October, 13-14, 2016 in Copenhagen, COBIS Building, M10
Welcome from Tobias, organizational tidbits
Ray Osborn unanimously confirmed as representative of ANL in NIAC
Stephen Cottrell unanimously confirmed as representative of the musr community in NIAC
Armando Sole unanimously accepted as ESRF representative in NIAC
Joachim Wuttke has expired as NIAC member and does not wish to maintain his membership. There will be no new representative for FRM-2.
Presentation by Mark Koennecke
Discussion on validating units and units itself.
Pete Jemian presented the canSAS application definition. There is a mapping from canSAS to NeXus.
There was some discussion on this. The result was that the definition will be revised to only represent the processed data (I(Q)) use case. To be revisited once this is done. May be tomorrow.
Ray presented his suggestion for a revised NeXus WWW-site. The main point is to make it look more modern. Also it is to be generated with sphinx which makes it easier to restyle it using sphinx styles.
Still needs some work, revisit at NeXus Telco. Then make it mobile friendly. If we can do this easily with Sphinx.
Herbert is presenting proposed changes for NXmx-1.5.
NXmx is adding a NXdetector_group for documenting detectors better to humans. They have funny detectors with segments which move diagonally with regards to the detector center.
Added an optional third detector dimension. This is a data organizational dimension.
There is some discussion about understanding this.
Changes in NXbeam for polychromatic beams in the application definition. As optional, added in order to be able to validate this.
Flux may need two values, NXbeam defines flux is flux per unit area. The additional flux is a rate: photons/second. To be called total_flux.
Adding incident_beam_size which is a 2 value array of FWHM (when gaussian) or a diameter (if it is a top hat) or 2 sizes for a rectangular beam.
Changes to NXmx accepted by consensus after a little discussion with no objection.
Herberts presents changes to NXtransformations.
Axis rotation ranges are being added. Crystallographers want that for emotional reasons. The use case is the oscillation method.
Under discussion this changed to:
These three proposals to change NXtransformations below where unanimously accepted for by vote:
The general axis is highly contentious and not very well understood. Herbert says that a general axis could also be expressed as a translation.
The end of this: Aaron and Herbert provide better documentation and examples. We also half agreed to drop the transformation_type attribute for the general axis. This will be deferred until we understand this better.
Very nice dinner at taarnet.dk in a high security environment with x-ray.
David Mannicke initially missing.
Tobias Richter confirmed as chair with one abstain, 11 yes
Mark Koennecke confirmed as executive secretary with one abstain, yes 11
Eugen Wintersberger confirmed as technical release manager with one abstain, yes 11
Pete Jemian confirmed as documentation and definition release manager, 12 yes
Mark Koennecke presented the results of the code camp discussion as a proposal.
We accepted the renaming NXevent_data fields to what Freddie presented and is being used at SNS and ISIS. All yes.
No majority for summarizing the streamed data in NXdata.
Action Item: Mark Basham to report on VDS performance in june. Based on that report we decide if NeXus does need an own segmentation scheme.
Presentation by Aaron Brewster. The use case is to store intermediate results
Discussing NXreflections in more detail:
Accepted as a base class with 12 yes, with the changes listed above. We trust Aaron to apply the changes.
Tobias Richter presents suggested changes to NXsample.
Accepted with 12 yes
Some details deferred.
Pete Jemian presenting the canSAS application definition revised from feedback of yesterday.
There was some doubt about this is what the canSAS group wants.
We accept NXcansas with NXinstrument made optional, 8 yes, 4 abstain
Short discussion on fixed versus variable length strings.
We accept both variable and fixed length strings, readers have to support both, 11 yes, 1 no
Short discussion on the versioning proposal as presented by Mark Koennecke from the result of the code camp.
Herbert made the comment to improve the workflow in this way:
The proposal was accepted on probation with 12 yes. To be revisited at the next NIAC in order to check if it works.
We ended the discussion on a time limit.
The problem: required group, some cases where no sensible default plot can be provided.
Change the documentation: make it clear that NXdata should be there but can be omitted if no sensible plot can be provide, 12 yes.
Pete Jemian has a modified version of application definitions with the 2014 attributes applied. This is the signal, axes definitions as group attributes. When changing application definitions consider the code in that branch.
This is informational.