Scheduling breaking changes we have on our radar

Replying in reverse order to DaxServer, Thad, and Antonin:

  1. Using OpenAPI/Swagger to document the API and generate clients is definitely the way to go, but I'm assuming this would be more than just a straight documentation of the existing API, but also a cleanup with consistent return formats, etc.
  2. Thad's quotes from Stefano are from:
  1. I don't see a rewrite such as proposed by Antonin as something which is feasible. Big bang rewrites always seem more attractive and easier than the messy alternative of dealing with existing code, but they always take longer than predicted and often fail altogether. Joel summarizes my view on this pretty well.

    Focusing on extension stability first makes sense to me, but I don't think the current situation is as dire as Antonin feels it is. I think it's mostly a matter of being a little bit more disciplined about evolving our APIs and adding tooling support to help us know when we're breaking an API unintentionally.
    Tom