bedhost select and retrieve on backend

[nsheff]

1. should PipestatManager implement select_distinct and select methods, just as passthrough to the .backend object? Yes, but these should be retrieve and retrieve_distinct, and should be backend-agnostic.

2. should bedhost change to interact using pm.backend.select() instead? NO because this is backend-specific

3. bedhost should use .retrieve instead of .select

4. create retrieve_distinct on abstract class

5. have the dbbackend implement this (using select_distinct)

6. have the file backend raise a NotImplementError() when this is called

7. eventually have the file backend implement this (using a dict hash table)

8. have bedhost use the pm.retrieve_distinct

9. remember, avoid if at all possible using use any backend-specific functionality

[nsheff]

Related to: https://github.com/pepkit/pipestat/issues/70 where we started to move select to the abstract class, and realized that should be retrieve.

[donaldcampbelljr]

• retrieve_distinct has been implemented.

pipestat retrieve function can now take a list of record_identifiers and/or result_identifiers which will limit needing to use select as frequently. https://github.com/pepkit/pipestat/pull/101

However, to the best of my knowledge, there are still a couple of places where Bedhost must use select to utilize the filter parameters. @khoroshevskyi can you confirm that the places using select cannot be replaced with the new retrieve_multiple functionality?

[nsheff]

I think this was basically revisited with a new approach in pipestat, here: https://github.com/pepkit/pipestat/issues/103

The above suggestions are all basically changes for pipestat, which are superceded by that issue. So, I think once that is done, this is probably just to be closed...