HSF Software Forum on EIC Software Consortium, November 21, 2018

Introduction

Reminder about the HSF/OSG/WLCG Workshop, 18-22 March 2019 at JLab
- We hope to open registration soon - before Christmas
Physics Event Generator workshop is next week at CERN

All current frameworks are more or less based on ROOT.
Are the detector concepts designed to read out cleanly, event-by-event? A lot of Nuclear Physics experiments have slow detectors that read-out many BCs in a single “chunk” (ALICE, FAIR Experiments).
- No pile-up at EIC - physics happens every 5th bunch crossing.
- DAQ chain and possibilities for real-time processing will drive the detector design (e.g., favour silicon over TPCs).
- EIC concept is much more integrated design between the accelerator, detector and the software stack.
Protobuf - nice and fast, lightweight; is it sufficiently flexible?
- Would like to avoid ROOT IO as we think this would be limiting in what we can achieve, limited by computing infrastructure and the need to “fit in”. How can we make things easier for the user? Think about new ways to do things. Protobuf buys language agnosticism, and that’s very useful also for transferable skills.
- Flexible and modular approach - interoperability is very important (C++ to Go).
- It’s also good for streaming readouts.
- Sebastien - how many people are trying out these different languages and backends?
  - This will be an outcome of working in the user group as people get started.
  - Current packages (EicRoot, Fun4All, GEMC) are actually very successful. ANL software is the most modular. Here C++ dominates as it is what’s been used historically.

Next meeting is 5 December - Spark based analysis in HEP (ROOT) and LSST
- https://indico.cern.ch/event/754811/