Stakeholder meeting notes – 2015/01/29

​Attendees:

David Adams, Chris Jones, Tom Junk, Rob Kutschke, Adam Lyon, Erica Snider

artists: Lynn Garren, Chris Green, Kyle Knoepfel, Jim Kowalkowski, Marc Paterno

[1] Issue discussions:

7538: art service/module command-line helpers

[[ https://cdcvs.fnal.gov/redmine/issues/7538 ]]

Kyle introduced the issue by saying he envisages a facility whereby users should be able to assemble a FHiCL file without looking at the code.  Kyle asked if it would be something that was helpful.  Chris Jones said CMS had such facilities and that they are valuable — there seemed to be some agreement from other users.  Jim recommended Kyle go ahead with items 1 and 2 listed in that issue.  Item 3 will take more work, but Dave Dagenhart is presenting CMS’s parameter-set validation method on Wednesday 2/2 to the Frameworks group at Fermilab.

2473: No error when services present in two places in .fcl

[[ https://cdcvs.fnal.gov/redmine/issues/2473 ]]

Everything in user block has equal status to something outside of it.  The user block has now been deprecated, though still supported.  Rob said he wasn’t exactly intending this type of behavior, but he said it was okay for now.

7560: fcl does not understand files that have lines terminated by cr-nl

[[ https://cdcvs.fnal.gov/redmine/issues/7560 ]]

Chris said that the include expansions are done before fhicl-cpp parses the code, which is where this problem was identified.  It was a quick fix.

7619: art-qualified builds of ifdhc v1_7_1 and related ifdh_art

[[ https://cdcvs.fnal.gov/redmine/issues/7619 ]]

Lynn says ifdhc v1_7_1 has been built per request.  Marc Mengel has a bug fix, however, that was cut and released on ​1/30 as ifdhc v1_7_2, which is available with ifdh_art v1_7_1 in the ifdh v1_7_1 bundle.  The ifdh bundle contains libwda, ifdhc, ifbeam, nucondb, and ifdh_art.

7662 : Allow absolute include paths in fcls

[[ https://cdcvs.fnal.gov/redmine/issues/7662 ]]

For now, the artists request that NOvA discusses this as a collaboration, and then inform the artists of their decisions.

[2] Discussion regarding a new art release for Yosemite:

Mac OSX (Yosemite) requires a gcc 4.9.2, or a patched version of 4.9.1. Marc asked if there was enough interest to release a version of art that has gcc 4.9.2, along with some other version changes for other packages, before the complete list of Sirius A issues was addressed.  Adam said that g-2 favors moving ahead.  Suggestion to release art 1.13.00 with a pared-down version of addressed Sirius A issues.

[3] Discussion regarding clang support:

Qt graphics in G4 might push art to supporting (only) the native clang.  In that case, a decision may be made to not support gcc on a mac.  Chris Jones said that CMS has had troubles incorporating clang support as clang and gcc have different interpretations of the standard.  A decision to support clang, however, is independent from committing to a Yosemite build of art.

[4] Lynn mentioned that a new version of ups (5.1.3), will be distributed.​​