MarcoClemencic

Line: 1 to 1
Added:
>
>
META TOPICPARENT name="CoreSoftwareMeetingMinutes"

LHCb Core Software Meeting

Date and Location

May, 21 2008
10:30 - 12:45
CERN (2-R-030)

Attendees

Gloria, Hubert, Marco Ca., Marco Cl. (minutes), Patrick (phone), Philippe, Thomas

Apologies

Markus

Subjects

News

From Architects Forum (Marco Ca.):
  • We will have a build of the AA software stack on Win32 with VisualC9, linking against VC71 externals (preliminarily tested by Stefan and seems to work).
  • GCC 4.3 not yet available because of problems with GCCXML 0.9. Under investigation.
  • Decision about HepMC2: as soon as ROOT provides the plug-in to allow the transparent migration (should be in ROOT 5.18 patches), we can try the migration and then move to HepMC2.
  • Atlas don't like the idea to move to the de-SEALed stack before data taking

Software Releases

Gaudi (Marco Cl.)

LHCb, Boole, Brunel (Marco Ca.)

Gauss (Gloria)

HLT (Patrick)

DaVinci (Juan)

OnLine (Markus)

Panoramix (Thomas)

Release and Deployment (Hubert)

Round Table

Philippe

-- MarcoClemencic - 22 May 2008

 
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