[NeXus-committee] Variants

[Koennecke Mark (PSI)]

Hi,

we briefly discussed the issue with the variants at the last telco. A means to encode variants is needed quickly for 
a pending update to NXmx for supporting the upcoming EIGER 2 detector. Any variant scheme will do for this. 
The agreement on the Telco was that I write this mail which reflects the current state of things as we know it. And 
that we use it until another scheme has been approved by NIAC. If someone is really unhappy about this, please 
speak up!


This is what I found in my Telco notes for the week 14 2014. 

-------------------------------------------------------------------------------------------------------------------------------------------------------
Variants
---------

It was aggreed to treat variants in a similar way as we treat
errors. Thus the actual field with the NeXus name will always hold the
best value. And in an attribute the names of any variants for the
field. Example:

distance
    @variant=distance_from_reading
distance_from_reading
    @variant=distance_as_guessed_by_cleaner
distance_as_guessed_by_cleaner

To be ratified at NIAC.

----------------------------------------------------------------------------------------------------------------------------------------------
And in the NIAC 2014 notes I found that this was actually ratified. 

We know that this scheme will fall over with external links. However, variants will usually only affect small 
data items like distance to sample and such. And this will not be very likely to be stored in an external file. 


Best Regards,

     Mark Koennecke