Core Minutes 20 October 2009

Core Minutes 10/20/2009

Present: Joanne Bogart, Anders Borgland, Philippe Bruel, Toby Burnett, Jim Chiang, Richard Dubois, Warren Focke, Tom Glanzman, Navid Golpayegani, Elizabeth Hays, Heather Kelly, Michael Kuss, Leon Rochester, John Vernaleo , Eric Winter

ScienceTools: (Jim) gave the report; see it in Confluence.
FSSC: (Eric) is working on getting tools to build on [mac] snow leopard; it will take a while.
Documentation: (Richard) Chuck is out sick; no report this week.
GR RHEL4 validation (Michael) See details here. Summary: the differences seen between builds on RHEL3 and builds on RHEL4 are likely due to slightly different values (in seventh significant digit) for m_cutOffRigidity, set by astro::IGRField::Model.verticalRigidityCutoff(). This code is using primarily float variables whereas the calling code, in the flux package, uses primarily doubles; this could be a factor. It's not clear that the RHEL4 results are worse (or better) than RHEL3. He is not planning to attempt any fixes. (Richard) We might ask Babar if they have seen similar effects.

On a slightly different topic, he compared results (1000 events) of running a RHEL3 build on RHEL3, RHEL4 and RHEL5. All are identical. Similarly, RHEL4 builds produce the same results whether run on RHEL4 or RHEL5. (Navid) That is to be expected. Running a build from an older OS uses compatibility libraries. These libraries are identical to the ones on the original OS, but can only be used for running; you cannot link against them.
GR Pass7.2: (Anders) We have a new tag (v17r35p1), necessary in order to change one of the cuts. Follow progress on this Confluence page. We have several items under Science validation (Vela, GRBs, Extra-Diffuse etc), but the main validation will be people doing real science analysis and seeing that the results are fine. (Richard) How long will it take to reprocess all the data? (Tom) 5-10 days; maybe less.
Materials audit: (Leon) Tracker is essentially done; no news since last week. Robert [Johnson] is interested in tackling ACD next.
Announcements: Fermi Symposium talk and poster deadline is Oct. 22nd.
Core talk:
- GR (Heather) For the record latest tag for Pass7.2 is v17r35p1. Latest tag for use in L1 processing is v15r47p12gr08, made in order to reference latest (gl5) ROOT tag.
- RM (Navid) Windows checkout and compile are ok but there are some unit test failures; ditto for RHEL5 64-bit. Some of the latter problems had to do with faulty SConscript files and have been fixed. Others appear to be real coding errors in case words are 64-bit and will probably require work by package maintainers.
- RMViewer (Navid) There is a new release of RMViewer: RC02. New features include
  - Developer downloads
  - Smaller screen footprint for individual build windows
  - Auto update. Display for builds which are in progress updates every 5 minutes.
  [Note: RMViewer has been installed for RHEL3 and RHEL4 in /afs/slac/g/glast/applications/install/@sys/usr/bin so should be in your path if you log onto one of those machine types and use standard set-up on SLAC Linux .]
- SCons checkpoint release 1.2.0.d20090919 (Heather, Joanne) Heather has been able to use the latest SCons checkpoint release successfully, even in an environment with multiple versions of VS installed, so the problem Joanne sees is almost certainly some sort of set-up anomaly, unrelated to either the SCons source or our SCons infrastructure.
- ASP (Joanne) has found a way to get SCons to write wrapper scripts for python programs belonging to ASP; it should also work for RSP. The remaining work on individual packages is routine and should only take a few days. Verifying that everything works as intended will take some time and consultation with Jim.

minutes index

J. Bogart, Last Modified: 01-Jun-2010 15:45:22 -0700