Core Minutes 9/17/2013

• ScienceTools: (Jim) addressed some memory issues with Likelihood:

  • Spacecraft data memory can be released after it's used.
  • cfitsio the entire input file holds in memory when using row filtering. Jean wanted to retain row filtering, but by adding column filtering when reading in FT2 file we could reduce the memory usage somewhat.

  There are also memory issues to be resolved in gtmodel.

  Joe will work on changes to gtrspgen needed in the new release.

  (Richard) What is the status of handling of time-dependent IRFs? (Jim) Code exists and has been tested, but not expected to be needed right away. Handling different event types is of more immediate concern.

  Jeremy has opened a JIRA concerning a gtltcube crash under unlikely conditions. Jim will look into it. (This is fixed in Likelihood-18-02-03.)

  See Science Tools Development Notes for more information on several of these issues.

• FSSC: (Joe) is working on building the release candidate. It looks good so far. (Elizabeth) is working on documentation for the data release. That release is on track for next Tuesday.

• Pass7 reprocessing: (Tom G.) We're caught up to August 13. He plans to interrupt P8 (for about a day) to finish off.

• Pass7 L1 release: (Tracy) Need to look at sys tests. Recall Overlay was broken with new Gaudi. We revised the P8 version somewhat to incorporate into P7. The P7 code reads only the P7-appropriate stuff. This makes it difficult to compare: can't expect event-by-event identity. The shape looks ok; numbers of events are different. Probably it's all ok.

• Pass8 reprocessing: (Tom G.) We're currently 56% done; expect to finish in late November.

• Storage: (Richard) Recs have been approved and may have already gone out for 350 Tbyte of disk and a Pbyte of tapes. We should have it by the time we need it.

• New flavor: (Joanne) met with Leon yesterday and wrote up instructions on use of the stand-alone program copyCalibration. His overall plan:

  1. Do the first urgent part of the copying (all calibrations in use now or about to be in use by L1) before leaving on vacation on the 20th. This stage also involves changes to jobOptions to use the new flavor and making a new GR. Make the switch by changing over to the new GR. New calibrations will get the new flavor.
  2. Once this all looks good, make copies of all calibrations previously used by L1 and change Pass8 reprocessing to use the new flavor.

• Geometry: Leon has made a new tag of xmlGeoDbs adjusting materials for ACD support (previously used first-guess materials which were less massive than the real thing). The new geometry is being tested to see if it improves comparison between MC and real data. See also JIRA LPATE-162.

• L1 Upgrade: (update from Heather) L1 is preparing to upgrade on Sept. 24th. It will include the changes for new calibration flavor as well as updates to stay consistent with ST tags. ROOT version is a special case, still under discussion.

• Core talk:

  • Windows boxes (Richard) Several have been shut down, but not yet removed until we're sure we don't need them. Our RM workhouse glast-win04 remains as well as web front-ends glast-win01 and glast-win02.

  • Mac Pro (Tom S.) currently has no access to this machine. He'll see about getting an account. We need to find a way to get access to nfs space from that machine (won't get it automatically).

  • Release diffs (update from Heaather) There are again calls for Release diffs, e.g. as in CMT RM where we had release.notes diffs. Others feel having easy, per package access to a full diff would be more useful. I've sent a query to Tom S and Tony concerning how we train Jenkins to do that and what info we can access from the RM DB to figure out the location of the code, etc. Tom is hoping to get more details about what people want today. I suspect Leon has plenty of opinions - but I'd like to keep this simple and we can try iterating on it.

previous

minutes index

next

J. Bogart, Last Modified: