Core Minutes 11/1/2011

• ScienceTools: (Jim) has been investigating a couple issues but there are no milestones to report this week.

• FSSC: (Tom S.) No news.

• LIGO workshop (Tom S.) was invited to give a talk on sharing data. Most of the LIGO people come from particle physics and don't have experience with the concept. The talk went well; participants were eager to hear about Fermi's solution. Overall the workshop was quite intense.

• Pass7: to reprocess or not to reprocess: (Leon) Pressure is increasing to get this decision made. It was on the C & A agenda for this week but got displaced by other topics so it's deferred to next week. The pros and cons are pretty well understood. To date there have been no pleas from science groups to do the reprocessing.

• Sys tests (Leon) This is taking more (of his) time than he'd like, but most likely now past the worst of it. An enhanced version he had created had unintended — and still un-understood — differences from the official version. But he was able to recover the older version. (Heather) The CMT build v17r35p24gr12 (for which there also exists an SCons build) compares well with the version being used for L1proc. (Anders) identical? or similar? (Leon) at the plot levl, yes. All the KS tests except for ACDMIPS_RIBBONS_KS come out 1.0.

• GR/Pass8 (Tracy) Much effort — especially by Bill — has been going into preparation for his status talk at the International Finance Committee meeting next week. He gave a practice talk on Thursday which was downright inspiring.

  The other major focus recently has been on truncation. Designing and implementing will be a lot of work. Furthermore it will introduce a new era in data-taking: the data will be different, which is anathema to some people. But the gains are significant. It impacts resolution in thick layers and at large angles.

  Philippe is making a lot of progress on recon at high energies (over a TeV). Tracy is completing the recon flow structure while devoting part of his time to the Stanford online class on Machine Learning. (CS Department is offering 3 online classes. This one has an enrollment of 55,000 !)

• Core talk:

  • Moving to redhat 5 (Heather) M.E. has declared herself done with getting L1proc working with SCons builds. Yippee!

  • ST and redhat 6 (Heather) Kim has finished building all the externals. Next we need build ST itself, both manually and with RM. (Tom S.) must create a rhel6 build of the RM code and then set database entries.

  • glast-win04 (Tom S.) Builds had been failing again, so he logged on and then builds started going through properly. (Heather) glast-win04 has some bad memory which might have something to do with all the problems we've had. Replacement memory is on order.

  • Incomplete CVS response (Tom S.) RM decides when to make a new LATEST tag by periodically issuing a cvs rlog command on SConscripts of packages belonging to a container package like GlastRelease-scons. It's looking for a new tag of the (contained) package. If it finds one, it creates a new LATEST tag for that container. The process has been going awry occasionally for GlastRelease and GRBAnalysis because of a problem with astro. astro/SConscript has many tags and once in a while the cvs rlog command output is missing the last bit (which is unfortunately the interesting part), causing RM to think it needs to make a new LATEST tag. The problem does not occur for ScienceTools, even though it also contains astro, perhaps because RM checks only once an hour for ScienceTools. Tom will modify parameters for GlastRelease and GRBAnalysis to do the same.

  • obf (Joanne) finally managed to get SCons to build B3-0-0 on Windows (vc90 compiler), thanks to a steady stream of advice and files from Tracy. So far just the minimum (include files and libraries) are being built. Ultimately the build should include project files as well, but the more urgent next step is to test it by building the OnboardFilter test program against it. That can't happen without builds of two externals currently missing for this platform: TMineExt and geant4 9.4.

    rhel5 builds (32-bit and 64-bit) of B3-0-0 can now be found in GLAST_EXT. Some minor changes must be made to allExternals.scons before they will be usable.

• Announcements

  • Richard will also be giving a talk at the International Finance Meeting on Friday, then will leave on Saturday for ADASS in Paris.

  • SCons meeting Wednesday at 10:30

• AOB

  • (Richard) Apparently there is expected to be a shortage of hard drives (floods in Thailand?). We are at 100 TB or so now. And need ~500 TB for the coming year. 400 for new data; 100 to replace old servers.

previous

minutes index

next

J. Bogart, Last Modified: