Stakeholder meeting notes – 2014/11/13

​​Attendees:

Chris Backhouse, Kurt Biery, Chris Jones, Rob Kutschke, Brian Rebel
artists: Lynn Garren, Chris Green, Kyle Knoepfel, Jim Kowalkowski, Marc Paterno
art 1.12.04
– Builds are proceeding and will be located at scisoft.fnal.gov/scisoft/bundles (see below).
Highlights:
1. messagefacility has syslog destination
2. new null output options
   “-o /dev/null” means delete at most one output
   “–no-output” means delete all outputs including the module definition clause and any paths
3. Issue 7224 re. sam_metadata_dumper and various issues related to JSON parsing
   – To ensure legitimate JSON input, the art JSON flag needs to be enabled, otherwise art will “guess” as to what it should be.
4. cetlib: new function signature that returns a canonicalized string by value instead of through passing by reference.
5. ROOT 5.34.23 included with art 1.12.04 distribution
   – includes additions by Chris J. and Philippe Canal related to thread safety
6. novaddt and nu distributions now include artdaq_core
7. Some documentation to be fleshed out.
Discussions:
1) Rob brought up a concern regarding the unintuitive meaning of an empty end_path vs. a non-existent end_path
– empty end_path executes no paths
– absent end_path executes all paths
Rob would like empty end_path’s and absent end_path’s to mean the same thing.  Marc recommended that Rob discuss the issue with various users to formulate a proposal to make to the artists.
2) Chris B. noted a regression in time to make histograms, etc. for a recent ROOT version.  He wondered if these issues were addressed in the latest release.  No one knew if that issue has been resolved, so Chris G. recommended putting in a JIRA ticket to the ROOT people if it is still too slow.
Discussion on Issue #7303
The original requirement was that the –sam-application-family, –sam-application-version, and –sam-file metadata routed to SAM needed to be consistent, which was checked at the post FHiCL-processing stage.  This can be done as late as a FHiCL file (does not have to be atthe command line), but it cannot be done in code.
General tone of original discussion was that the primary responsibility of ensuring consistency of metadata was on the user end, with the exception of some information that is naturally available w/in art — runs used, etc.
At the moment, however, it is not clear whether the behavior Chris B. is observing is enforced by art proper or the NOvA software layered on top of it.
The artists’ recommendation is that experiments need to first investigate their own code and see if the behavior they are observing is due to user code or art code.  If the behavior is enforced from art and different behavior is requested, then the interested stakeholders should meet, along with the artists, and the SAM folks to discuss a resolution.
scisoft.fnal.gov has been restructured
1) art 1.12.04 and all future releases are found in scisoft.fnal.gov/scisoft/bundles.
2) pullProducts is no longer versioned.  Instead, the version is part of the bundle spec: art-v1_12_04, mu2e-v1_12_04, nu-v1_07_00.  Note that the nu distribution uses the nutools version since nutools may change independently of art.
Announcements:
There is a new art Sharepoint site:
It is useful, but not complete, and is being updated.  The artists request patience as the design and content are finalized.​