ND Production TODOs for the Next Software Freeze

[alexbooth92]

A list of ND production related action items to address before the next software freeze - it is not necessarily complete.

Large tasks:

• Infrastructure for and testing of Pandora for NDLAr step.
• Consolidation of geometry configurations into one configuration file (possibly beyond this freeze - it's an "it would be really nice").
• Inclusion of light simulation (starting with larnd-sim).
• Sanity check that rock generation is behaving as we expect it to.
• Metadata generation.

Small tasks:

• Correct POT information propagated to the CAFs.
• Sort the y offset in NDLAr positioning as larnd-sim sees it.
• Drift length of NDLAr in larnd-sim see here.
• Response file of NDLAr see here.
• Add ability to produce a special sample, for example, neutrino-electron scattering.
• Grab spine configurations from spine_prod (2x2_sim issue #55).
• Switch on h5py dataset compression in flow (Andrew Cudd saw halving in size of flow file).
• Absorb change in SensDet name see here.
• Enable compression at flow stage see here.
• Use whatever the latest and greatest geometry is (both fid and antifid).

Discussions:

• Inclusion of rock in upcoming samples.
• Clarification of what is meant by rock and what is meant by fiducial (related to rock generation sanity check).
• Understand changes to what perlmutter use should be charged to which allocation.
• Understand why in the 2x2 workflow, edep-sim level files are passed to the CAFMaker. Is this something that full ND wants to do?
• Does the simulation of micro-timing struct of the beam need to be updated from NuMI to LBNF.

Things to remember:

• m3249 should be used for production jobs and should be run as dunepro. Details of how here.
• dune can still be used for R & D, analysis etc. Can be run as own user.
• There was discussion about picking a particular flavour from the flux some notes from Rob H on slack here.