Core Minutes 3/13/2007

• VS 2005, CMT and so forth: (Toby) See his remarks on the page Heather instituted on this topic and also many more specifics in a child page. Note that some serious thought has to be devoted to impact on RM and on integration with MRvcmt/MRstudio (Riccardo has made some progress here). (Heather) has a couple questions:
  • What about Linux? Do we also attempt to circumvent CMT-as-builder there? (Toby) There is no immediate need for this since CMT does an adequate job of building on Linux, though Jim Chiang has made the point that the makefiles generated by CMT are particularly unpleasant.
  • How closely coupled is this to integration with VS 2005? (Richard) Should this scheme first be made to work with the old Windows compiler? (Toby) Old scheme works, so why would one do this? (Richard) There will be some transition period when we move to VS 2005. People will still want to run old builds. (Riccardo) believes he can suuport both at the MRvcmt/MRStudio level.

  (Richard) on a related topic, is line mode in MRStudio complete? Could RM use it? (Riccardo) Yes, for configures and compiles, but not yet for test programs.

  (Heather) Tracy will be an alpha-tester for Toby's work to date.

  (Richard) While we need to look forward to VS 2005 support, it is no one's highest priority currently.

  [Correction for those who heard Toby's remarks at the meeting: I had nothing to do with glast_make_shlib.bat. According to CVS history, Traudl made the first version, adapting it from cmt_make_shlib.bat, a file which is part of the CMT distribution. Joanne]

• Saving MC trajectories, file sizes, n-tuple variables: (Tracy) doesn't yet know whether the additional data written for gammas will be more than offset by the savings for backgrounds or not. He suggests doing some test runs. (Richard, recalling that this project was instigated by Bill's request for additional variables) Is everything Bill asked for there yet? (Tracy) His part is mostly there; not entirely. Could be ready by tomorrow. (Leon) is ready with his part of the additions. (Richard) Remainder has to do with screw holes which are not currently in the geometry. What are the prospects for getting them in? (Joanne) On this timescale probably nil. The old attempt used an oversimplified pattern; probably we would want something more accurate. Eric has forwarded good information from Luis for top tiles. I have nothing comparable for side tiles, but the more serious hang-up is that I can't get started on this right away; other projects have priority.
• Interleave testing (Tom) Using Pipeline 2, Tom has a skeleton system; it needs fleshing out for robustness and usability. He's using David Decotigny's installation of Python 2.5. He has job options from Tracy for a backgrounds-only interleave test. He needs to add in the Sky, then will aim for a 3-day test run this week.

  (Richard) Speaking of pipelines, he has had an update from Claudia Lavalley who was looking into running pipeline jobs in Lyon. They've been having power problems there, so her work has been on hold. Looks like we'll be using SLAC exclusively to start with.

• ROOT news: (Heather) She has been able to build the patched 5.14 that we require from sources and it seems to be ok, though more testing is in order. Which version of Python should it be built against? (Toby) votes for 2.5, and so does everyone else.

  (Heather) Old hadd problem may have reappeared.  Here's the complete story from Heather:

  Back in the summer we saw trouble when merging multiple CAL ntuples which includes multi-dimensional arrays. ROOT-central provided a bug fix which was included in root v5.13. Back in August we decided to use hadd from v5.13 to perform merges. Now, Franz reports that newly merged CAL ntuples from the BT pipeline appear corrupt, and we again suspect hadd. What is not clear though is that Zach reports he does not see any trouble merging ntuples by hand using hadd. So first we need to be sure hadd is again causing trouble and how far-reaching the problem is.

• BEA: (Heather) At the time of the last ACD update we had decided to ignore it, but it's a sizable mass, so she got the ball rolling and Ken Segal of Goddard has taken it from there. (Joanne) Ken sent some information and then more this morning in response to some questions. Details of material, etc., would be nice but are not necessary for a reasonable first pass, however I can't get to it right away. (Richard) It's needed around July. (Joanne) Shouldn't be a problem then.
• Announcements: (Richard)
  • This fall Jonathon Dorfan is stepping down as director of SLAC, but not retiring.
  • Heather will be running these meetings starting next week. As Infrastructure Queen she has instituted a spot to bring up burning issues.
  • Job description for a new hire at Goddard has been finished.

previous

minutes index

next

J. Bogart, Last Modified: 01-Jun-2010 15:45:20 -0700