Closed stuartcampbell closed 3 years ago
Actually, based on the previous discussion on the beam_size_x
type issue, I would propose either entry_uuid
or entry_identifier_uuid
seem to fit with the recommendations.
NeXus allows the addition of field not defined in the NXDL base class. This proposal is to identify and reserve the names of specific fields for two terms that are part of Bluesky. Such inclusion in the NeXus base class will standardize how NeXus files are generated from Bluesky data (and the reverse as well).
Citing from the manual about the NeXus Community:
NeXus began as a group of scientists with the goal of defining a common data storage format to exchange experimental results and to exchange ideas about how to analyze them.
Already, NeXus is proposed as an interchange format to exchange Bluesky data with high-performance computing and machine learning.
What is the bluesky entity identified by entry_identifier_uuid
?
Examples of current use (not necessarily consistent with this proposal) are here: https://apstools.readthedocs.io/en/latest/source/_filewriters.html?highlight=NeXus#hdf5-nexus-file-structures
What is the bluesky entity identified by
entry_identifier_uuid
?
The start document uid
Proposal:
entry_identifier_uuid
as an optional field in NXentry
BLUESKY_
(or BS_
) as a reserved prefix (reserve option to use either after discussion with Bluesky project)DOI for UUIDs: https://doi.org/10.17487%2FRFC4122 (or https://www.rfc-editor.org/info/rfc4122)
The proposal was accepted with 13 votes for (0 against)
@stuartcampbell has said that the Bluesky project prefix should be BLUESKY_
UUID specification: https://www.rfc-editor.org/rfc/rfc4122.html
The current
NXentry
has the fieldentry_identifier
which is defined as a "unique identifier for the measurement, defined by the facility."Facilities that are using the Bluesky data acquisition have both a unique identifier (which is a uuid) and a 'transient identifier' which is along the lines of a traditional scan/run identifier.
As more facilities are adopting Bluesky, it would be nice to come to a recommendation on how to store both the
uuid
and scan id before we all develop our own standards. At the moment, I can see an argument for storing either the 'scan id' or the 'uuid' in theentry_identifier
- but then where to store the other one.Either adding an optional field like
entry_uuid
orunique_entry_identifier
or allowing the adding auuid
attribute ?