Skip to content

Full persistreco

This will be a shared RTA WP1 - DPA WP1 task as for Sprucing and Tesla++ to move forward Moore needs to persist the full reconstruction as done in Run 2 with the persistreco flag.

We will follow the run 2 method in https://gitlab.cern.ch/lhcb-rta/direction/blob/master/wp1/persistency.md

For persistreco in Moore, on top of !628 (merged) (@apearce ), we need

Converters

Note that (initially) much of the run 2 code can be recycled. At the moment the selections in HLT2 run over run 2 like objects - converters take the new event model objects from reconstruction and make them look like the objects of run 2. For the persistency we could utilise these converters letting us keep the old packers/serialisers etc. It does not seem that these converters will cause any timing bottlenecks. Whether or not we wish to persist the old event model into the offline or the new one is a DPA decision but the above should be our starting point.

Edited by Ao Xu