[NeXus-committee] Linking NXdata to NXdetector groups

[Ray Osborn]

Dear colleagues,
Peter will put out a formal call for votes on this week's issues, including
the one about linking NXdata to NXdetector groups.  We wrote a summary of
what we thought were the main proposals to come out of the discussions in
the last few weeks, and posted them at the beginning of the Swiki page
<http://www.neutron.anl.gov.:8080/NeXus/32>.

I would request that everyone does two things before voting.  Firstly,
please check that the summary accurately reflects the proposal that you
favor.  If it doesn't, post a message to this list as soon as possible.
Secondly, please check the examples and try to imagine circumstances in
which the scheme will not allow you to do something you need to.  If you
feel strongly about this, let us know what key limitation led you to vote in
the way you did.

I will use my chairman's prerogative to say that I strongly favor Proposal 1
as the only one that gives the flexibility to put any axes into an NXdata
group, including non-NXdetector axes. I am worried that the debate became so
fixated on how to link different detector banks that we forgot that people
might need to store other variables, such as sample temperature or
monochromator energy, as one of the axes.  NeXus must be flexible enough for
all types of data, and I'm not convinced that the other proposals provide
that flexibility (but correct me if I'm wrong).

By the way, with the suggested amendment by Freddie (of linking counts as
well), I think Proposal 1 is essentially identical to the PSI scheme that
Mark proposed; I was just slow to realize it.  I did suggest a slightly
different linking scheme to the one we use (i.e., storing absolute paths as
an attribute), but that is an API change that we need to discuss separately.

Best regards,
Ray
-- 
Dr Ray Osborn                Tel: +1 (630) 252-9011
Materials Science Division   Fax: +1 (630) 252-7777
Argonne National Laboratory  E-mail: ROsborn at anl.gov
Argonne, IL 60439-4845