Core Minutes 2/25/2014

• ScienceTools: (Jim) There are several items since our last meeting (Feb. 4):

  • Add ability to tie parameters together in SummedLikelihood.
  • Satisfy requests which came up at the Collaboration Meeting, among them
    • remove automatic differentiation between front and back irfs.
    • support event type selections that are coming with Pass8. It is convenient to have the new EVENT_TYPE column as a bit array so that row-based selections using bit masks can be used, e.g., EVENT_TYPE & 0b01001. cfitsio supports this and has very efficient filters that can be used to apply these types of selections. Unfortunately, tip does not support bit array columns, so I asked James et al to provide it. Currently, the EVENT_TYPE column is just an int (signed since cfitsio doesn't support unsigned int).

  See Science Tools Development Notes for details.

• FSSC: (Joe) Not much news. James Peachey will, directly or indirectly, see that bit array request is satisfied. (Jim) Comments concerning this request should go to JIRA STARCH-18.

• Pass8: (Leon) Philippe has made some plots indicating that some remaining issues concerning energy when comparing data and MC are likely attributable to energy dispersion (on for simulation but usually ignored in source analysis).

• Reprocessing: (Tom G.) New diffuse response correction will be applied to P7 recon data. It should be a fairly quick reprocessing task. It's ready to start up today.

  Within a month or so expect task for new classification, merit (hence also new FITS). May also need new recon for ACD; that has not yet been settled. (Leon) Raising the ACD threshold is under consideration. If adopted, there is more to it than just cutting out hits.

• Servers: (Tom G.) Of the 25 old glastlnx servers, all but 5 are retired. Three more can be retired right away (probable date is March 3); the other two will take longer.

  Hopefully by end of March we will have transitioned to new Oracle servers. This will unfortunately involve some long (approx. 8 hours) outages.

• Core talk:

  • GR (Heather) made GR incorporating the new tag of EarthPhenom at Elliot's request. She hasn't heard any complaints so presumably everything is in order. No other new work (Eric C. has been busy in ACD packages, for example) has been incorporated.

  • Externals (Heather) It might be time to advance to a newer ROOT 5.34 (we're at 5.34.03; they're up to .09). She will check release notes. We cannot advance to next major release (6) just yet as there is so far no Windows support.

    There has also been some agitation for a new Python external, probably just for ST.

  • Event display woes (Leon) WIRED display is dead due to new Java release. Dima is looking into it. The first step will be to get him set up with a newer GR for debugging on a supported OS.

    FRED has died on Linux, probably due to new Ruby. Still works on Windows. (Heather) volunteers to look into it if someone will send her the details. Could be that only minor patches will suffice to get FRED running with the new Ruby. Alternatively we can distribute (the old) Ruby as part of the FRED external.

  • Mountain Lion (Tom S.) Everything is set up. ssh keys are now installed. RM tools are installed and working (although builds fail with compile errors). There is one remaining annoyance: it is not possible to ssh in as glastrm directly; he must ssh using his personal account, then log in (e.g. via sudo) from there as glastrm. The same is true for Heather. It's inconvenient but not a show-stopper. If we can't get this fixed Tony J. will also need a personal account. Advice from the Computer Center has involved looking at the console which is not possible for him or Heather. (Richard) The machine is in Building 84 so perhaps some one local could help out.

previous

minutes index

next

J. Bogart, Last Modified: