potiuk commented on PR #58497: URL: https://github.com/apache/airflow/pull/58497#issuecomment-3567159195
> @potiuk, the PR that broke it was not included in changelog as it was titled as mypy fixes. Not blaming here or anything, just letting you know - not sure how we can improve the process for RM so it's easier to decide what should and should not be included in Changelog. Whether something is in a changelog or not, does not matter when it's breaking something. So It does not matter for RM to be honest. RM (and you can look it up in https://infra.apache.org/release-publishing.html#releasemanager - release manager's work is purely mechanical. Release Manager manages the *mechanics* of the release, not whether particular PR is breaking some subtle aspects of the product. It's more of the initial contributor's role to make sure the regression tests are good enough, and reviewer's case to verify. RM has to trust that if the PR is green and been reviewed, that it does what is described in the PR description. This failure here is not really a failure of a release process but either: * tests were not comprhensive enough to catch a regression or * reviews did not catch it Once those two are passed, RM **might** make a third pass (and sometimes does) - but there is no physical possibility (nor is not RM responsibility) to be able to check if a change has side effect. I guess in this case it's the matter of tests that should be written in the way that they fail if things change and it should be clear that the tests are not "accidental" and that they expect tuple to be returned. Or am I wrong. How else would you want to expect that RM should know that things are broken if the tests pass and reviewer said "ok" and someone merged it? -- 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]
