potiuk edited a comment on pull request #21273:
URL: https://github.com/apache/airflow/pull/21273#issuecomment-1028832350


   > Since it doesn't take long to run the migration, I'm thinking there's no 
harm in running it all the way to base. For instance, we run upgrades all the 
time when we run tests.
   
   I Love this idea. and yeah - we should run in down to 0. I think we 
sometimes correct even old-ish migrations (for example we corrected a number of 
them when we added MSSQL. If it does not cost much and we make the investment 
now, let's do it "fully". I am all for doing all kinds of tests we can imagine. 
We have now also plenty of credits in CI to run even more tests and we can 
speed up more of them by parallelilising,
   
   My point of view is that for Airflow, absolutely everything that can be 
automatically tested and does not introduce unnecesseary burden after initial 
period of stabilisation and flakines (which is kinda inevitable)  SHOULD be. 
Sometimes even if it does introduce some burden (because sometimes that's the 
only way to keep it in check). 
   
   I think also - that we should try more than we end up with. If we see 
opportunity for automation - we should try. if turns out that it's too high of 
a burden to maintain we can **always** adapt or remove it (it did happen a few 
times). But if we don't try - we will not know. There is plenty of room for 
incremental experiments here.


-- 
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