20070917 Tuesday, Core Software Mtg
Attendees: H.Kelly, L.Rochester, T.Burnett, J.Chiang, R.Dubois,
C.Patterson, N.Golpayegani, M.Kuss, E.Winter, I.Pavlin, T.Glanzman,
T.Usher, W.Focke, E.Charles
** GlastRelease News:
(Heather and Richard)
We are currently at v11r15 (including Geant4 windows fix,
GlastClassify Pass4 irf fix)
With new GlastClassify fix, have reprocessed merit files for the
recent allGamma and will soon do the 55d run. Tracy is working with
Bill on the Pass5 classification tree code (now working on Linux and
Windows). New Particle ID classification tree will be integrated,
either as a separate tree or combined with the Pass5 analysis.
- Discussion of ROOT pointer problem:
(Tracy)
RootTupleSvc::GetItem() vs. direct TLeaf pointers; do not yield same results.
When did this problem first show up? (March/April?)
- G4Generator and G4Hadron status:
[Michael was unable to speak up at the meeting, 'I was "Anders'ed" by
EVO', and provided these comments after.]
On Francesco's request, I added QSGC and QSGC_LEAD to the hadron choices
of G4Generator::PhysicsList (G4Generator tag). Because all hadron physics
choices classes are in G4HadronSim, actually, and not in the standard
include dir but under the src branch (very ugly), also the requirements of
G4HadronSim had to be modified (G4HadronSim tag). I wonder if G4HadronSim
should be reorganized.
- When to open the flood gates for a GR v12, and release the pent up
demand of tagged packages?
(Richard)
v11 series appears in good shape with regard to the ServiceChallenge, so okay to proceed with v12
(Heather)
Will begin now to assemble pieces for v12 build...
Also upcoming:
ROOT 5.14g (bug fixes) now built for all GLAST platforms, ready for further testing.
ROOT 5.16 migration planned for August-ish
** GLAST Citrix box at SLAC still being configured. Heather has
communicated to Rodney a full list of needs; hope to have this ready
for general use soon.
** Continuing search for Goddard software developer. One interview so
far, which went well.
** SCONS progress (Navid)
Latest version of SCONS installed: /afs/slac/g/glast/applications/SCons/0.97/
Working to get SCONS going with ScienceTools: /nfs/farm/g/glast/u06/golpa/ST-v8r2-scons
- note new directory structure (no package tag subdirectories)
- Special file, SConstruct (a Python script fragment), contains
config/build info for each package
- use the "scons" command to invoke functionality (try "scons -h" for hints)
- system does not yet know about GLAST_EXT directory structure,
but Navid expect to add that support soon.
- dependencies put in by hand (for now)
** Event Collections
Heather has taken recent changes in RootIo (David Chamont) with new
metaRootData package (Eric Charles) and has started the process of
adding in the reading and writing the pointer/link files in support of
event collections (see last week's minutes for additional details).
** SAA handling problem in Gleam
Current problem is that once the SAA is entered, it stays there and
never emerges - for the duration of the batch job. This was noticed
during the recent 55-day interleave simulation, in which each batch
job runs for 10 orbit-minutes and it was observed that there were
sometimes up to three consecutive runs with no (or few) events
indicating an SAA transit. GLAST normally spends ~12% of the time
in the SAA so these 30-min transits were suspicious. Toby will
investigate...
** A croc of shoes (Has everyone noted Richard's footwear?)
http://www.slate.com/id/2170301/nav/tap1/
On that note, the meeting ended at 08:50 PDT