potiuk commented on pull request #20510:
URL: https://github.com/apache/airflow/pull/20510#issuecomment-1002583733


   Not sure what it really means, but if this is a "behaviour" change (looks 
like) rather than API or "effect" change, it's not a breaking change IMHO. 
Especially if it can be disabled with a parameter (which I believe can be in 
this case).
   
   If the "effect" is the same, but there are various things done different on 
the way, this is not a breaking change really. As mentioned elsewhere - 
determining if something is a breaking change or not is never "0/1"  - things 
that are breaking for some people might not be breaking for others, and 
Python's versatility is such that potentially any change can be breaking if you 
extend or monkeypatch a class.
   
   I think when determining if a change is breaking we should really consider 
one thing; how likely it is that significant portion of our users (and 
especially all that are using the operatrs in their simplest form precisely as 
intended) will get broken dags.
   
   If it's likely -> it's a breaking change, If it's very unlikely and depends 
on some "magic" done by the user -> it's not breaking.. Anythng in-between 
should be our decision on how confident we are it will be "smooth sailing" for 
our users.


-- 
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]


Reply via email to