Stakeholder meeting notes – 2015/02/05

​Attendees:​

Herb Greenlee, Rob Kutschke, Brian Rebel, Erica Snider

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

Agenda:

Chris brought up the Sirius A-release roadmap, which shows the 32 issues slated for release of Sirius A.  Of the 32 shown, 5 remained to be resolved.  [2/10 update: all issues have been resolved.]

Sirius A will be released as art 1.13.00. The default build will be e7, which will be compatible with OSX (Yosemite).  The e6 build can be produced on request—e5 will no longer be supported.

Items of interest:

fhicl:

Chris mentioned that in the past, allowing white spaces on either side of “::” has been accepted.  From release 1.13.00 onwards, this error will no longer be accepted.  If you have any FHiCL configuration where this is the case, you will get a parse error.

New services:

[[ https://cdcvs.fnal.gov/redmine/issues/3598 ]]

[[ https://cdcvs.fnal.gov/redmine/issues/7751 ]]

There are two new services: TimeTracker and MemoryTracker, which are sqlite-driven services.  They are expected to functionally replace Timing and SimpleMemoryCheck services, which are now deprecated.  User FHiCL files do not necessarily need to be updated, but a “deprecated” warning will appear when either Timing or SimpleMemoryCheck are activated.

There was some concern expressed about the ease of using an sqlite database.  Kyle showed some example sqlite commands.  Some simple commands will be documented on the art Wiki.

TFileService:

[[ https://cdcvs.fnal.gov/redmine/issues/6590 ]]

Now includes supporting TGraph objects, which must be separately registered.  This is done with one command called makeAndRegister<TGraph>(…).

Floating point exception:

[[ https://cdcvs.fnal.gov/redmine/issues/4318 ]]

One of the exception bits was not being set — this has been fixed.

All other business:

Herb mentioned that it would be desirable to have a facility where the canonicalized FHiCL file could be annotated with the source/line number where the FHiCL parameter was set (or overridden, as appropriate).  There was some protracted discussion on this point.  Herb was encouraged to open a new issue in the tracker with a more specific description of the desired feature.​