Missing guide for how to set up an implementation from existing database

[ml-evs]

We should write some brief instructions on how to go from e.g. a Mongo database of structures with their own field names, to an OPTiMaDe API, mostly explaining which classes should be inherited from/rewritten, and what can be used as is.

[CasperWA]

Sure. Again, this is not relevant if one is not using MongoDB in the backend, e.g., is is not used for AiiDA/MaterialsCloud, but since the standard/example implementation here is based on MongoDB, it sure makes sense for people to want to reuse this repository as a package if they have a MongoDB backend and then some documentation on this would be quite helpful. However, I consider this low-priority, since an implementer can always look into the excellent matador/odbx implementation to see examples of how to do this - right? 😉

[ml-evs]

...since an implementer can always look into the excellent matador/odbx implementation to see examples of how to do this - right? wink

Pffff, I hope not...

I agree re: the rest, but I'm imagining just a brief break down of the FilterParser/FilterTransformer/EntryMapper/EntryCollection classes to make it clear we aren't backend-specific.

[jamesrhester]

(Part of JOSS review)
Just bumping this again.

The one thing missing from the documentation in my view is a simple guide to going from an existing database (of whatever type) to a database serving optimade queries. It seems to me this guide would slot in well under "Getting started"->"A single database". Currently it looks like the user is expected to study how the provided database does it, which is not terrible but could be improved.

[ml-evs]

(Part of JOSS review)
Just bumping this again.

The one thing missing from the documentation in my view is a simple guide to going from an existing database (of whatever type) to a database serving optimade queries. It seems to me this guide would slot in well under "Getting started"->"A single database". Currently it looks like the user is expected to study how the provided database does it, which is not terrible but could be improved.

👍

I see I assigned myself on this issue in February 2020, so I guess I have no more excuses... I will prepare something for next week (and will set up the PR so that it closes this issue when merged).

[ml-evs]

I think #915 is sufficient to close this issue @jamesrhester.

I've just pushed a new release, so the expanded "Getting started" section will appear on https://www.optimade.org/optimade-python-tools/getting_started/setting_up_an_api/ imminently.

The previous single db/multi-db stubs have been restructure as "Example use cases" and another guide for integrating within an existing database has been added.