ninsbl commented on issue #58385:
URL: https://github.com/apache/airflow/issues/58385#issuecomment-3551405549

   Thanks @Lee-W for the feedback.
   
   To me, the essence of AIP-83 seems to be to rename `execution_date` to 
`logical_date` and to make it nullable. What I get as the motivation behind the 
name- and constraint change has been the ambiguity of the term 
`execution_date`... Citing from AIP-83:
   
   > Logical date is not a bad name because, it implies that it is a date that 
carries meaning for the dag, that drives the behavior of the dag, as 
distinguished from execution_date which, could be misunderstood as "when it 
ran".
   
   So, in that sense, The current behavior in a way contradicts the idea behind 
AIP-83, as data-aware scheduling always assigns "when it ran" to 
`logical_date`, no?
   
   A practical example: on _2025-11-16_, I re-run a DAG that uses data from an 
external source that represents a snapshot of lets say _2025-10-05_ (satellite 
images in my case). Data-aware scheduling as of now always assigns _2025-11-16_ 
as `logical_date` for all DAGRuns triggered by the `Dataset`/ `Asset` update. 
That however is the actual `execution_date` and has no logical meaning for the 
DAG / DAGRun at all...
   
   That said, I am not suggesting to change the current behavior, but to at 
least give users an option to overwrite the current behavior...


-- 
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