ScienceTools: Seth presented the report. (Jim) The current ScienceTools release, v9r13, is suitable for Pass7 reprocessing, but does not yet include Pass7 IRFs; they are still being evaluated.
Pass7 reprocessing: (Richard) Tom started it up before he left, submitting 2500 runs out of 4300 total. Pipeline status so far shows ~1100 completed. There were a fair number of failures, possibly disk-related. (Anders) The failures appear to be intermittent; I think retries are succeeding. (Richard) At this rate, expect the reprocessing to take the rest of the week, to be followed by generation of FT1 and FT2. (Anders) There are no other problems (known problems, that is).
(Leon) and Anders examined a small number of events analyzed two ways:
and verified that the two methods found somewhat different tracks, not the desired result. The GR release was GRv17r31 for both analyses, but calibrations were not: newer calibrations were installed before doing the full recon. It's questionable whether updated calibrations could account for all the differences seen in track-finding. [Post-meeting update: they're assembling machinery to re-do the full recon with the older calibrations.]
GlastRelease: (Heather) Most recent tag is v17r31. There is a separate branch for Eric Charles' new ACD code.
ScienceTools on Windows: (Toby) In recent RM builds several packages are failing to compile on Windows. (Jim) would be willing to work on packages he is responsible for but doesn't have access to an acceptable Windows development environment. He has been unable to use CVS from the UW server for 6 months. (Heather) Emmanuel has just started using the UW server successfully. He might have some advice. [Also see doc].
Be sure to vote if you haven't already on date for upcoming GLAST Core Week. Leading contender so far is last week of July.
SCons RM (Navid) A couple weeks ago he had gotten a message from SCCS that Mac batch jobs were were working, quickly followed by another message saying they weren't. Sporadically a job will run correctly, but most still have problems with afs tokens. The Mac crashed on Friday and Navid immediately sent a message to SCCS asking them to bring it back up. No response yet.
The RM builds on Windows have gotten slower and slower. Disk fragmentation may be a factor. Navid is cleaning up unneeded files, then will see if it's worthwhile to defragment.
SCons and Windows (Joanne) Now that all ScienceTools externals have been built for the new Visual Studio, she has been adding packages to her mini-container. st_graph is the latest problem package; may discuss in more detail at tomorrow's SCons meeting.
CHS (Joanne) has added or modified SConscript files, made tags, etc., so that CHS should now be ready for SCons RM to build.
Externals (Emmanuel) is working on getting the new Fred going on Windows. Glast05 proved to be an inadequate development environment; he is now set up on the UW terminal server. Some changes in the requirements file may be needed.
previous | minutes index | next |