Core Minutes 6/29/2010

• ScienceTools: (Jim) No development news this week. Heather managed to get new versions of ROOT and python ready to go before she left; a new tag using them will be made before long.

  (Richard) A new advisory board has formed, chaired by Julie and Peter. What will the board be advising us on? (Richard) Content of ST? IRFs? Will know better after it's met.

  (Richard) Apparently problems seen last week with gtsrcmaps have been addressed? (Tom G.) work-around suggested by Jim took care of it: just run gtsrcmaps on two sources; use binned likelihood for the rest.

• FSSC: (Eric) No news. It's too hot for news. (John) Test file arrived and looks good under initial testing. End-to-end test will probably occur by the end of the week.

• Documentation: (Chuck) sends the following:

  There's not a lot of news, but I did make a new tag last week (workbook-07-00-11), and Berrie has updated the mirror site.

  Elizabeth Ferrara has taken over the responsibility for updating the FSSC's Fermi Science Tools documentation and has apparently been updating them for some time without realizing that we were supposed to be in the loop. I've asked her to make their files available to me on an ftp site and send me the link, so I can download them and do a compare op with prior version. We're still working out the details, but that would enable me to pick out the changes pretty rapidly.

  Currently, I'm working on the new Instrument Analysis section and updating some existing files as I go.

• Processing issues (Richard)

  • Pass6 reprocessing: original idea was to go with P6_V8 irfs, but that is now on hold

  • Merit reprocessing of Pass7 is also held up due to the major change in event classification scheme. Meanwhile, Eric C. is looking into the extended classes now stored on Astro server. Data volume for Pass7 verus Pass6 is up by a factor of 10. If we don't need to distribute the extended classes it would help a lot.

  • Anders W. Borgland We are doing P6 Earth limb runs with (only!) a new CAL asymmetry file right now (full recon). 1000 cores flat out since yesterday. [update from Richard: limit to 1000 cores is understood. Such a limit was put on the long queue to manage a rogue user one weekend! Tom switched to xlong, peaked at 2200 cores and is almost done. No overloading issues.] (Tom) There are, as usual, a few problem runs but on the whole it's going well. In particular, there haven't been any xrootd problems. However, this job is relatively cpu-intensive. Things might not be as smooth for io-intensive jobs.

  See the C & A schedule of projects for more.

• Pass8: (Leon) is making progress on the mass audit of the tracker. He has a model which splits each of top and bottom trays vertically into two pieces which will allow him to add extra mass in roughly the correct place.

  (Leon) In Pisa Philippe has gotten a new calorimeter clustering algorithm to work; it seems promising. (Richard) Will the new clustering replace the old? (Tracy) There can only be one "algorithm" in place, but, for example, that algorithm could use Philippe's method where applicable, but fall back to the old scheme in other parts of the phase space.

  (Tracy) Bill has doen some tuning in G4 simulation to better represent end effects in logs. Meanwhile, the project to update G4 is stalled; need to get it going.

  (Tracy) and Leon have been exercising WIRED on Windows. Some things work; some don't. It's missing some capabilities we need, things FRED had. WIRED was supposed to be preferable because support was available, but at this point there is zero support. (Richard) Existing manpower is stretched too thin. For example, Tony J. (and Flight Software personnel) are concentrating on EXO which is supposed to take data this summer. Recs are out for 2 new people; not a short-term fix but ultimately (3 months?) should help. This is a topic for the software meeting in July.

• Core talk:

  • SCons topics

    • Command-line tag collector for SCons (Joanne) has written a python script which should satisfy requirements as soon as the tagger script (gatekeeper for cvs tagging operations) has been updated to be a bit more permissive. So far the tag collector script has only been tested in a dummy mode where it says what it will do but doesn't actually do most of it.

    • GR and SCons (Joanne) is not a happy camper although there has been some progress. There are no known issues of any significance on Linux but there are several murky problems on Windows, preventing certain kinds of builds of certain things and, probably more difficult to address, there are problems with getting full and correct environments transmitted at run time.

      On the plus side:

      • after a few minor changes to SConscripts and many, mostly straightforward, changes involving externals, SCons initializes happily.
      • A patch in facilities::commonutilities provides a work-around for one of the run-time problems seen; now most ST test programs are succeeding in RM builds. Too bad it's not understood why this should be necessary.
      • Script for creating a suitably-reorganized obf external seems to do the right thing, but can't really be tested until packages using it can be built.
      • GlastSvc builds and its test programs runs successfully, but only if built non-debug. Otherwise there is a run-time error which is probably due to mixed (debug and non-debug) run-time libraries.

previous

minutes index

next

J. Bogart, Last Modified: