Core Minutes 8/12/2008
Core Minutes 8/12/2008Comings and Goings: (Richard) It was nice having Philippe at SLAC for an extended stay; he'll be going home on Thursday. Welcome t John Vernaleo, joining us from GSSC.
GR updates: (Richard) GR v15r36 is the current tag. There is a problem with Recon. processing data which is handled successfully with v15r33. (Leon) It doesn't look like the kinds of problems we've had in the past with the propagator. There is a message
AcdReconAlg ERROR caught exception starting propagator on track 0
then AnalysisNtuple crashes with a different error. (Tracy) AcdReconAlg is behaving as it was designed to do, putting out that message but then attempting to keep going. He suspects this is a case of a situation we've seen before, where the propagator is asked to go outside the mother volume.
(Philippe) Why was the trapezoid geometry [one of several upgrades in the newer tag] put in now? He suggests it be put on a branch and that we do a sample day and a muon run with it. (Leon) Was the one to suggest bringing in the trapezoids; the new geometry does plug some holes. In retrospect maybe it wasn't such a great idea to do it now. (Joanne) The trapezoid geometry has been available for about 10 months. There apparently is no good time to put it into production. However if we want any further refinements to the geometry, which we probably do, this upgrade has to be given some priority — which is not to say that it should go in before it is adequately tested [that is, tested as well as we can manage with finite resources.] (Anders) We should make the data sets and examine them right away rather than letting the issue hang around.
(Richard) What is the status of spacecraft alignment and stellar aberration? (Toby) The alignment, using the new calibration type, is in and verified. Stellar aberration seems to be ok, but the correction is very small so it is harder to be sure the code is really doing the right thing. (Jim) meanwhile wrote something for his own use to do the job for ScienceTools; he will make it generally available.
(Heather) will make another release, backing out trapezoids and putting in alignment and stellar aberration. Since we are still frequently making significant changes to L1—e.g., a recent new fsw release and another one on the horizon—she does not want to embark on maintaining branches.
(Leon) Back to exception messages: it would be nice if there were uniform handling which put out event and run number. It has been done by individual maintainers in several places.
ScienceTools (Jim) went through this week's report.
Data handling: (Richard) Tony is about to release a new astro server (and leave on vacation the next day). [See his descriptions of several new tools in Confluence.] (Heather) There is a new skimmer, v6r2p1. Tony ran into a problem when he tried using it; no response from David yet. [Post meeting: Heather has a note from Tony saying he is updating to v6r0p2, and there will be a way to specify skimmer version, for easier testing in the future.]
Reprocessing: (Tom) [kindly provided the following:] The reprocessing task development continues but with a to-do list that looks much like last week's due to a couple of time-consuming land mines. One notable problem is that fmerge (used to merge FT1 files) was discovered to silently truncate file names which exceed ~100 characters, creating an inconsistent and confusing failure mode. Fortunately, a "next generation" fmerge, called ftmerge, was at hand. But unfortunately, the version of ftmerge installed at SLAC does not match the heasarc online documentation (e.g., the copyall option is not recognized).
This raised the issue of how the ftools are maintained at SLAC and whether we should upgrade to the latest version (6.5 from 6.0.4 dating from November 2005) and whether our installation ought to be stored in a release-specific directory. Currently, a single release is stored in $GLASTROOT/applications/astroTools.
(Heather) volunteers to look into regularizing our installation of ftools
SCons: (Navid) There's not much to report on SCons end. I'm continuing work converting the 3 parts of the RM (RM code, batch submitter, and workflow). The RM code is done and the batch submitter is pretty much converted to using the non gui portions of QT rather than using several third party libraries to accomplish the same task. Once I finish testing the batch submitter, I will work on converting the workflow part of the rm. When that's done I'll enable the new RM to submit jobs for GR and ST.
(Emmanuel) is now compiling GlastRelease with Scons, finding and fixing compile errors as he goes.
(Joanne) GoGui is now able to build a package using SCons; still need to try various permutations involving variant packages. Clean is implemented but isn't yet doing quite the right thing. Support of CVS operations is nearly complete. From GoGui one can
Various configure options are supported and saved from session to session; there is still more to do. Later this week she hopes to start porting to Windows.
AOB:
(Richard) Target date for the Core meeting is the last week in
October 
previous
|
minutes index
|
next
 |
J. Bogart, Last Modified:
01-Jun-2010 15:47:05 -0700
