potiuk commented on PR #73670: URL: https://github.com/apache/airflow/pull/73670#issuecomment-5823882579
Let's see the result of the discussion https://lists.apache.org/thread/5k06417wrbb10hkfk8ogz8py7qgcbpqz and possibly VOTE if we cannot reach the consensus. I would really like to understand why "Human Summary" as a header is wrong. I fundamentally disagree that PR content is **primary** for humans to write. I think it's primary to show intentions and reasoning - and whether it's human or AI written is secondary. I think **clear sections** showing Human/AI **when** (and only when) most of the code was generated by AI and human only specified and verified the intention - is a good idea. This nicely separates "human intention" from the "machine generated implementation". I believe this is absolutely the future way of thinking of PRs: Humans specifying intentions, Machines implementing it (with human overlook). The next step I see: attach a mini-design explaining the intentions of features and bigger changes (ADRs or design docs). And building harnesses that will allow maintainers - via combination of deterministic harnesses and agents verify if the intention is properly implemented, verifying performance issues, security and the like. This is a very nice step where we clearly separate what's human's and what's not. -- 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]
