Core Minutes 27 May 2008

Core Minutes 5/27/2008

Present: Joanne Bogart, Anders Borgland, Philippe Bruel, Toby Burnett, Emmanuel Cephas, David Chamont, Jim Chiang, Seth Digel, Richard Dubois, Dan Flath, Warren Focke, Berrie Giebels, Tom Glanzman, Navid Golpayegani, Heather Kelly, Michael Kuss, David Landriu, Chuck Patterson, Leon Rochester, Tracy Usher, Eric Winter

News: (Richard) International Finance Committee has approved our request for money for work on Fred and MRStudio, presumably to go to DataMind. It will be awkward and difficult to justify awarding the money on DataMind, but when Richard suggested the possibility of somehow routing the money through Italy he was reminded not to underestimate the power of Italian bureacracy.

There will be a meeting with SCS on computing needs after launch.
April/May priorities: (Richard) See them all in Confluence.
- Filter bits (Heather) is, for the time being, retreating from initial stab at an implementation whose virtual constructors were giving her grief to something less ambitious. This concerns the TDS; Root part is already working. (Richard) Is AnalysisNtuple ready? (Leon) hasn't really thought about it yet; expects it will be straightforward once Heather's part is done.
  
  (Richard) As with GLT, will now have "traditional" filter bits — those generated by Offline OBF — and the flight collection. We should ensure that both sources produces information in the same format. Ideally code accessing Offline bits should switch to flight version for real data, but chances of finding and modifyiing all relevant code in the next week would seem to be about nil. (Tracy) But if we don't make this a priority it won't happen. (Anders) There haven't been any systematic checks to see how well they agree even for just the gamma filter, though there is some evidence that it's ok.
- On-the-fly Filter config. (Heather) is concerned that a week might not be enough to work all the kinks out of the proposed mode of OBF running: use real flight software when running on Linux at SLAC and Tracy's obf external library otherwise. (Richard) Note hard freeze starts next Monday. Warren disappears Thursday or Friday. (Anders) But Maria Elena will be here; she can handle some but not all classes of problems that Warren normally takes care of.
  
  (Tracy) is ready and waiting. Has set up a local MOOT archive on his laptop, copying files of interest from the production archive. (Joanne) All code necessary for production running has been written and builds. The mootCore parts which read in the filter configuration information and make it available to clients have been tested. Testing the outer MootSvc layer which will provide Tracy's interface will take some doing. There are no production MOOT configurations which include the filter configurations and hence no data created with suitable, but non-existent configurations. However, by adding another short-cut job options parameter which specifies which MOOT configuration — from a test MOOT, not the real thing — to use (information which would normally be looked up from the data) it will be possible to test most of the MootSvc code and will give Tracy a way to try out his code as well. The only part of MootSvc that wouldn't be exercised is code which fetches the acquisition start time from the data. [As of Wednesday the new job option has been added. The code is tagged and Tracy is giving it a whirl.] (Richard) suggests finding out whether Online can give us something more realistic.
- ROOT upgrades (Heather) As of GR v15r3 we're using ROOT 5.18c on Windows and 5.18c-gl1 everywhere else (it's only needed for writing to xrootd, done only from Linux).
- GR validation (Richard) Out of 30,000 jobs run to validate v15r2 only 7 failed. We can safely dispense with the v14r11 series.
- Interleave (Richard) We're planning some new simulation runs: 1) a nadir run, possibly a full sim of 90 orbit minutes and 2) one of the L&EO data taking, now that Julie has the orbit file in hand.
  
  (Tracy) A new GRBgrid problem has arisen: GRBgrid livetime is zero in event header. This is very mysterious since no one is aware of code changes in GRB or elsewhere which could cause this.
  
  (Leon) We still need to do something about multiple copies of pointing info algorithms. He would like confirmation that Tracy's fix (giving the version in FluxSvc precedence since AnalysisNtuple has a use — albeit private — of FluxSvc) is ok. (Toby) For real data FluxSvc should not be involved so the version of this code in FluxSvc is not a good candiate. Originally the code was in FluxSvc; he made what was then a copy in AnalysisNtuple for L1 processing. Since then they have diverged. By the way, he'd like to enable stellar aberration (in AnalysisNtuple version of the code).
Science Tools : Seth went through the Science Tools Update for May 27th.
Data Handling: Dan reviewed the group meeting and subsequent developments in this report.
Documentation: (Chuck) Added information from Philip Hart to How-to-fix on FASTCopy input and output sections. Tony contributed a page on the Data Crawler.
Core talk:

Core Face-to-face Meeting (Heather) The week of July 7th is being proposed. If it's not good for you, speak up!

Skimmer and CEL (David C.) Last week was spend improving checking. He is now working on CEL integrations; he hopes to be done with it next week.

Linux source distribution glitch (Heather) A core dump, apparently caused by running Doxygen on CalUtil, has made its way into the Linux source distribution, making said distribution essentially unusable. She will look into it and alert Zach (package manager for Calutil.

New RM, SCons (Navid) 64 bit compiles of ST with SCons are now working. There was one last glitch concerning linking a shareable to a static library. For this to work correctly on all 64-bit platforms the objects in the static library must be compiled with -fPIC ("position-independent code"), which entails a small performance hit.

minutes index

J. Bogart, Last Modified: 12-Jun-2008 17:12:56 -0700