EUTelescope currently duplicates larger part of the data when going through the processing chain:
The EUTelClusteringProcessor always creates collection "original_zsdata" (hard-coded). This is then expected to be present by almost every subsequent processor including EUTelMille since every call to EUTelUtility::HitContainsHotPixels() requires it.
Furthermore, HitContainsHotPixels() also requires the clusters and the hotpixel map to be present.
This is a significant overhead with no benefit if hot pixels are excluded from clusters in the first place.
Other usage scenarios (e.g. alignment without "hot clusters", track fitting with) should just either
- create two cluster collections (with and w/o hot pixel) and match if needed, or
- prepare a "filtering" processor which (optionally) removes hits/clusters w/ hot pixels (exploiting e.g. the LCIO collection linking feature)
Judging by my understanding of the usefulness of the current approach, I would suggest to remove any reference to "original_zsdata" to reduce the overhead for most of the EUTel users.
Please weight in if this would affect your analysis or any study that you know of!