Refset Descriptor Inactivation
I'm just revising our validations for the Refset Descriptor Refset. (Upon realising after realising we had none!).
And have wondering if the entries should remain active for retired reference sets?
The specs (5.2.4.1 Reference Set Descriptor) seem to be silent on the issue... (Unless noted elsewhere)? So I've looked for guidance in the data...
The non-human refset is the only refset that I can think of that's been inactive. And it's descriptor history is a bit murky..
id | effectivetime | active | ... | referencedComponentId | value1 | value2 | value3 |
|---|---|---|---|---|---|---|---|
c3bca49d-3f8b-5f92-90e4-6001bc898173 | 20020131 | 1 | ... | 447564002 | 900000000000461009 | 900000000000461009 | 0 |
0328c732-d156-42ef-b54f-d8d459fe7f56 | 20170731 | 0 | ... | 447564002 | 449608002 | 900000000000461009 | 0 |
There's two entries, it is (was) a simple type Reference Set. I'm not sure if the 2017 entry is just noise, or incorrect somehow.
I don't mind what the answer is. I just want to write an appropriate validation rule.
Comments
Considering the design ideas behind RF2 I would prefer to keep the refset description active in the same way and with the same reasons as why descriptions to inactive concepts stay active. They still appropriately describe the inactive component.
(The only exception is, of cause, if the refset descriptior contains an error, but in that case it should be replaced with a new and correct active refset descriptior.)
My gut feeling is that they should remain active. The reference set itself is inactivated, but the metadata for it is still accurate and probably should be there to aid working with the metadata for the retired reference set.
I feel like we've actually made a mistake with this in the relationships file. When concepts are inactivated their relationships and axioms are inactivated, but that does make it harder than perhaps it should be to dig that out if needed for inactive concepts - e.g. aggregation/analysis of data including inactive concepts. Given the many reasons for inactivation there may not actually be anything wrong with the modelling, and in fact, fixing the modelling as best as possible before retiring it (as crazy as that sounds) would be good in a number of cases.
Leaving the descriptions active for concepts is good in this regard.
Anyway, back to the original topic - I think leave it active.