potiuk commented on PR #34257: URL: https://github.com/apache/airflow/pull/34257#issuecomment-1712917897
TL;DR: I think lack of min-version does not "invalidate" RC1. It might go in the next release. So there is no need to "cancel" google RC1 because of that change. But it is good to have it merged for the next release. Bumping min version for common-sql is generally "good" . Bumping any "min-version" dependency when we have a good reason for it is generally a GOOD thing. We've been discussing it at several other occasions (azure provider recently) and generally setting "min-version" for a dependncy that we already have higher in our constraints is a good thing - especially if it does not have a potential of conflict. Comon-sql is 1.y.z and we should keep it like that forever, and that means it's backwards-compatible. This means that anyone should be able to safely update to latest 1.* version and there is no potential for conflict. Good example of such potential conflict - setting pydantic min_version to >= 2.0.0 is a bad idea now as some of our dependencies have not yet upgraded to Pydantic 2. Also Pydantic is the "core" dependency, and we could be far more relaxed in case of provider dependencies. When we set min-version in provider, user can always stick to previous provider version. But this is NOT the case with common.sql - we CAN (and should in this case) bump the min version because potential of conflict is `0`. But we do not have to invalidate current release because of that. -- 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: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
