Core Minutes 9/22/2009

• ScienceTools: (Richard) The written report is nearly skeletal this week. The PSF update proposal presented by Matthew yesterday at C & A will impact ST; may want to reprocess FT1. What about FSSC? Will it pick up the new data set? (John) I suspect that we want to avoid letting the FSSC dataset fall behind the SLAC one. (Anders) We may combine the release of the new PSF with Pass7, depending on when Pass7 happens.

• Pass7: (Anders) Bill made a new worksheet in which extra-diffuse is an additional class. But he's out of town now and Tracy does not have all the information he needs to integrate it into GR. When this is done, we will reprocess the data (most or all) to produce new merit and FT1. It's likely we'll have to iterate once before we're ready for a public release, including reprocessed data of all kinds except FT2 and perhaps livetime cubes.

• FSSC: (Eric) No news.

• GR RHEL4 Validation: Validation of RHEL4 has been deemed "ok" but we still haven't understood all the differences in the CAL plots. (Heather) Sasha is looking at it; could be helpful if other people — Tracy and Philippe spring to mind — did as well. Relevant materials have been forwarded to them.

  RHEL4 compiler warnings (Heather) People who were inclined to fix them have done so. The resulting tags will be merged into GR v17rsomething.

  We now have a native RHEL4 build of the obf external, which should eliminate a pile of warnings. Heather will inquire about 64-bit and RHEL5.

• New mass model: (Leon) We have had an inaccurate model for a long time. Robert has tracked down a lot of information as to what should be there. Leon is implementing it, a cumbersome process partly because what we know are masses whereas what G4 wants are densities. The xml format can handle it [was augmented to do this very thing for spacecraft volumes some time ago. ed.] but it's not much fun. He now has something which is much closer to Robert's finding, but not identical. He expects by next week to have a model with the same volumes and correct masses. He'll then try this out in the beam test model.

  Meanwhile Robert is looking into other inaccuracies in the model, the most flagrant probably being the missing Base Electronics Assembly associated with the ACD. [Up until the current work on Tracker essentially nothing has been done in this area for a couple years. For status at that time, see p. 5 of a presentation for C & A and associated tables. ed.]

• Documentation: (Chuck) sends the following:

  I'm finishing up a rewrite of the LAT GRB Analysis tutorial. Nicola provided valuable input late yesterday afternoon that I plan to incorporate over the next day or two, and then I'll be working on the segue to "rmfit", which is the primary burst analysis tool currently in use.

  I've also incorporated a number of updates, which you can access via the change log, which is linked in from the WB's splash page. I plan to create a new tag towards the end of this week, so we can update the mirror site.

  (Toby) There was a request to document pointlike for the Workbook. He has been using native python tools having functionality which is a superset of what Doxygen can do, but the result is not compatible with the Workbook. (Richard) See existing tutorials in the Workbook to get an idea of what's needed. They are inevitably hand-crafted, not automatically generated with a tool.

• Core talk:

  • RHEL4 validation for MC; Systests (Liz) The system test comparisons of Monte Carlo simulations still show differences for composite source mixtures. My system test data test (in dev still) run is failing for rhel 4. That is probably an error somewhere in my set up; I anticipate help from the experts. [Thanks to Liz for supplementing my fading memory. ed.]

    (Michael) has been making comparisons of MC on RHEL3 and RHEL4. Energy deposits (for what is nominally the "same" run) differ, as do step sizes. He would like to determine whether the behavior of our code is different or if the differences are predominantly due to G4. He will check if our code is coming up with the same particles on both OSes.

  • Wrappers/setup scripts (Joanne) has tested Navid's new code on Linux and Windows (new, improved single-step design for setup) and it works!
  • GR and SCons (Joanne) Reorganized G4HadronSim has been committed but is untested.
  • ASP and SCons (Joanne) In order to test that proposed changes to ASP for building under SCons have not broken CMT builds, she needed a CMT build of ScienceTools. Since this work is being done on a 64-bit RHEL5 machine (building 32-bit binaries) that involved breaking some new ground, but it's something we need to do in any case. There were some minor glitches (e.g. 32-bit/64-bit problem in SwigPolicy; gcc41 compile error in healpix), now fixed. All of ScienceTools now builds in this environment.
  • RHEL5 externals (Heather) Kim has built all ScienceTools RHEL5 64-bit externals [Update: There was a problem with the python build. Navid has redone it. Now ROOT has to be rebuilt, using the new python. Should be done some time this week.]
  • Windows and SCons (Toby) Is it ready for developers to switch over? Sounds like not quite yet. (Joanne) There will be difficulties for anyone who tries to do serious work with it — that's a promise — but we can't get past this to a better system unless someone does try. The installation now on the Terminal Server should be upgraded to the most recent ScienceTools release (with slight modifications needed for building project files) and the very latest GoGui release (0.9.7) should be installed. It is compatible with the new setup file.

previous

minutes index

next

J. Bogart, Last Modified: 01-Jun-2010 15:45:13 -0700