Cube model - relation of DataProduct to DatasetMetadata
CresitelloDittmar, Mark
mdittmar at cfa.harvard.edu
Wed Mar 21 20:21:52 CET 2018
All,
We have been busy these past few weeks exercising the ds/cube/coords/meas
models in the time series case.. creating a toy SimpleTimeSeries model and
serializing example files to it. This is in context of a study being
conducted by the TDIG.
I am now wanting to tie-up some loose ends and turn over a set of models
and examples for review which cover the cube cases (event list, NDImage)
and is compatible for supporting the above type extensions. (NOTE: the
Spectrum model would follow a very similar pattern).
Anyway.. working these examples, I come back again to the relation of a
DataProduct (eg SparseCube, SimpleTimeSeries ) with the corresponding
DatasetMetadata instance.
Earlier versions of the cube model included a fairly tight relation between
these models, and
last year, there was an extensive discussion on this list which included
this topic. (noted on the cube twiki.. "Time Series Cube DM"). This
discussion resulted in my removing the connection entirely.
I strongly feel that it is a mistake to remove the connection entirely.
One must be able to associate a DataProduct with the corresponding
DatasetMetadata.
This is especially true for any serialization that may hold multiple
instances of them..
ie: multiple TimeSeries.. multiple Spectra..
I'll note that other example models
+ the sample model in Jiri's Time Series in Cube note
+ toy models by Mireille and Laurent made for the TDIG study
all have a strong relation between these elements (extension or
composition).
Keeping in mind the earlier discussion, I will be updating the model so
that:
+ DataProduct contains a reference to the corresponding DatasetMetadata
This is the lightest relation we can make which satisfies the needs and
provides the most flexibility w.r.t. what kind of Dataset description we
have (eg: ObsDataset vs "SimDataset" vs "??Dataset")
I just wanted to let people know what has been going on, and to give a
head's up about the change.
Mark
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.ivoa.net/pipermail/dm/attachments/20180321/8f3a6f13/attachment.html>
More information about the dm
mailing list