GlueX Software Meeting, October 13, 2020
GlueX Software Meeting
Tuesday, October 13, 2020
3:00 pm EDT
BlueJeans: 968 592 007
- Review of Minutes from the Last Software Meeting (all)
- Recent workshops:
- HDF5 and GlueX (Richard)
- Review of recent issues and pull requests:
- Review of recent discussion on the GlueX Software Help List (all)
- Action Item Review (all)
Present: Alex Austregesilo, Nathan Brei, Thomas Britton, Sean Dobbs, Mark Ito (chair), Igal Jaegle, Naomi Jarvis, David Lawrence, Susan Schadmand, Simon Taylor, Beni Zihlmann
There is a recording of his meeting on the BlueJeans site. Use your JLab credentials to authenticate.
- Alex called our attention to the recently announced upgrade of the Tape Library and the associated down time, scheduled for November 2-4. See Bryan Hess's message to the SciComp Briefs mailing list.
- Sean pointed out that we did not have a plan for the software/analysis talk on Saturday morning of next week's Collaboration Meeting, so we made one. Alex will give an update on processing. Mark may contribute a slide or two on software issues.
Review of Minutes from the Last Software Meeting
We went over the minutes from September 29 without significant comment.
We discussed two workshops, sponsored in part by the Lab, held in recent weeks.
- Joint GlueX-EIC-PANDA Machine Learning
- David was one of the organizers and he reported that the workshop was generally well received and another one is planned for next year.
- Richard had good things to say about the ML for beginners sessions given by Thomas Stibor.
- Naomi lamented the lack of time, both during and after the workshop, to work through the practical exercises provided.
- Future Trends in Nuclear Physics Computing
- Thomas commented that some of the topics (shared software, career paths for scientists working on software) are becoming a common feature of many workshops.
- Mark mentioned (as he did last week) Rucio as a product that has promise for use by GlueX. It does scientific data management, that is, a system for keeping track of files and their various instances across multiple storage sites.
HDF5 Support for HDDM
Richard gave a presentation on work he has been doing with using HDF5 as the underlying I/O system and associated record format for HDDM. Please see [his slides for the details. His slide titles:
- What is HDF5?
- How does HDF5 relate to HDDM?
- How does HDF5 support get included?
- How does HDF5 support get built?
I/O using HDF5 will likely be better optimized than the code used by HDDM currently. It also provides a mechanism for random access. The C++ interface is done and Richard is working on Python support.
Port of halld_recon to JANA 2
Nathan gave us an update. All of the JEventSources have been completed and he is on the plugins and then to testing.
Mark raised the issue of trying to incorporate unit testing into our software stack. Here the first problem is to define what we mean by unit testing. Nathan was in favor of implementing some sort of system and has thought about the issue a bit. Although he thinks that it would be a great thing to have, he is worried about a particular feature of our code: configurable operation of software components. For each configuration or option for a factory, for example, different testing code may be called for, at least in detail, and as data is passed from stage to stage, the number of configuration combinations may make for a large number of cases to test. We need to think about and discuss the problem further.
Copied from the last meeting:
- Ask the Computer Center about staging the start of jobs in time. (Mark)
- Make sure that the automatic tests of HDGeant4 pull requests have been fully implemented. (Mark, Sean)
- Finish conversion of halld_recon to use JANA2. (Nathan)
- Ping Dmitry on CCDB Issue #68. (Mark)
- Remind folks about their problematic cache files. (Alex)
- Finish CentOS 8 compatibility project. (Mark)