Core Minutes 7/3/2007

• End-to-end tests: (Richard) ETE 1b and ETE2 will happen this weekend. The bulk of the Offline work is being done by Anders and Warren (Pipeline) and Heather (new EM release). Data will not include Pointing and Livetime History — still too problematic. (Heather) The EM memory leak has been found. It was due to running OnboardFilter without EbfWriter. (The reason EbfWriter was omitted is that it was crashing..would be nice to know why it crashes with real data!); however we don't actually need OnboardFilterfor these runs. The fix for now is to exclude it as well as EbfWriter.

  Since the EVO panda ate Anders' oral presentation, he kindly sent the following:

  We are supposed to have two science runs for ETE2 which takes place on Saturday. They will be processed in Pipeline 2 with the new monitoring i.e. Fast Mon, Digi and Recon end-of-run plots and trending quantities. Pisa, David Paneque and Martin Kocian have been busy making tags of the monitoring applications. Warren is even busier trying to make them all work in the pipeline. Max is working on the web interface. Tight schedule as ETE2 was originally planned for July 16...

• 55-day run: (Richard) A last problem [something concerning orientation during pointed observations] has been side-stepped by eliminating pointed observations. We're ready to roll; in fact we're rolling. If we get all the resources we're hoping for from the farm this will only take about a day to complete. You can see for yourself how it goes.
• System Tests: (Leon)
  • MC particle Z-termination discrepancy See his presentation for details on different behavior of GlastRelease v11r8 versus v11r7, most likely related to Tracy's fix in G4 of uninitialized quark content. Bottom line is that the discrepancy appears not to be as serious or disturbing as originally feared.
  • Leon suggested he and Heather get together to track down some most-likely-minor EM problems.
  • System tests account password needs to be changed. Leon, with some trepidation, will consult Navid about using escrow.
• MRStudio update: (Riccardo) The version now in CVS fixes all but one of the outstanding Linux problems. He expects to fix that one (multiple builds or cleans when the all target is specified) shortly and then will make a tag. At that point Linux support will just about be caught up with Windows (still missing: ability to invoke debugger) and he will return to adding new functionality.

  (Heather) When/if we replace CMT, it will be with an SCons-based system. Enhancements to provide something like current per-package functions are to be determined. (Riccardo) In MRvcmt, references to CMT were everywhere, but in MRStudio they are restricted to only a couple classes. Conversion to a different build/package management system should not be too difficult.

• Connecting ROOT files: (Heather for Eric Charles) The new package metaRootData can be used to access ROOT files as a group. It doesn't quite compile on Windows and some changes will have to be made to RootIo to use the new facility both for reading and writing.
• Skimmer: (David) There has been a debate last week about what to do with pointing history when skimming merit ROOT files, and where to store the result. We agreed that the pointing history from all input ROOT files should be concatenated and written to a separate output file. A new version of skimmer should be released this week with this new feature.
• Root packages evolution: (David) Once the last segmentation fault has been defeated, I should commit a new RootIo with a centralized management of ROOT trees (task started by Tracy some weeks ago). I hope some users will be able to test it. Next step is to move the non-Gaudi code to commonRootData, so that more clients should be able to benefit from this service. [Thanks to David for the text of these two items. I couldn't catch most of it at the time; I was EVOed! ed.]
• ACD geometry: (Heather) We are not quite ready to use the new xmlGeoDbs tag involving shingled side tiles and more extensive ribbon segmentation but should be within a day or two, when Eric finishes work on AcdRecon. (Joanne) A still newer version of ACD geometry including top screws has been committed to CVS but not yet tagged. She is waiting until it can be tested (when AcdRecon work is finished) before embarking on more of the same for side tiles. Since there are no measurements of side tile screw positions as there were for (most) top tile screws, she will circulate a proposal for uniform invented positions to those who might be interested. It is not likely that the side screws will be implemented before she leaves on vacation this Friday.
• MOOT offline: (Joanne) Nothing to show yet, but MOOT Online is coming along. Code and a suitable XML schema have been written and mostly tested for handling the XML "vote" files generated by Eric Charles' gui to express configuration intent for both physics and charge injection runs. Jim Panetta's client code has been updated for the new stuff and has exercised most of it. Required changes to database structure have been made to production as well as test databases. We are nearly ready to cut over to the new system.
• Pass 5 classification tree: (Richard) Bill's new classification tree is ready but the work involved in integrating it into Gleam is, as usual, substantial. Tracy is working on it.
• AOB
  • 1-year run (Richard) We won't make the original target start date of August; probably it will be more like November.
  • ROOT upgrade (Heather) We should consider moving to the new production ROOT release in August.
  • Core Face-to-face meeting (Richard) Second week of August is still the likely choice. It is unfortunately impossible for Tony, but we can't expect to find a date which works for everyone. One important topic will be the new build system. (Heather) We should have Science Tools representation for those discussions.

previous

minutes index

next

J. Bogart, Last Modified: 25-Jul-2007 11:25:21 -0700