TODO list

This list is not yet ordered by anything meaningful.

Identified items

1. Complete the "usage context" document, turn into a tech note. The purpose of this document is to enumerate the variety of "processing contexts" (e.g. types of workflow) in which HEP event data are used.

2. Identify the "mini-app" to be implemented. The goal is to complete something in 4-8 weeks that can be written up (in a tech note) and then made public. This work should integrate into the portfolio of work that is part of our CCE contributions.

3. Begin paperwork to obtain time on NERSC machines for doing I/O studies. This may have to wait until we have a running mini-app.

4. Arrange appropriate repository to share code with Dana, move relevant code from GitHub marcpaterno/gallery-demo (hdf-branch branch) into it.

mini-app considerations

1. Running on Cori or Edison is necessary.
2. Demonstrating the use of HDF5 i/o is useful.
3. A multi-part mini-app is OK (e.g. art jobs each writing an HDF5 file, one analysis program or Spark program reading multiple files).

The goal is to demonstrate ease of use in writing, and also in processing, analysis-style ntuples in a format different from what is usually used in HEP, and which format can be sensibly used on HPC facilities. Testing will be done at NERSC. This also builds upon the container technology use (which demonstrated running MicroBooNE MC programs on Cori) last year.

We want to provide this as a CCE mini-app.

There is not yet a decision (from CCE) regarding exactly what the deliverables should be.