GlueX Offline Meeting, September 19, 2012
GlueX Offline Software Meeting
Wednesday, September 19, 2012
1:30 pm EDT
JLab: CEBAF Center, F326/327
- Review of minutes from the last meeting: all
- Report from Data Challenge Meeting
- Reconstruction sub-group reports
- Scan-Build Analyzer: David
- Truth information in Monte Carlo events: Richard
- Action Item Review
- Review of recent repository activity: all
Talks can be deposited in the directory
/group/halld/www/halldweb/html/talks/2012-3Q on the JLab CUE. This directory is accessible from the web at https://halldweb.jlab.org/talks/2012-3Q/ .
JLab: David L. (chair), Elliott W., Dmitry R.
UConn: Richard J.
CMU: Curtis M., Paul M., Will L.
IU: Matt S.
Review of Minutes from Previous Meeting
As we reviewed the previous meeting's minutes, we took a moment to refresh our memories on the status of CCDB deployment. Dmitry reported that he has run reconstruction using both CCDB and the files from subversion. The results were (nearly) identical. He also reported that Mark has produced a script that can generate a CCDB compatible SQLite file. Some HOWTO documentation is forthcoming so others can start beta-testing the CCDB. It should be available by the collaboration meeting.
Truth on Monte Carlo Events
Richard gave a report on truth information that is embedded in the Monte Carlo events. This has come up as an issue since he has been working on the data file size issue as part of the Data Challenge effort. He showed some examples where truth information (i.e. information only available in simulation and not available in real data) was placed into attributes of the HDDM XML specification making it difficult to strip from a file. He proposed using subtags instead that could hold the truth information while maintaining an easily identifiable association with a "hit". He showed an example and it was generally agreed that this would be a better approach. There was some concern expressed that the subtags could lead to more complex reconstruction code using the C-interface. Richard pointed out that the C++ interface makes the access as simple for the attribute style as the subtag style.
A proposed, new HDDM for Hall-D was shown. Richard requested that responsible persons for each subsystem look it over and verify that the information contained in the sections relevant to that subsystem are appropriate. Since there are several changes, these will require several changes to the JANA HDDM interface. It may well motivate a re-write using only the C++ interface of HDDM. This will probably not happen before the collaboration meeting.
There was some brief discussion on the Data Challenge activities. The main consensus was that the efforts seem to be progressing along nicely, thanks much to Mark's efforts at JLab on using the farm and others' efforts to work on the data file format (REST) and event size.
David reported on his continuing work on using static C++ code analyzers to try and identify coding errors or potential issues with the GlueX code. This is in response to a recommendation from the Software Review committee report from the Software Review last June.
Two free products are being investigated. The scan-build analyzer utilizes features in the clang-llvm compiler suite. A report on sim-recon software was generated a few weeks ago and circulated via the offline mailing list. The second is cppcheck and David showed examples of reports generated by that and HTML-ified by another free script with info from both merged by a script he is working on.
Both products seem to produce lots of false positives which may limited their use as these must be sifted through by hand in order to find legitimate issues. Further investigation is needed to see what it will take to suppress the issues identified in order to make the tools useful. It is hoped that at least one of these can be used as part of the standard release procedure.
- Tracking: No report (Simon out today) - PID: Nearly ready to check things in, but hesitant to without sufficient documentation. Probably won't happen until after collaboration - Calorimetery: David is ready to make the BCAL time spectrum method the default and will do so soon.