Present: Toby, Thomas, Ian, Richard, Masanobu, Pat, Traudl, Joanne
Pat reported that the old glast computer on campus has been replaced with a newer, more powerful machine. The change-over is now complete; all functions of the old computer have been successfully taken over by the new one.
The question was raised whether it is time to make the old Stanford CVS repository read-only. Since there is still active development in some of the directories in this repository, the answer would seem to be "no".
Toby requested the ability to do anonymous (read-only) checkout from the Stanford mirror of the SLAC repository. Pat believes he can find a way to implement this [Ed. done].
Thomas has installed a custom set-up script in the EXTLIB package to check (so far on unix platforms only) that the environment variable GLAST_EXT has been defined. It is run whenever the main CMT-generated set-up script is run. He will look into creating a makefile target so that this script is run automatically as needed.
Traudl has made some small modifications to the proposed MC classes and reflected this in her RD.PDF document. She suggests that we order the various lists (containers possess a Compare function to allow such ordering - Root TOBjArrays have that feature as well).
Traudl reported on conversations she has had with the CMS group working with Objectivity. On the whole they are positive. Improvements which Babar has not made (perhaps could not make since they are a running experiment) have allowed them - CMS - to avoid problems Babar has experienced. Is Root adequate for us? Should we also be considering Objectivity? Richard believes that facilities which have been missing in Root (such as the use of stl in persistent objects) either are already or soon will be incorporated.
Toby has divided tb_recon into packages corresponding to the subfolders. This will have to be modified to eliminate circular dependencies. There is also the problem of name clashes with "standard" packages.
Toby has been attempting to rationalize the package structure so that code which is common to glastsim and tbsim need not be duplicated. This is more complicated than it might be, due to the ubiquity of inlining in C++. In particular, although the application and reconstruction sources are identical for tbsim and glastsim, they must be maintained as separate packages because each depends on a detector-specific version of the instrument package.
There hasn't been much during Heather's absence. Ian is getting up to speed. He is attempting to clone the ACDTiles example for the MCTrack class. There was some discussion about how long we will stay with the irf files as a data source for the gaudi work.
We need to smooth the process of running Rootcint as needed when building Rootwriter, probably by defining an appropriate makefile target. LCD should have an example to steal from. Thomas will set up a test suite for RootWriter.
Are we going to use Root for histogram services?
Joanne reported on progress since last week:
Thomas installed the new version of the parser, 1.3, on the V disk at SLAC; we still need to install Linux and, ultimately, Solaris binaries for Xerces in glast afs space.
new versions of instrument and xml have been tested using the pre-existing test program in the instrument package. They appear to be working properly, at least for reading in instrument.xml and glast.prs.
of the 7 remaining packages requiring conversion, 4 are done (that is, compile). glastflux, application, and reconstruction are left.
J.Bogart Last Modified: 06/01/2010 15:46