Core Minutes 9/18/2007

• OktoberTest data generation prep: (Heather) The GR tag v12r4 made last week includes the new ACD geometry with shingled side tiles and ribbons. (Tom) An all muon run (4 million events) and background run (23 milliion events) went smoothly. From the Pipeline perspective the new code seems ok. (Heather) There were some warnings; nothing fatal. One, "ToolSvc.ObfVals...WARNING FilterStatus not found", is understood; it occurs when Ebfwriter filters an event, which in turn happens if no trigger bits or only ACD LO is on. The other, concerning ROOT, should be tracked down but in the meantime is not holding us up. [More detail from Heather: "there is a problem with the OBF object in ROOT. Some of the hairy details of OBF are not being stored correctly. This needs to be fixed, but it is not a fatal error, as this refers to detailed OBF data that is rarely really used."]

  (Heather) Made v12r5 yesterday, including latest astro, CRflux, etc. Still outstanding is a new OnboardFilter. Tracy had a breakthrough last night; a new tag is expected imminently. The astro unit test has been failing; Toby has been asked to look into it. (Richard) Once Tracy's new OBF is in, we'll be ready for the data generation phase of OktoberTest.

• Other GR news
  • mootCore (Joanne) Navid has added the python and swig interface packages to GR LATEST, making it possible to also add mootCore; could do this today. Upgrades to CalibSvc are in progress and are expected to be ready in time for OktoberTest pre-testing (approx. Oct. 1).
  • Calibration TDS paths (Heather) All calibration clients now either use hard-coded paths (AncillaryDataRecon) or the new ICalibPathSvc method (TKR, ACD, CAL) [thanks to Leon, Heather and Zach, resp.] There is no longer any need for the old CalibModel ugliness. Joanne will convert AncillaryDataRecon to the new scheme if Nicola doesn't get to it.
  • use-ing and losing (Heather, Tracy) Some reshuffling of use statements caused problems for TkrUtil and other clients of G4Generator. The latter needs the data file (ParticleProperties.txt) which currently lives in FluxSvc, which in turn uses astro which uses cfitsio.. If the file were moved somewhere else we could get rid of these dependencies. The plan is to move it to extFiles.
  • Composite Event List (CEL) meeting (Heather) We met with Tony J. and David Chamont to discuss the Data Catalog implications of the Composite Event Lists. It was agreed that Tony J and company would help out David as necessary on the Data Catalog details. We will be storing the Data Catalog logical file names in the Composite Event List ROOT files. To avoid continuously querying the Data Catalog, jobs that utilize the CEL files will likely use a utility that will crack open the CEL file and use Data Catalog utilities to determine the actual location of the ROOT files. That list can then be used as input to the skimmer or whatever else is accessing the CEL files. David promised that the skimmer will be updated to allow for CEL input, but will also retain the ability to accept a list of run/event ids as is now the case. Richard, ever concerned about timelines, is asking that we begin using CEL files for the traditional Interleave to replace the map which is written to merit to store the location of the full ROOT files. The timescale for this would be about a month. [thanks to Heather for this text]
• MRStudio meeting (Heather for Riccardo) There is a new tag of MRStudio, v0r6, which fixes the tagging problem on Linux. Although there are still outstanding requests for better support on both Linux and Windows, this tag should be usable for some things. Riccardo would appreciate feedback from those willing to try it out. We plan to meet approximately every 2 weeks to monitor progress. The proposed time is 11 AM every other Friday; next meeting will be Sept. 28. See also Heather's notes on the last meeting.
• SCons (Navid) See his extensive notes on what he's done and how to use it. It's essentially ready to go on Linux, at least for ScienceTools, and usable on Windows though still missing the ability to build developer-friendly project files. Navid would like to go ahead and add ScienceTools-scons and soft links to ScienceTools packages in production CVS so that it's available for everyone to use. There may be some new wrinkles in supporting GlastRelease to the same degree, but the work already done for ScienceTools should cover the bulk of what GlastRelease needs.

  There were a couple questions:

  1. Leon: Will there be a way to "lock" packages to keep them from being rebuilt as one can currently do with MRvcmt on Windows?
     Navid: If the overriding is done in the build path itself then there is no there's no way to do that; SCons will build the dependencies. However if the overriding is done by grabbing an Installer build and having it in a separate directory from the override packages, then the original packages downloaded by the Installer are treated as sort of like external libraries (they are not rebuilt).
  2. Leon: Will it be possible to have variant builds with more than just "Linux" (e.g. OS version) in the name?
     Navid: Currently that is is not programmed in but he'll look into it. In the meantime, there is already the option of specifying the variant name yourself.
  [Thanks to Navid for supplying text versions of the answers.]
• New hires: (Heather) No word yet from the candidate for the Goddard position. The hunt may be on again.
• Travel: (Richard) leaves for London and ADASS on Friday. Sure would be nice to get the OktoberTest runs going before then.

previous

minutes index

next

J. Bogart, Last Modified: 01-Jun-2010 15:45:16 -0700