December 11, 2006 Software
- Review minutes from December 4, 2006 Software meeting
- FCAL update (Mihajlo): Single cluster energy (profile)and polar angle (profile)vs MC generated values.
- Make system behaviors: BMS Note Make system Note Source tree
- Status of CD-2 milestones
Attendees: Elliott W. Simon T. David L. Mark I. Matt S. Mihajlo K. Zisis P.
Mihajlo showed a couple of plots (see links in agenda above) indicating the successfull integration of the Rad-Phi calorimeter reconstruction code into DANA for single cluster events. The plots do show a noticible descrepancy between the reconstructed cluster energy and the thrown photon energy. Mihajlo and Matt believe this comes from a non-linearity correction due to the Cerenkov angle that is present in the reconstruction, while the affect is not being modeled in hdgeant. The reconstruction in DANA was made to match at 1GeV. They plan to modify hdgeant to include the proper effect rather than take it out of the reconstruction. Once this is done, Mihajlo will move on to looking at πo events with both γs in the FCAL.
Mihajlo also reported that he has made modification to the particle gun in HDGeant that allow one to specify an energy range as well as a range for θ and φ. These changes have not been checked into the repository yet, but he will do that soon.
Make system discussion
David reviewed some of the features of the BMS system curently used to build most of the Hall-D software. This is just a collection of makefiles written by David. The main issues were what we want the default behavior to be and what is easy/possible/reliable to implement. Recently, the default behavior was changed so that invoking make with not target automatically invokes the install target.
The discussion really focussed on the issue of when to relink programs. Ideally, when a library used in linking a program is updated, you'd want to relink the executable. This could be problematic since libraries are found in a search path and one would need to write a script to mimic this search and include the right files in the dependancy list for the executable. At least that's what David thought. Mark pointed out there is an option in gmake to include libraries with the -l (that's "dash L") and put the search paths in the VPATH variable to have gmake do this for you. David will look into this. It was suggested though that if this is unsuccessful, then we should make the relink target the default and add a norelink target for those times we don't want to trigger auto relinking.
Matt will give an update on the background studies.
David will give an update on charged particle tracking.
- Freezing tags in subversion through hooks (David)
- Put tarballs of external packages on Wiki(David)
- Survey alternatives to doxygen(Elliott)
- SVN web interface for browsing change logs (David)
- Put mods to hdgeant particle gun in repository (Mihajlo)
- Design and implement exception and logging scheme for JANA