dstandish commented on PR #24743:
URL: https://github.com/apache/airflow/pull/24743#issuecomment-1173948817

   OK addressed the comments
   
   > We still need to address the HA/lock issue 
   
   As presently constructed we shouldn't get conflicts because I think each 
dagrun should get a unique run id based on timestamp. But yeah we do need to 
think through how dagrun creation should work in this area. Let me know if you 
have any thoughts around what we should do.
   
   > you need to explain/highlight where you've diverged from the AIP (which is 
from what I can see the behaviour around multiple datasets.)
   
   Yeah so the AIP does state intention to support multiple upstream datasets I 
guess the change is doing it now rather than later.  The other difference is 
AIP shows support for single rather than in a sequence -- we could pretty 
easily update this code to tolerate either sequence or dataset obj -- but i 
kindof like just requiring sequence better.  WDYT?
   
   There is dagrun creation behavior that isn't addressed in AIP that we do 
need to sort that out.  But if there's a need to unblock other work perhaps we 
can do that in followup after discussion and debate.


-- 
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