Present: Joanne Bogart, Toby Burnett, Xin Chen, Johann Cohen-Tanugi, Richard Dubois, Traudl Hansl-Kozanecka, Michael Kuss, Sean Robinson, Leon Rochester, Alex Schlessinger, Tracy Usher, Karl Young
Build issues: (Toby, Alex) Toby has a short list of build problems which the Release Manager could help to address. Alex responded that
in case there is a crash, it should be possible to get the system to automatically start up gdb on the dump and do a backtrace.
he has asked for more disk space. When we have it, it will be possible to save everything from the Release Manager builds and rerun test programs interactively.
he hopes to include nightly builds of HEADs some time this week. (In the meantime, we could turn Karl's nightly builds back on.)
Various people have had mysterious configuration problems which we need to understand. What is different about Marco's Linux set-up so that test_Gleam doesn't fail for him? Why is test_Recon unable to find G4Generator at run time when test programs from other packages with no visibly significant differences in their requirements files have no trouble? And so forth.
To understand problems with finding and loading shared libraries, it can be useful to set breaks in the GaudiKernel System class with a debugger.
Gleam to-do: (Toby) Here is the full list. See Karl's report on Random status below.
The minimal ntuple is in active development. Toby can handle most of the items (for example, it was easy to adapt old code for CAL items) but needs help for the Tracker. (Tracy) Although it's possible to compute all the Tracker items, some of them may be useless or worse than useless as indicators of performance of current Tracker recon code. Better alternatives could be found. If we do continue with the old quantites for now, we have to be prepared to get rid of them later and not let them get permanently embedded in the system.
Right now the ntuple is being generated in meritAlg, but ultimately it needs to move to Recon.
Random report: (Karl) Gaudi tool will likely be ready some time this week. Karl has been involved in discussions with Marc Fisher of the Zoom project at Fermilab about whether it's OK to reseed every event. Current plan is to go ahead with current design, which does reseed.
previous | minutes index | next |
J. Bogart Last Modified: 01-Jun-2010 15:46:08 -0700