You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Complete the list of mixin classes and add important methods
Make Records of Unions instead of Unions of Records for the cases with links into multiple collections? (maybe better try not to make unions at all, e.g. for truth collection create a homogeneous record)
Find out why ak.where triggers materialization of a few (but not all) columns (probably different situation now with dask)
Silence warnings or add a filter function to be passed to TBranch.iteritems
Zip the HLT collection (this changed in newer PHYSLITE versions, see next point)
Find out how trigger info/matching works and how we make it work in columnar analysis
Find a way to load the hash to branchname mapping from the MetaData tree (or calculate hash manually) (maybe we better hardcode targets for cases where there is only a single one and make a special treatment for truth collection)
Try out what is/isn't working with newer versions of PHYSLITE
Some collections (e.g. TruthMuons) have fiels that can't be zipped (empty lists for some events, so can't use common offsets). How to handle this? Have a list of collections not to zip? Don't zip anything at all?
The text was updated successfully, but these errors were encountered:
This is to track the TODO list/open questions for the PHYSLITE schema for nanoevents
Make Records of Unions instead of Unions of Records for the cases with links into multiple collections?(maybe better try not to make unions at all, e.g. for truth collection create a homogeneous record)Find out why(probably different situation now with dask)ak.where
triggers materialization of a few (but not all) columnsTBranch.iteritems
Zip the HLT collection(this changed in newer PHYSLITE versions, see next point)Find a way to load the hash to branchname mapping from the MetaData tree (or calculate hash manually)(maybe we better hardcode targets for cases where there is only a single one and make a special treatment for truth collection)The text was updated successfully, but these errors were encountered: