Present: Joanne Bogart, Toby Burnett, Xin Chen, Richard Dubois, Riccardo Giannitrapani, Heather Kelly, Francesco Longo, Sean Robinson, Leon Rochester, Alex Schlessinger, Tracy Usher, Karl Young
Gleam: (Toby) See his to-do list. The "Noise in cal" item has already been done by Sasha. Richard notes that Gleam v1r1p3 builds readily on Linux.
Packaging, namespaces: After some email discussion of these topics, we've converged on the following decisions:
GlastDigi will be split into separate packages (by algorithm) to better enable parallel development of classes.
GlastEvent will continue as a single package since different subsystem pieces share some common elements and developement is largely over in this package. The package name will be changed to Event.
namespace names, when used, will be identical to package name.
There are still some naming/namespace issues to be resolved for the TkrRecon classes that have been moved to GlastEvent.
ROOT: (Heather) See her presentation on ROOT branching, which shows how applications can choose to read in only part of a ROOT file (however all application which write a particular kind of file should use the same branching scheme). Even though some branches in Heather's proposed Digi branching scheme may be small, there appears to be little or no performance penalty for branches, so we might as well branch as far as is consistent with the logical organization of the data.
New VCMT & glastpack: (Toby, Alex) Work continues to take advantage of the improved GUI and better OO design. The new VCMT will (but doesn't quite yet) use glastpack internally. Alex is using SOAP [a popular remote procedure call protocol using XML for the physical representation of the messages], in particular the SOAP::LITE Perl modules, to communicate with glastpack over the Web.
We might want to require cygwin on Windows installations so that glastpack doesn't have to program around the deficiencies of the Windows implementation of CMT.
Test/release strategy: (Karl) See this flow-chart with description of the steps involved. The use of branches allows ongoing development in the HEAD during testing/patching of a release. Of course everything ultimately has to be merged.
Probably there will be a web interface to the test/release facility, e.g. via the new Perl web server.
Using Gleam MC ROOT output: (Richard) See his report on first experiences. Things are looking positive. Some oddities were found, notably too many McParticles being recorded and some problems in interactive Root for accessing the McIntegratingHit energy map. See the summary slide.
Calibration status: (Joanne) See this write-up for the latest take on infrastructure design and a couple pointers to the pieces that exist. A ROOT client could be readily added, as well as a client (Web based?) to access the metadata directly.
Geometry: (Joanne) Tracker additions (MCM boards, tracker walls) are in the HEAD of xmlGeoDbs. [as of 9 May tagged v1r5].
You can now view summary information (values of constants, materials summary) for the HEAD as well as the latest release of xmlGeoDbs from the geometry page.
J. Bogart Last Modified: 04-Aug-2004 15:40:45 -0700