matterrr opened a new issue, #37977:
URL: https://github.com/apache/airflow/issues/37977

   ### Apache Airflow version
   
   2.8.2
   
   ### If "Other Airflow 2 version" selected, which one?
   
   _No response_
   
   ### What happened?
   
   The arg `include_prior_dates` in `TaskInstance.xcom_pull()` has the 
following definition:
   
   >         :param include_prior_dates: If False, only XComs from the current
   >             execution_date are returned. If *True*, XComs from previous 
dates
   >             are returned as well.
   
   In practice, `xcom_pull(key="<some_key>", include_prior_dates=True)` only 
returns a single xcom (the most recent prior DAG run) when only the `key` is 
specified. I think it would only return multiple xcoms if multiple `task_ids` 
are specified, or the xcom(s) was pushed from a mapped task.
   
   ### What you think should happen instead?
   
   I think that the `include_prior_dates` being plural implies that 
`xcom_pull(key="<some_key>", include_prior_dates=True)` _should_ return an 
`array` of xcoms for all `dag_runs`. This is even consistent with the wording 
in the `get_many()` function called in `xcom_pull()`:
   
   >         :param include_prior_dates: If *False* (default), only XComs from 
the
   >             specified DAG run are returned. If *True*, all matching XComs 
are
   >             returned regardless of the run it belongs to.
   
   But strangely, the `get_one()` function in `BaseXCom` has this wording:
   
   >         :param include_prior_dates: If *False* (default), only XCom from 
the
   >             specified DAG run is returned. If *True*, **the latest 
matching XCom is
   >             returned regardless of the run it belongs to**.
   
   At the least, the wording `include_prior_dates` should be changed to 
`include_prior_date` (singular) if xcoms from _only_ the previous DAG run will 
ever be returned.
   
   ### How to reproduce
   
   h3. Example Code
   
   ```
   import logging
   import sys
   from datetime import datetime
   
   from airflow.decorators import dag, task
   
   default_args = {
       "owner": "airflow",
       "depends_on_past": False,
       "start_date": datetime(2021, 1, 1),
       "retries": 0,
   }
   
   logger = logging.getLogger(__name__)
   
   
   @dag(
       dag_id="test.xcom",
       default_args=default_args,
       schedule=None,
   )
   def test_xcom():
       @task()
       def get_recent_xcom(**context):
           from datetime import datetime
   
           ti = context["ti"]
           xcom_results = ti.xcom_pull(key="test_xcom", 
include_prior_dates=True)
   
           print(f"{xcom_results=}")
   
           ti.xcom_push(key="test_xcom", value=f"{datetime.now()}")
   
       get_recent_xcom()
   
   
   test_xcom()
   
   dag.doc_md = """This test DAG logs the xcom."""
   ```
   
   h3. Steps
   
   1. Run the DAG, noting the xcom value will be `None` for the first run
   2. Run the DAG subsequent times, noting the `xcom_pull()` only ever returns 
a single value from the previous DAG run
   
   ### Operating System
   
   Debian GNU/Linux 11 (bullseye)
   
   ### Versions of Apache Airflow Providers
   
   _No response_
   
   ### Deployment
   
   Docker-Compose
   
   ### Deployment details
   
   _No response_
   
   ### Anything else?
   
   _No response_
   
   ### Are you willing to submit PR?
   
   - [ ] 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