Core Minutes 9/29/2009

• ScienceTools: (Jim) went through the written report.

• FSSC: (Eric) has been trying out regression testing done at SLAC; will incorporate into FSSC testing. Also getting ready for the data workshop on Thursday. (John) Db activity continues to show a big spike on Monday mornings, then trails off during the week to essentially nothing over the weekend.

• Documentation: (Chuck) sends the following:

  Nicola has provided more input for an rmfit tutorial running on SLAC Public. At the moment, we're trying to resolve what appears to be an environment variable problem when using an SCons build.

• User disk space & ramifications: (Richard) u31 filled up. This took down its sulky server which also serves the build disk, hence affected L1. Unix admin was called and the problem was fixed, but L1 continues to be vulnerable to this kind of thing. A back-up copy the GR being used for L1 is kept in afs space for just such a situation and similarly for ST, but the ST setup didn't work properly on the copy. (Warren) I tried running config for every package from the command line which is what I do for GR. (Richard) In order to avoid this cascade of problems we could

  • Move nfs user disks to another server
  • Force everyone to use the afs user disk

  Neither would be painless.

• GR RHEL4 Validation: (Michael) took a look at optimized RHEL3 versus unoptimized and discovered that differences there are about as great as between the RHEL3 and RHEL4 builds he had been comparing. There is a subtle difference involving event times even before G4 gets into the picture. It's not due to order of random number invocations; that hasn't changed.

• New mass model: (Leon) has a version which leaves dimensions unchanged and almost has masses matching Robert's numbers. It's still missing about 700 grams per tower, probably in the alignment hardware and flexures. The mass of the wall actually goes down a bit. Zero-degree radiation length is essentially unchanged. Incorporating the missing mass in a reasonable way will require a more detailed model of the bottom tray.

• Pass8: (Tracy) Changes for Pass8 will affect CAL moments analysis and cluster analysis as well as tracking and ACD recon updates. He has been thinking about global pattern recognition algorithms; something will be in Confluence soon.

• Announcements:

  • Abstracts for the Fermi Symposium are due Thursday
  • There will be a partial computing outage at SLAC tomorrow (Wednesday, Sept. 30th). Confluence should stay up but other essentials, like the batch system, will be down from 6 AM to at least 1 PM local time.

• Core talk:

  • GR There is a new tag for L1 use, v15r47p12gr4, containing the following goodies:

    • Updated GlastSvc where the random service seed is set per run rather than per event by default.
    • LdfConverter and RootIo have been updated to use EvtMax==0 to denote reading in a full input file without specifying the number of events they contain.
    • CalRecon was updated to remove two outdated unused tools from CalEventEnergyAlg as requested by Philippe.
    • Trigger now sets its mask via a string in the JO.
    • The enums package now properly defines the TEMBUG constant.
    • Gleam pipeline JO files were updated to use EvtMax=0, and to remove unnecessary use of TriggerAlg.mask.

    [thanks to Heather for the above list]

  • RM and such (Navid) On the externals front, -gl4 build of python is now working on all platforms including RHEL5 32-bit and 64-bit. It can now be used in a build of ROOT. Another external needing a little work is obf. The script producing it has been copying flight builds from the wrong place.

    Developer releases are almost ready to go. RMViewer needs a tiny update to be aware of them.

  • Tagging (Joanne) had some ASP packages to tag and decided it would be nice if stag would edit the new version string into requirements files as well as SConscript, something Leon had suggested a while back. In the process of doing this I discovered that tagging along a branch with stag was broken. It should be another couple hours' work to get everything in order.

  • Externals for GR (Heather) has gotten MySQL 5.1.39 to build on Windows (VS 9) and has written instructions for building on Linux. Kim will do these builds. Tracy has been looking into building our current G4 version on VS 9. It's probably ok, but we'll need to stick with CLHEP 1.9.2.2 for GR at least (we had recently advanced to 1.9.4.2 for SCons-built ScienceTools).

  • CHS and SCons (Joanne) Yesterday I met with Jim Panetta, a real user! We found and fixed some problems; we'll have another round soon. (I was able to build all of CHS with SCons on my laptop some weeks ago, but didn't commit all my local changes and the world has moved on since then.) [Update:  Jim found and fixed a couple minor things on his own and was then able to build CHS; no second round needed!]

previous

minutes index

next

J. Bogart, Last Modified: 01-Jun-2010 15:46:39 -0700