Core Minutes 8/28/2007

• EM retirement: (Heather) The first attempt at a GlastRelease tag (v12r1) to be used for what had been EM territory ran into problems with digis. The digi code was making up volume identifiers for non-connected PMT's which other, geometry-related services were complaining about. This problem is now fixed. Still need to separate out the tri-row bits algorithm. She hopes to have a new tag ready for Anders to try out by September.
• New GR release coming soon (Heather) She has put latest tags of practically everything in GR HEAD. There will be another astro tag soon (patch for Windows) which will also go in. Omitted for now is the latest ACD geometry with slanted, shingled side tiles. (Joanne) The xml part — tiles and ribbons — is done and top screw holes are properly located in xmlGeoDbs tag v1r45p1. Still on the to-do list, but with relatively low priority, are side screws and some minor adjustments to top x-measuring ribbons. (Heather) In spite of valiant efforts Eric Charles did not get AcdRecon fully working with the new geometry. Active distance calculations are sometimes incorrect, hence disabled for now. It will have to wait until he's back in town next week.
• Trigger information (Richard) For real data the true trigger information comes from FSW. Some of the "same" information is calculated by OBF on the ground. We don't have a mechanism for storing both in the tuples.

  (Richard for Eric G.) It is possible to set trigger configuration from within Gleam? (Toby) Believes it is.

  [And now for this week's Core Minutes Quiz
  Q: Does this item look familiar?
  A: Through the word "tuples" it's a direct quote from last week's minutes. It isn't quite what was said this time around, but close enough. Leon pointed out it only took about 3 years to resolve the similar issue of the GLT word.]

• Code status (various)
  • OBF (Tracy) The new fsw library has changes to some of the tracking data structures, hence the Offline code accessing them has to change, as do the relevant TDS data structures. Changes have been made and the new code builds, but crashes at run time. Tracy expects no special problems in resolving this, just routine debugging of new code. The new format is not backwards compatible.  If new code encounters old data, or vice versa, it will go blamo™
  • Acd digitization and calibrations (Heather) This is in; that is, calibration services are used to obtain pedestals. Expect System Tests to look a bit different. There is also a mode (the default, in fact) which uses an ideal calibration not requiring database access.
  • Acd digi smoothing(Richard) At this week's C & A meeting Robert Johnson compared ribbon energies (see p. 9) from real data and from surface-muon MC. The latter distribution has discrete chunks. Isn't there supposed to be smoothing of ribbon response as well as tiles? Could he be using old data? (Anders) The data was generated with tag v6r070329p28. (Heather) That is very old and would explain what he saw.
  • Event Collections (David Chamont) changed the job option used for digi input to be a list rather than a single string since all the support for multiple files is there in the code. He is currently looking into how Event Collections will interact with the Skimmer.
  • mootCore (Joanne) has added a utility which will find the LATC source files, such as GEM configuration, associated with the master key appearing in data. The bulk of the work remaining to make LATC configuration source file paths easily available within Gleam is concerned with writing a new MOOT conversion service within CalibSvc and perhaps changing the organization of the calibration TDS somewhat. It's all supposed to be ready to go for OktoberTest and may yet be.
  • Job options revealed (Heather) Tom would like a log of all job options, not just the ones explicitly set in job options files. Heather asked Gaudi Central and was promptly directed to GaudiHistory, a smallish library which, up till now, has not been included in our Gaudi builds. She has built it for a new version, v18r1-gl4 and has also made code modifications in GlastSvc to make use of it. For now she's not promoting GlastSvc since it also has some changes for the new geometry. [However there would be no harm in promoting GlastSvc ahead of the rest; it doesn't require anything which is not already in HEAD.]
  • ROOT v5.16 Speaking of new versions of external libraries — this one has been tried out on Linux and seems to be fine. At some date in the not-to-distant future Heather would like to move to it.
  • Tracker splits (Leon) The "calibration" splits file in use currently has each plane split evenly between the two readouts, but in reality one (top plane in tower 10) is read out all from one side. He would like to make this change, expected to be entirely innocuous. (Joanne) Once the MOOT conversion service is in place, the true splits configuration for real data will be available that way.
• LEO run: (Richard) The replacement interleave events (Earth10) will give us the proper coverage of rocking angle etc by sampling the orbit we plan to use - 1 sec every minute of orbit time.

  There are some 81000 pipeline jobs, all quite short and all the pipeline messaging to keep track of them seems to be overwhelming Pipeline 2. Tony is back from vacation and looking into what is going wrong.

  After these jobs run, Tom will have to sort the events by PtSCzenith angle and do the xml bookkeeping. Then the interleave run can go. There was some discussion about redoing the charged particle backgrounds due to orbit oddities, but Toby argued that gets lost in the noise in interleave and we can ignore it. We're going with that. This is all been fixed in GR, so will not be an issue next time.

• SCons update: (Navid, Heather) He's working on the ability to override packages with a different version. He has updated his installation in CVS. See this Confluence page for the latest information on this and related topics. The mechanism for maintaining multiple builds (e.g. debug and optimized) made from the same source tree is in place.

previous

minutes index

next

J. Bogart, Last Modified: 11-Sep-2007 14:46:38 -0700