[NeXus-committee] Next DECTRIS Summary

["V. Armando Solé"]

Hi Pete,

On 17/01/2013 18:27, Pete Jemian wrote:
>
>
> On 1/17/2013 8:07 AM, "V. Armando Solé" wrote:
>> I would like to add that from the discussion I had (via mail) last
>> September with Ray Osborne and Carlos Pascual I got the idea the "new
>> way" of moving signal and axes attributes to the NXdata group was
>> already official.
>
> Actually, I brought this to the NIAC as an outcome from the 2012 
> canSAS workshop in Uppsala.
>
> This change was not decided.  One significant point resisting the 
> change was that we are dithering our standard.  Changes such as this 
> should be considered deeply, with consideration of existing data files.
>

If a standard is not up to the task, it is time to define a new standard 
(how often has the "meter" been redefined? At least twice during my 
life). The disadvantages I see with moving the attributes to the NXdata 
group itself (changing again my codes) are largely compensated by the 
advantages (much cleaner code, datasets being reusable as axes or 
signals, DECTRIS issue, ...)

The consideration about existing data files does not hold as they will 
be supported by the codes that already support them and during some time 
even the datasets inside the NXdata group could carry the "ancient" 
attributes. It would not be the first time the NXdata group is revisited 
(use of "axis" or use of "axes").

My way of thinking is that a new need implying a new solution certainly 
does not break compatibility while a bad solution to define a plot 
(attributes to *each* of the datasets involved in a plot) will always 
remain a bad solution. It seems logical that if NXdata is a group 
defining a plot, the attributes to define the plot belong to the group 
and not to the datasets inside (kudos to the guys that came out with 
that idea). If it is logical to call it NXdata is something I guess has 
been discussed in the past and that to change would certainly break a 
lot of code.

Best regards,

Armando