[Nexus] NXlog for event mode

[Tobias Richter]

Hi all,

In scan mode we tend to record a set of dynamic metadata (sample temperature, orientation, incident flux) per detector image. For neutron event mode there are no images recorded in the raw data and recording all metadata per neutron would be excessive. But for some fast dynamic data (electric fields) you would want a relatively fine temporal granularity. The solution so for seems to be to use NXlog and correlate the times using binning, averaging, interpolation, or the like. So far so good.

The current use of NXlog in the wild as far as I can tell happens in complete parallel to the NeXus “expected” hierarchy of entry/instrument/chopper_one/phase instead you find something like entry/logs/chopper_one_phase. Nothing but the name tells the user (or reduction code) what the data is. My wish would be to stay close to the scan usage and my primitive idea would be to replace every applicable (HDF5) dataset with an NXlog group.

Unless there are good reasons (not counting the grey area of lack of time of effort) why that wasn’t done originally that I cannot think of, this seems a better approach than relying on names in new structures. Admittedly in both schemes the main data is inaccessible to exiting scan aware software, but with mine scheme at least all auxiliary information would be in known places.

Any comments?

Tobias