Also - I always like those kinds of policies to work on examples. And we
have a cool one.

I added a proposed rule to make it easier a bit when we are accepting a new
provider, that is about not a service managed by a 3rd-party. This is when
we have another open-source software integration. In particular, when such
open-source software has governance by Apache Software Foundation or
similar foundation with good governance, I propose that such a provider
(when PR is of a great quality). LAZY CONSENSUS should be enough to accept
such a provider.

We have now a great example - Dylan from Astronomer had just completed
iteration on Apache Kafka Provider PR (donated by Astronomer and for a long
time tested and proven with their providers package) and it has precisely
the kind of quality we - I think - need to be able to accept such provider,
That is including integration test that make it easy to setup Kafka with
breeze and run tests with the real "Kafka" running in a docker-compose
managed setup. I will call for a LAZY CONSENSUS there - and if that will be
cool for everyone, we can treat it as a precedent for similar cases in the
future.

J.


On Sat, Apr 15, 2023 at 9:56 AM Jarek Potiuk <[email protected]> wrote:

> Hello Everyone,
>
> I prepared a PR, where I tried to capture the results of all discussions
> of the community/3rd-party providers
>
> https://github.com/apache/airflow/pull/30657
>
> This is just a proposal for now - and since English is not my
> native language (and I am known to write too many words rather than too
> little), I am happy to get the comments and corrections, and updates to the
> PR. The aim of this page is to explain once and for all what is our
> approach to accepting new providers, explaining what are the consequences
> of having a community vs. 3rd-party provider so that everyone can be aware
> of it without reading long devlist discussions we had about it,
>
> Part of this PR is extracting what we had so far in README to separate
> PROVIDERS.rst file - which I think is a better place as the small chapter
> about providers outgrew the README already (which should be small and
> concise).
>
> Take your time in reviewing and commenting on it. I think it is important
> to make it precise and clear. Anyone is welcome to contribute.
>
> J.
>
>

Reply via email to