Redesign of "datasets"
Calculations can be assigned to dataset. Before putting the repository on Elasticsearch, this used to be more complex allowing a form of nested dataset, which are not supported anymore.
This issue is about, what dataset should actually be, independent from what they are right now or have been in the past. The discussion should be from a user perspective, not from a technical point of view.
There are questions that can help shaping the meaning of datasets.
- Can a calculation belong to multiple datasets?
- Can a dataset belong to another dataset?
Based on answer combinations, we have:
tags
- datasets can not be part of other datasets
- a calculation can be part of many datasets
hierarchies
- datasets can be part of other datasets
- a calculation can be indirectly contained in many datasets
- for direct containment 1:n or m:n is possible
- datasets can comprise only other datasets or calculations or there can be mixed datasets
flat
- a calculation can be part of only one dataset
- datasets can not be part of other datasets
Personally, I prefer tags which provide good flexibility, users should be familiar with the concept, its relatively easy to implement, and it is a compromise between hierarchies and flat.