GlueX Offline Meeting, February 23, 2010
GlueX Offline Meeting
Tuesday, February 23, 2010
2:00 pm EST
JLab: CEBAF Center F326/7
To connect by telephone:
- 800-377-8846 : US
- 888-276-7715 : Canada
- 302-709-8424 : International
- enter participant code: 77438230# (remember the "#")
Talks can be deposited in the directory
/group/halld/www/halldweb/html/talks/2010-02 on the JLab CUE (you have to be a member of the "halld" Unix group). This directory is accessible from the web at https://halldweb.jlab.org/talks/2010-02/ .
- Review minutes from last meeting: all
- Action item review: all
IU: Matt Shepherd
JLab: Craig Bookwalter, Mark Ito (chair), David Lawrence, Lubomir Pentchev, Yi Qiang, Sascha Somov, Simon Taylor, Elliott Wolin, Beni Zihlmann
Review of Minutes
We went over them. Not much was said.
- New release: Mark reminded us that release-2010-02-11 has been out for a week or so now. No problems have been reported so far.
- Environment variable checking in the makefiles: Mark mentioned that one of the upper-level makefiles in the src tree (Makefile.dirs specifically) now performs a check for key environment variables and gives a simple error message on faults. This should avoid errors later in the build, errors that are likely harder to interpret.
- hdgeant++ fixed: Mark reported that hdgeant++ will now pop the required windows on start-up. Beni was consulted to identify the fix. A change was made to the src/programs/Simulation/hdgeant Makefile.bms to support this. David reported that this caused a build error for him on MacOS (missing library). He will report back with more specifics.
- Auto-notification of build errors: Mark pointed out that we are now getting email notices of errors and warnings from the nightly builds, on Mondays. David has already beaten back most of the warnings that we had some weeks ago. This feature of the nightly build system is documented on the Nightly Builds of GlueX Software wiki page.
- The problem with named pipes: Beni had been in communication with Richard on a problem using a named pipe between hdgeant and mcsmear in order to avoid creating the intermediate HDDM file. The problem is that the a seek on the output stream is performed to write the bytecount for an event after the bytes are counted. Seeks on pipes do not work. The code could be modified to support pipes; Mark will consult with Richard on cost vs. benefit.
A couple of other items arose during discussion of the last announcement (not closely related):
- Truth and Hits: We decided to have a policy that the data structure should support the simultaneous existence of Monte Carlo "truth" information and detector generated "hit" information, to allow comparison. In particular, a scheme where the hits over-write the truth is discouraged. David volunteered to look into cases where this might be occurring.
- Noise Hits in mcsmear: We decided to change the default behavior of mcsmear from noise-hits-generated mode to skip-noise-hit-generation mode. The -n switch is currently used to select skip-noise-hit-generation. A new option will have to be implemented to select noise-hits-generated. David will work on this.
Action Item Review
We went over the list. Discussion was generated.
- Moving bcal to mcsmear: The bcal smearing is presently done in the reconstruction and not in the smearing program. The structure of the "hits" is different than the "truth" in this case (double-ended readout) and the appropriate objects are created in the reconstruction. If the smearing is done in the smearing program however, that information must be transmitted in the HDDM file to the reconstruction. Matt asked about whether we need to modify data model to do this and if so, how to get started. We thought that this modification was best; David offered to help. Beni pointed out that the TOF must have the same issue; he volunteered to investigate and report back.
- MacOS Build: David has received and set up a new Mac-Mini to use as a testing platform. The main problem is getting a 64-bit build of CERNLIB for use by HDGeant. Another option is to try an all-32-bit build, not the natural choice for Snow Leopard. David and Mark are just getting started on this.
- GEANT4 for GlueX: The preceding discussion brought up the issue of migration to GEANT4. It seems likely that we need to spend some effort here.
- Adequacy of wire-based tracking for identifying neutral clusters: We decided to drop this item.
- Using a pre-built build: We need better documentation of the HALLD_MY scheme, where a small subset of the code can be checked out and modified and the bulk of the software can come from pre-built, public software.
New Action Items
- Modify data model to allow simultaneous existence of truth info and hit info. -> David
- Change the default behaviour of mcsmear to not generate noise hits. -> David
- Understand smearing scheme for time-of-flight. -> Beni
- Write some HALLD_MY documentation.
- Review problems with hdgeant++ build on MacOS and report to group. -> David