kolfild26 opened a new issue, #33373:
URL: https://github.com/apache/airflow/issues/33373

   ### Description
   
   Currently, in TimeDeltaSensor its `delta` interval starts from [its dagrun 
creation](https://github.com/apache/airflow/blob/main/airflow/sensors/time_delta.py#L44)
 but not from when the task's actually been running.
   
   ### Use case/motivation
   
   Imagine we've got some tasks before the TimeDeltaSensor task.
   In order to use TimeDeltaSensor we need to know their expected execution 
time and add this time to the sensor's delta parameter. Ok, it's not good for 
using but can be acceptable. But what if their expected execution time may vary 
for some reason from time to time? How to calculate the delta to be passed in 
sensor?
   
   So, is there any reason it's done this way? Wouldn't it be better if `delta` 
was attached to its task instance, which seems to be more intuitive. smth like 
this:
   ```
   L44 target_dttm = context['task_instance'].start_date
   ```
   
   ### Related issues
   
   _No response_
   
   ### Are you willing to submit a PR?
   
   - [X] Yes I am willing to submit a PR!
   
   ### Code of Conduct
   
   - [X] I agree to follow this project's [Code of 
Conduct](https://github.com/apache/airflow/blob/main/CODE_OF_CONDUCT.md)
   


-- 
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