potiuk commented on PR #35972: URL: https://github.com/apache/airflow/pull/35972#issuecomment-1869511325
> We _did_ mark AFS as experimental right? So, in theory we could be a bit more daring in not doing backwards compatibility? Well. Not if we are targetting our own, released providers .. That's like introducing a deliberate breaking change that breaks what's already released there. I think we have to be really careful when we are pushing out a chang that our providers depend on. Yes, it's more effort and yes, keeping the backwards compatibility will cost us, but I'd say if we want to release a code that our provider depends on, then we should NOT release it (feature flags from AIP-44 (internal API) for example now and AIP-52 (Setup/Terdown) had been doing that in the past - we released new Airflow version with those "partially done" ut disabled. We **could** have done it this time as well if we felt that current implementation is not ready for prime time (and for us to support backwards compatibility). I think in this case we shoudl simply acknowledge we were rushing too fast, bite the bullet, and introduce and maintain compatibility layer. - treating it as valuable lesson to be more careful wiith such changes in the future. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: commits-unsubscr...@airflow.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
