[WIP] SuperDuperDB simplification to assure its adoption and extensibility

[kislerdm]

Context

As a SuperDuperDB customer, I want to use it to integrate with different databases and to be able to extend the library with onboarding effort less than a day[1].

Problem

How to ensure usability, maintainability and extendability of SuperDuperDB?

• Simplicity: the modules' boundaries are clearly defined.
• Clarity: the architecture is self-explanatory.
• Conciseness: public interface is concise and intuitive, and the codebase does not contain "dead code".

The RFC illustrates the design approach aiming to enable us and our community to improve on the mentioned aspects.

Proposed solution

Status quo

Classes relation

Click to zoom

Package architecture

Click to zoom.

The above diagrams illustrate complexity of SDDB.

The question to the audience: is it avoidable complexity?

Notes

1. The figure is arbitrary picked for the sake of illustration.

References

• kedro dataset "port"
• ports&adapters/hexagonal architecture
• simple made easy