Add newsfragment and fix compat tests

newsfragments/43902.significant.rst

@@ -0,0 +1,5 @@
+Renamed ``execution_date`` to ``logical_date`` across the codebase to align with Airflow 3.0.
+
+The shift towards ``logical_date`` helps move away from the limitations of ``execution_date``, particularly with dynamic DAG runs and cases where multiple runs occur at the same time. This change impacts database models, templates, and functions:
+- Renamed columns and function references to ``logical_date``.
+- Removed unique constraints on ``execution_date`` to support concurrent DAG runs with the same logical date.

providers/src/airflow/providers/google/cloud/sensors/gcs.py

@@ -190,9 +190,14 @@ def ts_function(context):
     except KeyError:
         from airflow.utils import timezone
 
-        data_interval = context["dag"].infer_automated_data_interval(
-            timezone.coerce_datetime(context["logical_date"])
-        )
+        if AIRFLOW_V_3_0_PLUS:
+            data_interval = context["dag"].infer_automated_data_interval(
+                timezone.coerce_datetime(context["logical_date"])
+            )
+        else:
+            data_interval = context["dag"].infer_automated_data_interval(
+                timezone.coerce_datetime(context["execution_date"])
+            )
         next_info = context["dag"].next_dagrun_info(data_interval, restricted=False)
         if next_info is None:
             return None

providers/tests/standard/operators/test_python.py

@@ -941,7 +941,7 @@ def test_virtualenv_serializable_context_fields(self, create_task_instance):
             *intentionally_excluded_context_keys,
         }
         if not AIRFLOW_V_3_0_PLUS:
-            declared_keys.add(
+            declared_keys.update(
                 "next_ds",
                 "next_ds_nodash",
                 "prev_ds",