What to do with "empty" datasets
Closed this issue · 4 comments
Some datasets are still referenced by the system, while there are no more observations in them (for example, they only contains old records, now filtered by #215).
It is tempting to add some mechanism that, at the end of each data import, remove datasets that have no longer any observations so the different screens are not polluted by them.
Problem is: some of those datasets might be included in user alerts. Is it therefore acceptable to automatically/silently edit the alert configuration to remove the "now empty" dataset (please note that if this dataset reappears later because it has observations again, it will be with another ID, and therefore not automatically re-added to the alert configuration)?
I think we don't have to be that shy and that we can implement the mechanism that remove empty datasets at the end of each import.
Sure, we'll change the definition/configuration of some user's alert, but that alert would point to a dataset with no observations anyway (+ I think the case will be pretty rare)
I agree, referring to empty datasets will be confusing.
I agree too, @niconoe.
That's now implemented!