Retrievers: should document_store be a private attribute?

[anakin87]

• Retrievers of InMemoryDocumentStore (InMemoryBM25Retriever...) have the attribute document_store

• Retrievers of Elasticsearch and Opensearch Document Store have the attribute _document_store

We should probably agree on a consistent approach,
so that for example pipeline.get_component("retriever").document_store always works.

@silvanocerza

[silvanocerza]

Yeah, doesn't make much sense for the Document Store to be public. Also if you created an instance of that Component I would expect you must have also built the Document Store it's using.

[masci]

What if I'm reading a pipeline from Yaml? I can still get the retriever with pipeline.get_component("retriever") but how do I access the document store?

[silvanocerza]

Good point, that indeed could be a use case. 🤔

[masci]

Let's agree that yes, it's ok to expose the document store instance and close this issue.

I'm not sure we should enforce a convention right now, it's an easy fix and we can patch integrations as we need it, but in case you want to schedule some work to uniform the existing document stores, please open another issue with a checklist of what should be done.