The current architecture only fills truth info if a reco branch filler requests it. That means that events that happen to form no reconstructed objects in any of the detectors that are enabled (think, e.g.: lots of neutrons produced outside fiducial volume that make small blips inside it) won't have their truth info stored.
Fixing this would require the various RecoBranchFillers to invoke the TruthMatcher even when they don't have a reconstructed object. That's easy enough; but there's no way of enforcing that they do that, which means that events could easily slip through.
Should the trigger grouping machinery also group all the relevant true interactions as well? Should we just add in explicit calls to TruthMatcher in the reco branch fillers in an ad hoc way? More thought is needed.
The current architecture only fills truth info if a reco branch filler requests it. That means that events that happen to form no reconstructed objects in any of the detectors that are enabled (think, e.g.: lots of neutrons produced outside fiducial volume that make small blips inside it) won't have their truth info stored.
Fixing this would require the various RecoBranchFillers to invoke the
TruthMatchereven when they don't have a reconstructed object. That's easy enough; but there's no way of enforcing that they do that, which means that events could easily slip through.Should the trigger grouping machinery also group all the relevant true interactions as well? Should we just add in explicit calls to
TruthMatcherin the reco branch fillers in an ad hoc way? More thought is needed.