GR updates: (Heather)
See also Heather's comprehensive tag history.
Coming soon: Ric Claus has released a new version of LDF. Work is in progress on compatible upgrades to ldfReader. A new fsw release is expected around Aug. 28
(Anders) Would like to know how to plan for next L1 release, in particular whether to attempt to add trapezoids and alignment at the same time. He has a couple questions:
(Richard) Having just made an L1 proc release, we probably don't want to rush into another one. In any case, don't plan on combining trapezoids with alignment in an L1 release; timescale of validation for trapezoids is still unknown.
In summary, expect a release reasonably soon, but not immediately, including alignment and new LDF.
Disk usage (Richard) discussed various options to keep our disk usage within bounds. One thing we will certainly do to help in the short-term is spend the $60K or so we have remaining in the 2008 budget for more disk.
ScienceTools (Seth) went through this week's report.
Data handling: (Dan) He's working on a new Pipeline release. See also Tony's list of new data handling software. (Heather) Add to that email from Tony describing use of WIRED.
Reprocessing: (Tom) [provided the following:]
Good news is that nearly all of the pipeline mechanism to perform reprocessing is now running and a small sample of data has been reprocessed end-to-end. There are still a number of details to work out.Current problems:
- Fix SVAC production. Need either SVAC code that can read/write xroot directly, or get skimmer to work in a mode wherein events in a DIGI file can be extracted based on a TCut (e.g. "Entry$<12000"), or write root macro to do this.
- Use proper tstart/tstop in makeFT1/ftmerge when producing FT1 files
- Update run history table in database
- Cleanup root://glast-rdr//glast/Scratch after use
- Test new ftools/ftmerge using Heather's recent installation at SLAC
- Transfer full metadata from L1PROC into REPRO, or find a way to refer back
- Full-scale testing (I/O, CPU, and other loads)
- Test all possible starting points
- Excessive dwell time in READY (slows down testing)
- dataCat registration 'new' vs 'old' method (new method will not reregister)
- processing history DB object does not work
ftools: (Heather) ftools comes with its own Tcl/Tk [8.4] which is different from the one installed natively on RHEL3 systems, hence causing some grief. (Jim C.) We should be consistent and include the required version along with all our other externals. (Heather) Has no personal stake in the matter, but believes the discussion has to include Navid, who maintains all our externals. He returns tomorrow.
skimmer: (David C.) is working on various issues. Nothing to report yet.
SCons: (Emmanuel) has made SConscript files for all GR packages and is working through compiling wih SCons. He is currently dealing with a problem in AnalysisNtuple. He has a couple of ideas. If neither of them pans out he will consult Leon.
(Joanne) is concentrating on porting to Windows (which is not to imply that GoGui on Linux is done: there are still important functions which are missing, but enough is there to make a reality-check on Windows worthwhile). After various adventures, most of GoGui performs about as well on Windows as Linux, with a couple exceptions:
previous | minutes index | next |