potiuk commented on pull request #19637: URL: https://github.com/apache/airflow/pull/19637#issuecomment-971967128
Hmm. Maybe I am a little overcautious here, but it comes from the experience of responding and trying to help many of our users. I think it really boils down to what we want to optimize for. My current optimisation "goal" is to decrease the number of problems our users even approach us with. So if there is a problem we can anticipate and solve for our users without them even knowing it, I am all for that. Adding more and more escape hatches like that is going quite against that "optimisation" goal, because we admit that there is a problem and we try to "throw the problems at our users to solve" - and this (from the experience) will hit us back. This was the case with the warning for "removed" task instance entries. I know why we added it but IMHO it exposed an internal detail of airlfow to our users, and this kinda 'undermines' the feeling that airlfow is "reliable" and "safe to upgrade" - users who just want to install airflow without knowing how it works intenrnally should be first-class citizens as well, and those kind of "escape hatches" just leaks the "airlfow" abstraction. I think sometimes we tend to forget that our Ops (even admins) are not "developers" or even no "sql admins" should not eventually even know that there is a database behind. This is really an implementation detail for many. But If I am the only one to be worried, I am ok with the flag (I just hope in the next few months we will not have to handle a floo od of issues upgrading from say 2.0.2). -- 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]
