In todays fluorescence experiments users often raster the sample with the beam and record a single spectrum at each sample location. For beamline P06 at DESY this procedure became virtually the default mode of operation. The current NXfluo application definition does not include this kind of experiments as it only allows to store a single spectrum below /:NXentry/:NXinstrument/:NXdetector. However NXfluo could be easily modified to accommodate raster scans.
The changes concern only the NXdetector instance of NXfluo. The layout should be changed as follows
entry:NXentry
instrument:NXinstrument
flourescence:NXdetector
data[npts,nenergy]
energy[npts,nenergy]
where npts denotes the number of raster points. It is important to note that we do not care about the particular pattern used to raster the sample. This would be hard to achieve taking into account that there is a nearly arbitrary number of patterns available. Even entirely irregular paths would be possible (and have already been requested by one of the beamline scientists here at DESY). One advantage of this pattern agnostic approach would be that analysis software can stick to its current workflow: processing the spectra recorded during an experiment in a simple sequential order.
If the energy dataset is of shape [nenergy] (rank one) we can assume that the energy range was the same for each point recorded. A single spectrum can easily be stored by using the original schema of the definition which could be interpreted as ntps=0.
Just as a recommendation (must not go into the standard), it would make sense to store the position of the beam on the sample like this
entry:NXentry
sample:NXsample
x[npts]:NX_FLOAT
y[npts]:NX_FLOAT
The suggested change is in line with NeXus scan rules. Thus the change was accepted at NIAC 2014. No vote was necessary.