Present: Joanne Bogart, Toby Burnett, Richard Dubois, Riccardo Giannitrapani, Heather Kelly, Michael Kuss, Sean Robinson, Leon Rochester, Alex Schlessinger, Tracy Usher, Karl Young
Imminent GlastRelease release: We're almost ready. CalRecon needs some work; its test program failed in the latest build last night. Would also be nice to take a good look at System Tests output (which is available; Karl will do the looking) before blessing everything. At least one System Test (10 GeV gammas) is taking an awfully long time (20 hours!) to complete [Now looks like this is just due to contention for something. The 20 hours is clock time. Execution time is reasonable. Karl to ed.]
Visual.net: Steve would like a schedule for its support within GlastRelease. Now that workarounds (not always appetizing, but workarounds nonetheless) have been found for the last of the known problems we've had with VC 7 Toby will exercise it for a while to see if he can uncover anything else. Assuming that goes all right, we can start officially supporting VC7, and perhaps plan on cutting over to it entirely 6 months later, which should give users enough time to make the switch.
With the move to the relatively strict VC7 compiler we now have examples of code which compiles successfully on Linux but not on Windows, rather than vice versa. Furthermore, there is no easy way for Linux developers to learn about the compile errors since the nightly build runs only on Linux.
Calibration infrastructure: (Joanne, Heather) Calorimeter pedestals and gains have been implemented. The CalibSvc test program accesses pedestals correctly; gains are entirely analogous. Would like to hear from Sasha before proceeding to implement additional CAL calibration classes. Next focus will probably be ToT and also an investigation of XML Schema, using Xerces 1.7.0 (thanks to Alex, who has installed it on Linux) as a possibly better means of describing the XML for calibration data sets. (Richard) Should we consider using ZLIB to compress xml files? (Joanne) Yes, if we run into performance problems, it might be appropriate.
Heather hopes to set up a new rootUtil (name provisional) package this week which will handle the low-level ROOT file access. Should be sufficiently generic to be used in conversion of event data as well as calibration data.
EM data format: (Heather) She will be importing a new package [in fact, already has], ebfBuild, which will build an ebf shared library from code in the Online repository. Ric Claus warns that the EBF format is still not stable (he's just the messenger). Richard asked whether the output from reading EBF will be a ROOT file or TDS classes. Heather had been originally thinking of just going file to file, but says both kinds of output could be supported easily enough. In either case, the output is digis.
Mini-tower coming: It's expected on March 11. There will be a Tracker meeting here at the end of March.
FRED: (Riccardo) v0.3 is now available. You can use CorbaSvc (soon to be committed to the repository) to connect to a remote or local HepRep server. You'll also need HepRepSvc (package) and the ACETAO and ZLIB external packages. See the FRED status page for details. (Richard): Since FRED is to be our preferred display package, ACETAO and ZLIB should be added to the standard collection of external libraries.
Riccardo has grant money to travel here for CHEP03 so will be available for face-to-face consultation.
AnalysisNtuple: (Leon) ...made a presentation worthy of a mini-documentation review of this new, somewhat misnamed, package. It does finally produce an ntuple of recon results, but the bulk of the code is independent of the ntuple. See the lovely Doxygen output for yourself.
Doxygen proposals: (Toby) Motivated by Leon's use of Doxygen and the fact that we will soon have to start integrating new Science Tools packages, written in some cases by new people, Toby has suggested additions and revisions to our Doxygen recommendations. All comments were favorable. Heather would like to give members of the Documentation Task Force (many but not all at this meeting) one last chance to comment, then will see about updating documentation and affected files in the templates package, meant to exemplify our standards.
We may have to do some scrambling to implement one part of the proposal, namely that Doxygen output be generated as part of the Nightly Build and be available on the web. We're running out of unix web space. Web servers do not serve nfs space.
In support of another part of the proposal, Alex will look into providing a mechanism within the Nightly Builds system to build packages not yet mentioned in any version of GlastRelease requirements.
Release Manager wishlist: (Richard) Alex will soon be spending most of his time on the pipeline rather than the Release Manager. We need to collect and prioritize requests for enhancements.
Science Tools Core? (Richard) How do we help Science Tools get started on their core software, especially given that some of us will become some of them? Two core meetings a week isn't a pleasant prospect. Should we invite them to ours?
previous | minutes index | next |
J. Bogart Last Modified: 01-Jun-2010 15:48:21 -0700