Present: Joanne Bogart, Toby Burnett, Xin Chen, Richard Dubois, Marco Frailis, Riccardo Giannitrapani, Traudl Hansl-Kozanecka, Julie McEnery, Sean Robinson, Leon Rochester, Alex Schlessinger
GlastRelease: (Toby, Leon) We are ready for a tag. Might want to incorporate just a couple more things: latest AnalysisNtuple tag and, beyond that, a new variable Bill has introduced.
Richard raised the issue of stability of these variables. At some point this will become important. We need to get the Analysis Group to specify/bless a collection of variables and then control these carefully. Toby mentioned that a suitable interface, for example a visitor interface, would render at least some changes, such as additions, backwards-compatible.
EM and data parsing: (Richard) Heather plans to be back at work next Monday. In the meantime, Steve managed to find what appears to be the parsing code on her desktop, bundled it up and sent it to Richard. So far it compiles but does not link, most likely due to external library version mismatch. Xin will be helping to bring this online, perhaps also Leon for tracker-specific bits.
The EM itself is here at SLAC, in Building 33, but at last report not quite assembled: have to get cables, etc., to fit. Expectation is that it will be ready for data-taking in a matter of days.
Analysis GUI : (Richard, Riccardo) See Richard's comprehensive discussion of this issue: definitions and requirements, four possible implementation directions (Gaudi-based, native ROOT-based, JAS-based, Python with PyRoot) and tentative conclusions. So far native ROOT and JAS look the most tenable; his preference is native ROOT.
Riccardo also wrote extensive notes on the subject. He prefers a layered variant of the first (Gaudi-based) approach for general event analysis, both because the full TDS is then accessible and because of the flexibility permitted by layering, but admits that this is probably overkill (and less efficient) for analyses which are just filling and manipulating histograms. There direct access to the ROOT file is preferable.
The Qt widget library, used by the Go4 project, came up for separate discussion. There is potentially a licensing/cost issue on Windows. Riccardo's impression is that Fox functions at least as well (and is available for free on both our platforms).
CHEP03 notes: (various) Most of the talks are available online at the UCSD site. There were up to 11 sessions going on in parallel so we couldn't cover them all. Most of us attending have written up some impressions:
One big worry was the possibility of CMT and Gaudi being discarded by LHC experiments in favor of (pre-existing) SCRAM and (in progress) Seal, respectively, the choices of the LCG project. Judging by various private conversations, we can relax somewhat. LHCb and ATLAS are not likely to give up CMT, and Seal is going to look a lot like Gaudi. Traudl remarked that Seal is focussed on persistency services and development is quite rapid. However others have heard remarks that imply it will ultimately include the full palette of services provided by Gaudi.
If you're wondering what, if anything, we've been doing right all these years, you might be intrigued by Federico Carminati's plenary talk on Software development in HEP.
Calibration infrastructure: (Joanne) [Didn't get to this in the meeting, but it's short and I might forget about it by next week. J.] Sasha has made "ideal" calibration files for CAL gains, pedestals and mu slopes for the LAT instrument. This exposed a bug in the conversion of tower row and column, now fixed.
In the process of moving Sasha's files over a slow link (gain and ped are about 900k) I tried compressing them. They were reduced by just about a factor of 100.
previous | minutes index | next |
J. Bogart Last Modified: 01-Jun-2010 15:45:22 -0700