...
- Objects in the event should be immutable.
- If a TrackCluster is not in the event, then the record is incomplete.
- There is no record of who is using it.
- We do not need to solve this now but we need to keep it on the list of things to do.
Someone pointed out that the TrackCluster has time field and we asked why. All of the cluster classes discussed today have it. In retrospect we
are not sure that it belongs.
- Time does have meaning at the SimTrackerHit level.
- For strips and pixels it has no meaning at the RawTrackerHit and cluster level.
- For a TPC has meaning both at the RawTrackerHit and cluster level ( it encodes arrival time, and is not synchronous with the beam crossings so it should be a double, as it is in Dima's class ).
- For a clocked CCD the situation is less clear. Time does have a meaning: if you assume a different beam crossing, you get different solutions to the position in the clocked dimension. It's not clear to me that you can even do clustering without assuming a t0. Maybe it makes sense to return the assumed t0 with a getTime() method? In which case it should be an integer. I don't see any meaning for a double getTime().
We do not have a mechanism for recording the configuration information for the clusterfinder and TrackCluster code: what thresholds were used,
what algorithms were used to form clusters, what algorithms were used to compute position once the track was associated.