Presently, when reading Podio data, JEventSource::Emit() needs to call JEvent::InsertCollectionAlreadyInFrame<T>() for each collection already in the frame. This requires knowing all types T that might be in the file. This limiting the event source to producing a "closed universe" of types, and also requires some boilerplate and possibly a code generation step.
As per the discussion in #365, it would be nice to avoid this.
Presently, when reading Podio data, JEventSource::Emit() needs to call
JEvent::InsertCollectionAlreadyInFrame<T>()for each collection already in the frame. This requires knowing all typesTthat might be in the file. This limiting the event source to producing a "closed universe" of types, and also requires some boilerplate and possibly a code generation step.As per the discussion in #365, it would be nice to avoid this.